<%BANNER%>

Demand for Air Passenger Traffic and Its Impact on the Tourism Industry of Florida

Permanent Link: http://ufdc.ufl.edu/UFE0022668/00001

Material Information

Title: Demand for Air Passenger Traffic and Its Impact on the Tourism Industry of Florida
Physical Description: 1 online resource (299 p.)
Language: english
Creator: Cazanova, Jose
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: adjustment, airline, attacks, autocorrelation, behavior, consumer, demand, elasticities, equations, florida, model, partial, rigidity, seemingly, terrorist, tourism, transportation, unrelated
Food and Resource Economics -- Dissertations, Academic -- UF
Genre: Food and Resource Economics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The airline industry plays a key role in the tourism industry, serving as a vital link between consumers and the tourism industry. Since nearly one half of the tourists coming to Florida use air transportation, having quantitative methods for both explaining and forecasting air passenger traffic are vital to planning and infrastructure development for the tourist sector and the state in general. Also, knowing how air travel demand responds to unexpected shocks, such as hurricanes, provides guidelines for emergency planning, risk assessment, and prevention management as well as impacts on the state's economy. The primary objective of this research was to develop an understanding of the factors influencing domestic demand for airline travel to Florida. Factors such as prices, income, terrorism, seasonality, storms, wildfires, and advertising expenditures were included in the analysis. Estimation results from the partial adjustment model indicates that overall air passengers respond immediately up to a certain extent to changes in some demand drivers and that the full response to such change is realized in subsequent periods. Also, results indicate that demand for air passenger travel has not fully recovered from the terrorist attacks on September 11, 2001. Simulation analysis indicates that overall demand for air passenger travel to Florida is more sensitive to changes in income and that wildfires have no significant impact on demand for air passenger travel to Florida.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jose Cazanova.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Ward, Ronald W.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022668:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022668/00001

Material Information

Title: Demand for Air Passenger Traffic and Its Impact on the Tourism Industry of Florida
Physical Description: 1 online resource (299 p.)
Language: english
Creator: Cazanova, Jose
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: adjustment, airline, attacks, autocorrelation, behavior, consumer, demand, elasticities, equations, florida, model, partial, rigidity, seemingly, terrorist, tourism, transportation, unrelated
Food and Resource Economics -- Dissertations, Academic -- UF
Genre: Food and Resource Economics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The airline industry plays a key role in the tourism industry, serving as a vital link between consumers and the tourism industry. Since nearly one half of the tourists coming to Florida use air transportation, having quantitative methods for both explaining and forecasting air passenger traffic are vital to planning and infrastructure development for the tourist sector and the state in general. Also, knowing how air travel demand responds to unexpected shocks, such as hurricanes, provides guidelines for emergency planning, risk assessment, and prevention management as well as impacts on the state's economy. The primary objective of this research was to develop an understanding of the factors influencing domestic demand for airline travel to Florida. Factors such as prices, income, terrorism, seasonality, storms, wildfires, and advertising expenditures were included in the analysis. Estimation results from the partial adjustment model indicates that overall air passengers respond immediately up to a certain extent to changes in some demand drivers and that the full response to such change is realized in subsequent periods. Also, results indicate that demand for air passenger travel has not fully recovered from the terrorist attacks on September 11, 2001. Simulation analysis indicates that overall demand for air passenger travel to Florida is more sensitive to changes in income and that wildfires have no significant impact on demand for air passenger travel to Florida.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jose Cazanova.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Ward, Ronald W.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022668:00001


This item has the following downloads:


Full Text





DEMAND FOR AIR PASSENGER TRAFFIC AND ITS IMPACT ON THE TOURISM
INDUSTRY OF FLORIDA

















By

JOSE CAZANOVA


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2008

































2008 Jose Cazanova



































To my Mom and Dad.









ACKNOWLEDGMENTS

First and foremost, I thank God for the opportunity to pursue and complete this doctoral

program. I thank my family for their undying love and patience during this long academic

journey that started on a sunny day in Zamorano, Honduras ten years ago. I would like to thank

Mom and Dad for their unwavering support and great sacrifices made throughout. I could not

have asked for better parents. I also extend my deepest appreciation to my brothers, Mauricio

and Levi who have also made sacrifices to help me achieve this goal. I also thank Nini, my little

booboo, for making the days more enjoyable.

There are no words to express my gratitude and overall admiration to my chair, Dr. Ronald

W. Ward, a great mentor and even better person. He has been the best mentor and source of

inspiration someone could ask for during this immense task. I remember when we first talk about

the research idea and its complexity. He told me that it would be a lot of work that will require

great dedication. I took the challenging task because I knew that, with his help, the project was

going to be very rewarding. I greatly appreciate his hard work, excellent advice, and well-timed

guidance during my research. I am very grateful for his words of encouragement every time I felt

overwhelmed by the immensity of this research. All my struggles were lessened by his patience

and help. I would also like to thank him for giving me the opportunity to pursue my internship in

Ethiopia and also for encouraging me to study abroad in Spain. It has been a great honor to work

with him.

I also thank Dr. Richard Kilmer, who always seemed to have a comment of support when

he saw me struggle in the office. I am grateful for his valuable feedback given in the dissertation.

I have also enjoyed our great conversations on Gator sports. I would also like to thank Dr. Robert

Emerson and Dr. Stephen Holland. Their time and effort spent on my dissertation are greatly









appreciated. My appreciation also extends to Dr. Antonio Flores for his feedback on my

dissertation that has enabled me to improve its quality.

I also thank Dr. Burkhardt. I cannot thank him enough for his great support and advice to

pursue the internship in Ethiopia and the study abroad in Spain. When things seemed to be

impossible, he found a way to make sure I had these great opportunities to travel the world.

My appreciation also is extended to Dr. Jane Luzar. Without her help, I might never have

had the opportunity to pursue the PhD. I am thankful for her constant support and dedication she

has offered to every Zamorano who has knocked on her door.

I would like to thank all of my friends who have helped me while writing my dissertation.

These include my dear friends from the Food and Resource Economics Department and

Gainesville and to those who are living across the United States. My appreciation also extends to

my friends living in Honduras, Bolivia, Ethiopia, and Spain who have always sent me e-mails

expressing their great support. I would like to extend a special thank you to Marcos G. who has

always expressed nothing but sincere words of encouragement to me. I could not finish this

acknowledgement without expressing how proud I am to be a Florida Gator. Go Gators!









TABLE OF CONTENTS



A C K N O W L E D G M E N T S ..............................................................................................................4

L IST O F T A B L E S ........ ...... .... .......... .......... ............................................. 10

LIST OF FIGURES ................................... .. .... .... ................. 12

A B S T R A C T ............ ................... .................. ........................................ 2 2

CHAPTER

1 INTRODUCTION ............................... ... .. .... ..... ..................23

A airline Transportation and Tourism ............................................... ............................ 24
Problem Statement.................. ................... ............................... ........ 26
Research Objectives............................ .. .. .... .... ..................26
R research M methodology .............................. .............................................. ............ 27
Airline Passenger Traffic Partial Adjustment Model ............................................... 28
D ata an d S co p e ..............................................................................2 9

2 L ITE R A TU R E R E V IE W ............................................................................. .....................32

Forecasting Tourism D em and.............................................. .................. ............... 32
Scope of Recent Tourism Demand Studies ....................................................35
Econometric Techniques Applied to Tourism Demand Analysis .......................................38
D eterm inants of T ourism D em and .............................................................. .....................4 1
Static versus Dynam ic Regression M odels....................................... .......................... 45
Challenges in Tourism D em and Analysis ........................................ ......................... 47
Recent Developments in Tourism Demand Analysis.......................... .................51
Chapter Summary ................................ ... .. .... ..... ................... 52

3 DESCRIPTIVE STATISTICS ON THE AIRLINE PASSENGER TRAFFIC
TRAVELING TO FLORIDA AND OTHER INDICATORS ...........................................53

Airline Passenger Traffic ........................................................... ...... ................. 54
Description, Selection, and Aggregation of Domestic Airline Passenger Traffic
D ata ............................ .... ......... ......... ....... ....... .. ...................... 5 4
Description, Selection, and Aggregation of International Airline Passenger Traffic
D ata ............................ .... ....... .... ....... .............. ... .... ..... ............. 5 6
Domestic and International Airline Passenger Traffic Traveling to Florida...................57
Domestic airline passenger traffic by U.S. region............................... ...............58
Domestic airline passenger traffic by destination CSA .......................................61
South Florida CSA domestic airline passenger traffic..........................................62
Orlando CSA domestic airline passenger traffic............................................. 65
Tampa-St. Petersburg CSA domestic airline passenger traffic.............................67









Jacksonville CSA domestic airline passenger traffic ............................................70
Fort Myers CSA domestic airline passenger traffic...........................................72
International airline passenger traffic by world region ........................................75
International airline passenger traffic by destination CSA ......................................77
South Florida CSA international airline passenger traffic ....................................78
Orlando CSA international airline passenger traffic................... .......................81
Tampa-St. Petersburg CSA international airline passenger traffic ........................83
A airline Ticket Prices ......... ... ......... ......... .......................... ........ ................. .86
Description, Selection, and Aggregation of Airline Ticket Price Data for Domestic
F lig h ts .................................. ....... ....... ................................ ............... 8 6
Description, Selection, and Aggregation of Airline Ticket Price Data for
International Flights .................. ................................. .... .... .. ........ .... 88
Domestic Airline Ticket Prices by U.S. Region..... .......... .................................... 89
Domestic Airline Ticket Prices by Destination CSA............................................... 92
South Florida CSA domestic airline ticket prices ......................................... 93
Orlando CSA domestic airline ticket prices ........... ............................................96
Tampa-St. Petersburg CSA domestic airline ticket prices ....................................99
Jacksonville CSA domestic airline ticket prices .........................................102
Fort Myers CSA domestic airline ticket prices ...................................105
International Airline Round Trip Airline Ticket Prices by World Region ...................108
International Airline Round Trip Airline Ticket Prices by Destination CSA ...............110
South Florida CSA international airline ticket prices ...........................................111
Orlando CSA international airline ticket prices ...........................................113
Tampa-St. Petersburg CSA international airline ticket prices .............................114
Freight and Mail Transported via Commercial Passenger Airlines to Florida .................116
Economic, Social, and W weather Indicators................................................. ..... ............... 127
Gross Domestic Product, Personal Disposable Income, and Population ....................127
Brand and Generic Advertising Expenditures......... ............ ............................. ......131
Description, selection, and aggregation of advertising expenditures data .............131
Brand advertising expenditures................................................ ............... 132
G eneric advertising expenditures ................................................... ............... 136
Foreign Exchange Rate: Euro to U.S. Dollar ..... ...............................................138
H historic Jet Fuel Prices ........................ ..................................... .. ................ 139
Hurricanes and Wildfires Affecting Florida......................... ..... ............. 140
Average Temperatures in Origin Regions and Destination CSAs .............................144
Precipitation in Florida ................................................ .. ...... .. ........ .... 145
C rim e R ates in F lorida ................................................................................. ......... 14 6
Chapter Summary ................................. .. ... .... .................. 147

4 THEORETICAL FRAM EW ORK .......................................................... ............... 149

M otiv action ................................................................................................149
The Partial A djustm ent M odel..................................................... ................................. 152
Domestic Air Passenger Traffic Partial Adjustment Model.............................................. 154
Construction of Empirical Domestic Air Passenger Model ....................... ..................156
Description of Variables in the Static Component of the DAP-PAM ...........................157
Identification of the Variables in the Dynamic Component of the DAP-PAM ............161









E stim action P ossibilities...................................................................................... .......... 166
Chapter Summary ................................. .. ... .... .................. 169

5 R E S U L T S ..........................................................................17 0

Com prison of the Estim ation Alternatives................................... .................................... 170
Results for Demand for Passengers by CSA ............................................. ............... 171
F lo rid a-R esu lts ....................................................................................................... 17 3
Analysis of coefficients in dynamic component ...................................................173
Analysis of coefficients in static component............................. 173
South Florida C SA -R results .................................................. .............. ............... 174
Analysis of coefficients in dynamic component ...............................................175
Analysis of coefficients in static component............................. 175
Orlando CSA-Results ........ .... .. ........... .............. ........................................ 176
Analysis of coefficients in dynamic component ...............................................176
Analysis of coefficients in static component............................. 177
Tam pa-St. Petersburg CSA -R results ........................................ ......................... 178
Analysis of coefficients in dynamic component ...............................................178
Analysis of coefficients in static component............................. 178
Jacksonville C SA -R results ........................................... ....................................... 179
Analysis of coefficients in dynamic component .......................................... 180
Analysis of coefficients in static component............................. 180
F ort M years C SA -R results ....................................................................... ..................18 1
Analysis of coefficients in dynamic component ...............................................181
Analysis of coefficients in static component............................. 182
Chapter Summary ................................. .. ... .... ...................83

6 SIM ULATION AN ALY SIS ....................................................... ......... ......194

In tro du ctio n ................... ...................1.............................4
Sim ulations for Incom e............ .................................................. .............. .......... ....... 197
Florida CSA-Incom e Simulations ...........................................................................197
South Florida CSA-Income Simulations ........... ................................. ...............201
Orlando C SA -Incom e Sim ulations........................................................... .. ................ 204
Tampa-St. Petersburg CSA-Income Simulations.........................................................207
Jacksonville CSA -Incom e Sim ulations ........................................ ...... ............... 210
Fort M years CSA-Income Simulations................................ ......................... ....... 213
Sim ulations for A airline Ticket Prices .................................. ............... ............... 216
Florida CSA-Airline Ticket Price Simulations..........................................................217
South Florida CSA-Airline Ticket Price Simulations ....................................... 219
Tampa-St. Petersburg CSA-Airline Ticket Price Simulations .....................................221
Jacksonville CSA-Airline Ticket Price Simulations ............................................. 223
Fort M years CSA-Airline Ticket Price Simulations .............................................225
Simulations for Terror ............... .. ..... ...................... ........... 226
Florida CSA-Terror Simulations ............................................................................227
South Florida CSA-Terror Simulations..................................................................... 229
O rlando C SA -Terror Sim ulations...................................................................... ...... 231


8









Tampa-St. Petersburg CSA-Terror Simulations...............................................233
Jacksonville CSA -Terror Sim ulations.......... ..................................... ..... ............. 235
Fort Myers CSA-Terror Simulations................... ............................ 237
Sim ulations for H hurricanes ........................................... ........................... ............... 239
Florida CSA -H hurricane Sim ulations ........................................ ........................ 239
South Florida CSA-Hurricane Simulations....................................... ..............242
Orlando CSA-Hurricane Simulations........................................... ......... ... ............... 244
Tampa-St. Petersburg CSA-Hurricane Simulations......... ..... ...........246
Jacksonville C SA -H hurricane Sim ulations ........................................ .....................248
Fort Myers CSA-Hurricane Simulations...... ................. ...............249
Sim ulations for Seasonality ....................................................................... .....................250
F lorida C SA -Seasonality .................................................................. ...... .................250
South Florida C SA -Seasonality ......................................................................... ...... 252
O rlando C SA -Seasonality ..................................................................... ..................254
Tampa-St. Petersburg CSA-Seasonality................... ...... ........................ 256
Jacksonville C SA -Seasonality ............................................................ ............. .258
Fort M years C SA -Seasonality .................................................................... ............... 260
Simulations for Fire, Rainfall, and Temperature ............ ............................................262
Chapter Summary ................................. .. ... .... .................. 262

7 SUMMARY, CONCLUSIONS, IMPLICATIONS, AND FUTURE RESEARCH ............268

C o n c lu sio n s ........................................................................................................................... 2 6 9
Im p lic atio n s .................................................................................................................... 2 7 4
F u tu re R e se arch .............................................................................................................. 2 7 6

APPENDIX

A DESCRIPTION OF DESTINATION CSA AIRPORTS...........................................278

B TRANSFORMATION OF VARIABLES: TSP PROGRAMS ............. .......................279

C ESTIMATION OF MODEL: TSP PROGRAMS................... ..........................283

D SIMULATION ANALYSIS: TSP PROGRAMS..........................................................284

E POPULATION FACTORS BY U.S. REGION................................... ...............292

L IST O F R E FE R E N C E S ......... ................. ............................................... ..............................294

B IO G R A PH IC A L SK E T C H ............................................................................. ....................298










9









LIST OF TABLES


Table page

3-1 Geographic region scheme as defined by the United States Census. .............................55

4-1 Summary of the variables included in the DAP-PAM................... ...............163

4-2 Identification of destination CSAs and origin U.S. regions............... .. ............... 165

5-1 Coefficient estimates of 129 and their corresponding t-value using five different
estimation approaches.......................................................... 87

5-2 Florida CSA: coefficient estimates and their corresponding t-values for the demand
for air passengers traveling from four U.S. regions using the SUR-AR1-ALL
approach ................................ ..................... ............. ................ 188

5-3 South Florida CSA: coefficient estimates and their corresponding t-values for the
demand for air passengers traveling from four U.S. regions using the SUR-AR1-ALL
approach ................................ ..................... ............. ................ 189

5-4 Orlando CSA: coefficient estimates and their corresponding t-values for the demand
for air passengers traveling from four U.S. regions using the SUR-AR1-ALL
approach ................................ ..................... ............. ................ 190

5-5 Tampa-St. Petersburg CSA: coefficient estimates and their corresponding t-values
for the demand for air passengers traveling from four U.S. regions using the SUR-
A R 1-A LL approach. .......................... ...... ..................... .... .............. 191

5-6 Jacksonville CSA: coefficient estimates and their corresponding t-values for the
demand for air passengers traveling from four U.S. regions using the SUR-AR1-ALL
approach ................................ ..................... ............. ................ 192

5-7 Fort Myers CSA: coefficient estimates and their corresponding t-values for the
demand for air passengers traveling from four U.S. regions using the SUR-AR1-ALL
approach ................................ ..................... ............. ................ 193

6-1 Comparison between the simulated demand in the presence of the 9-11 terrorist
attacks and the simulated demand in the absence of 9-11 terrorist attacks. ...................266

6-2 Comparison between the simulated demand in the presence of hurricanes and the
simulated demand in the absence hurricanes during the hurricane season (June-
N ov em b er). ............................................................................ 267

E-1 Northeast region (1001): population factors by month and by year between 1996 and
2006 .............. ...................... ...................................... ......... ...... 292









E-2 Midwest region (1002): population factors by month and by year between 1996 and
2006 ............ ......................... ...................................... ......... ...... 292

E-3 South region (1003): population factors by month and by year between 1996 and
2006 ............ ......................... ...................................... ......... ...... 293

E-4 West region (1004): population factors by month and by year between 1996 and
2006 ............ ......................... ...................................... ......... ...... 293









LIST OF FIGURES


Figure page

1-1 Share of U.S. domestic airline passenger traffic by state in 2006. ..............................25

3-1 Monthly seasonal pattern of domestic and international airline passenger traffic
traveling to Florida between 1990 and 2006................................. ...............58

3-2 Florida CSA: total domestic airline passenger traffic from five U.S. regions between
1990 and 2006....................................................... ................... ... ....... ....... 60

3-3 Florida CSA: share of total domestic airline passenger traffic from five U.S. regions
in 1990, 1998, and 2006.................... .. ............... ....... ......... 60

3-4 Florida CSA: monthly seasonal pattern of domestic airline passenger traffic from
four U.S. regions between 1990 and 2006. ......................... ..................................... 61

3-5 Total domestic airline passenger traffic traveling to the destination CSAs in Florida
betw een 1990 and 2006. .......................... ...... ................................... .. .....62

3-6 South Florida CSA: total domestic airline passenger traffic from five U.S. regions
betw een 1990 and 2006. .......................... ...... ................................... .. .....63

3-7 South Florida CSA: share of total domestic airline passenger traffic from five U.S.
regions in 1990, 1998, and 2006 ............................................... ............................. 64

3-8 South Florida CSA: monthly seasonal pattern of domestic airline passenger traffic
from four U.S. regions between 1990 and 2006. ..................................................64

3-9 Orlando CSA: total domestic airline passenger traffic from five U.S. regions between
1990 and 2006....................................................... ................... ... ....... ....... 66

3-10 Orlando CSA: share of total domestic airline passenger traffic from five U.S. regions
in 1990, 1998, and 2006.................... .. ............... ....... ......... 66

3-11 Orlando CSA: monthly seasonal pattern of domestic airline passenger traffic from
four U.S. regions between 1990 and 2006. ......................... ..................................... 67

3-12 Tampa-St. Petersburg CSA: total airline passenger traffic from five U.S. regions
betw een 1990 and 2006. ......................... ...... ................................... .. .....68

3-13 Tampa-St. Petersburg CSA: share of total domestic airline passenger traffic from
five U .S. regions in 1990, 1998, and 2006.................................... ......................... 69

3-14 Tampa-St. Petersburg CSA: monthly seasonal pattern of domestic airline passenger
traffic from four U.S. regions between 1990 and 2006. ................ ............... ..............69









3-15 Jacksonville CSA: total domestic airline passenger traffic from five U.S. regions
betw een 1990 and 2006. ......................... ...... ................................... .. .....71

3-16 Jacksonville CSA: share of total domestic airline passenger traffic from four U.S.
regions in 1990, 1998, and 2006 ............................................... ............................. 71

3-17 Jacksonville CSA: monthly seasonal pattern of domestic airline passenger traffic
from four U.S. regions between 1990 and 2006. .............. ........ ................... 72

3-18 Fort Myers CSA: total domestic airline passenger traffic from five U.S. regions
betw een 1990 and 2006. ......................... ...... ................................... .. .....73

3-19 Fort Myers CSA: share of total domestic airline passenger traffic from four U.S.
regions in 1990, 1998, and 2006 ............................................... ............................. 74

3-20 Fort Myers CSA: monthly seasonal pattern of domestic airline passenger traffic from
four U.S. regions between 1990 and 2006. ........................... ..................................... 74

3-21 Florida CSA: total international airline passenger traffic from four world regions
betw een 1990 and 2006. ......................... ...... ................................... .. .....76

3-22 Florida CSA: share of total international airline passenger traffic from four world
regions in 1990, 1998, and 2006 ............................................... ............................. 76

3-23 Florida CSA: monthly seasonal pattern of total international airline passenger traffic
from four world regions between 1990 and 2006. ............................... ................77

3-24 Total international airline passenger traffic traveling to the top three destination CSAs
in Florida betw een 1990 and 2006 .......................................................................... .... 78

3-25 South Florida CSA: total international airline passenger traffic from four world
regions betw een 1990 and 2006............................................... .............................. 79

3-26 South Florida CSA: share of total international airline passenger traffic from four
w orld regions in 1990, 1998, and 2006........................................ .......................... 80

3-27 South Florida CSA: monthly seasonal pattern of total international airline passenger
from four world regions between 1990 and 2006. ............................... ................80

3-28 Orlando CSA: total international airline passenger traffic from four world regions
betw een 1990 and 2006. ......................... ...... ................................... .. .....82

3-29 Orlando CSA: share of total international airline passenger traffic from four world
regions in 1990, 1998, and 2006 ............................................... ............................. 82

3-30 Orlando CSA: monthly seasonal pattern of total international airline passenger traffic
from four world regions between 1990 and 2006................................ ................83









3-31 Tampa-St. Petersburg CSA: total international airline passenger traffic from four
world regions between 1990 and 2006. ............ ....................................... .....................84

3-32 Tampa-St. Petersburg CSA: share of total international airline passenger traffic from
four world regions in 1990, 1998, and 2006. ............................... ............................... 85

3-33 Tampa-St. Petersburg CSA: monthly seasonal pattern of total international airline
passenger traffic from four world regions between 1990 and 2006 ...............................85

3-34 Florida CSA: average one way airline ticket prices for domestic flights from four
U .S. regions betw een 1993 and 2006.......................................... ........................... 90

3-35 Florida CSA: average round trip airline ticket prices for domestic flights from four
U .S. regions betw een 1993 and 2006.......................................... ........................... 91

3-36 Florida CSA: quarterly seasonal pattern of average one way (OW) and round trip
(RT) airline ticket prices for domestic flights from four U.S. regions between 1993
an d 2 0 0 6 ............... .............. .............. ................................................9 2

3-37 South Florida CSA: average one way airline ticket prices for domestic flights from
four U.S. regions between 1993 and 2006. ..............................................94

3-38 South Florida CSA: average round trip air ticket prices for domestic flights from four
U .S. regions betw een 1993 and 2006.......................................... ........................... 95

3-39 South Florida CSA: quarterly seasonal pattern of average one way (OW) and round
trip (RT) airline ticket prices for domestic flights from four U.S. regions between
19 9 3 an d 2 0 0 6 ......... ................... ......... ........ .........................................9 6

3-40 Orlando CSA: average one way airline ticket prices for domestic flights from four
U .S. regions betw een 1993 and 2006.......................................... ........................... 97

3-41 Orlando CSA: average round trip air ticket prices for domestic flights from four U.S.
regions betw een 1993 and 2006............................................... .............................. 98

3-42 Orlando CSA: quarterly seasonal pattern of average one way (OW) and round trip
(RT) airline ticket prices for domestic flights from four U.S. regions between 1993
an d 2 0 0 6 ............... .............. .............. ................................................9 9

3-43 Tampa-St. Petersburg CSA: average one way airline ticket prices for domestic flights
from four U.S. regions between 1993 and 2006 ........................................ ..........100

3-44 Tampa-St. Petersburg CSA: average round trip air ticket prices for domestic flights
from four U.S. regions between 1993 and 2006 ........................................ ..........101

3-45 Tampa-St. Petersburg CSA: quarterly seasonal pattern of average one way (OW) and
round trip (RT) airline ticket prices for domestic flights from four U.S. regions
between 1993 and 2006. ........................... ........ .. .. ...... ............ 102









3-46 Jacksonville CSA: average one way airline ticket prices for domestic flights from
four U.S. regions between 1993 and 2006 .................................................................... 103

3-47 Jacksonville CSA: average round trip air ticket prices for domestic flights from four
U .S. regions betw een 1993 and 2006........................................ ........................... 104

3-48 Jacksonville CSA: quarterly seasonal pattern of average one way (OW) and round
trip (RT) airline ticket prices for domestic flights from four U.S. regions between
1993 and 2006........................................................................ ... .... .. .... 105

3-49 Fort Myers CSA: average one way airline ticket prices for domestic flights from four
U .S. regions betw een 1993 and 2006........................................ ........................... 106

3-50 Fort Myers CSA: average round trip air ticket prices for domestic flights from four
U .S. regions betw een 1993 and 2006........................................ ........................... 107

3-51 Fort Myers CSA: quarterly seasonal pattern of average one way (OW) and round trip
(RT) airline ticket prices for domestic flights from four U.S. regions between 1993
an d 2 0 0 6 ............... .............. ............. ............................................... 10 8

3-52 Florida CSA: average international round trip airline ticket prices from three world
regions betw een 1995 and 2006............................................... ............. ............... 109

3-53 Florida CSA: monthly seasonal pattern of relative change from the average round
trip airline ticket prices from three world regions between 1995 and 2006. .................110

3-54 South Florida CSA: average international round trip airline ticket prices from three
world regions between 1995 and 2006. ........................................................ ............. .111

3-55 South Florida CSA: monthly seasonal pattern of relative change from the average
round trip airline ticket prices from three world regions between 1995 and 2006..........112

3-56 Orlando CSA: average international round trip airline ticket prices from three world
regions betw een 1995 and 2006................................................................................ ... 113

3-57 Orlando CSA: monthly seasonal pattern of relative change from the average round
trip airline ticket prices from three world regions between 1995 and 2006. .................114

3-58 Tampa-St. Petersburg CSA: average international round trip airline ticket prices from
three world regions between 1995 and 2006 ............................................................. 115

3-59 Tampa-St. Petersburg CSA: monthly seasonal pattern of the relative change from the
average round trip airline ticket prices from three world regions between 1995 and
2 006 .............. ...................... ...................................... ......... ...... 116

3-60 Florida CSA: total domestic freight transported from four U.S. regions using
commercial passenger airlines between 1990 and 2006. ................................................118









3-61 Florida CSA: monthly seasonal pattern of total freight transported from four U.S.
regions using commercial passenger airlines between 1990 and 2006............................118

3-62 Total domestic freight transported to six destination CSAs in Florida using
commercial passenger airlines between 1990 and 2006........ ............ ................119

3-63 Florida CSA: total domestic mail transported from four U.S. regions using
commercial passenger airlines between 1990 and 2006........ ............ ................120

3-64 Florida CSA: monthly seasonal pattern of total mail transported from four U.S.
regions using commercial passenger airlines between 1990 and 2006............................121

3-65 Total domestic mail transported to six destination CSAs in Florida using commercial
passenger airlines between 1990 and 2006. ....................................... ...............122

3-66 Florida CSA: total international freight transported from three world regions using
commercial passenger airlines between 1990 and 2006........ ............ ................123

3-67 Florida CSA: monthly seasonal pattern of total freight transported from three world
regions using commercial passenger airlines between 1990 and 2006............................123

3-68 Total international freight transported to six destination CSAs in Florida using
commercial passenger airlines between 1990 and 2006........ ............ ................124

3-69 Florida CSA: total international mail transported from four world regions using
commercial passenger airlines between 1990 and 2006........ ............ ................125

3-70 Florida CSA: monthly seasonal pattern of total mail transported from three world
regions using commercial passenger airlines between 1990 and 2006............................126

3-71 Total international mail transported to six destination CSAs in Florida using
commercial passenger airlines between 1990 and 2006.............................................127

3-72 Annual gross domestic product from four U.S. regions between 1990 and 2006...........129

3-73 Annual per capital personal disposable income from four U.S. regions between 1990
an d 2 0 0 6 ............... .............. ............. ............................................... 13 0

3-74 Annual population estimates from four U.S. regions between 1990 and 2006 .............130

3-75 Total brand advertising expenditures from four tourism-related companies in Florida
betw een 1995 and 2006. ........................................... .. .... ........... ....... 133

3-76 Florida attractions: monthly seasonal pattern of total brand adverting expenditures
from three tourism-related companies between 1995 and 2006 .....................................134

3-77 Florida and non-Florida attractions: monthly seasonal pattern of total brand adverting
expenditures from three tourism-related companies between 1995 and 2006...............135









3-78 Total generic advertising expenditures spent by government entities from Florida
betw een 2002 and 2006. ............................................. .........................136

3-79 Monthly seasonal pattern of total generic adverting expenditures spent by city,
county, and state government to promote Florida between 2002 and 2006. .................137

3-80 Annual air passenger traffic traveling from Europe to Florida and annual exchange
rate (Euro to USD) between 1990 and 2006. ................. ...... ................138

3-81 Historic average kerosene-type jet fuel prices from five worldwide locations between
1990 and 2006........................................................................ ... ....... ....... 139

3-82 Monthly seasonal pattern of historic average kerosene-type jet fuel prices from five
worldwide locations between 1990 and 2006 .......................................... .............140

3-83 Number of tropical storms (by category) affecting Florida between 1990 and 2006......141

3-84 Number of wildfires (by category) affecting Florida between 1990 and 2006..............143

3-85 Percentage of wildfires and acreage burned in Florida (by month) between 1990 and
2 006 .............. ...................... ...................................... ......... ..... 14 3

3-86 Difference between average temperatures in Florida and each U.S. region between
1990 and 2006 ............ ............................................................. ....... ....... 145

3-87 Monthly average rainfall in each of the five destination CSAs in Florida between
1990 and 2006 ........................ ......... ..... ....... ................ ............... 146

3-88 Annual crime index in Florida between 1990 and 2006............................................147

4-1 Power spectrum of domestic air passenger traffic data for the Florida-Northeast (33-
100 1) pair. ................................................................................ 162

5-1 Florida CSA: effect of terrorism dummies on demand for airline passengers from
four U .S regions. ...................................................... ................. 184

5-2 South Florida CSA: effect of terrorism dummies on demand for airline passengers
from four U .S. regions. ............................... .......................................... ...... ............1.. 84

5-3 Orlando CSA: effect of terrorism dummies on demand for airline passengers from
four U .S. regions. .......................................... ........................... 185

5-4 Tampa-St. Petersburg CSA: effect of terrorism dummies on demand for airline
passengers from four U .S. regions. .......... ............................. ........... ............ 185

5-5 Jacksonville CSA: effect of terrorism dummies on demand for airline passengers
from four U .S. regions. ........................... ............. .. .. .......... ........ ....... ........186









5-6 Fort Myers CSA: effect of terrorism dummies on demand for airline passengers from
four U .S. regions. .......................................... ........................... 186

6-1 Florida CSA-Income simulations: relationship between different levels of personal
disposable income and domestic demand for air passengers from four U.S. regions. ....199

6-2 Florida CSA-Income simulations: short and long run changes in domestic demand
for air passengers from four U.S. regions at different levels of per capital personal
disposable income..................... .......................... ........ 200

6-3 South Florida CSA-Income simulations: relationship between different levels of
personal disposable income and number of passengers for each of the four U.S.
reg ion s ................... .......................................................... ................ 2 02

6-4 South Florida CSA-Income simulations: short and long run changes in the number of
passengers at different levels of per capital personal disposable income for each of the
four U .S. regions. .......................................... ........................... 203

6-5 Orlando CSA-Income simulations: relationship between different levels of personal
disposable income and number of passengers for each of the four U.S. regions. ...........205

6-6 Orlando CSA-Income simulations: short and long run changes in the number of
passengers at different levels of per capital personal disposable income for each of the
four U .S. regions. .......................................... ........................... 206

6-7 Tampa-St. Petersburg CSA-Income simulations: relationship between different
levels of personal disposable income and number of passengers for each of the four
U .S re g io n s ...................................... .................................................. 2 0 8

6-8 Tampa-St. Petersburg CSA-Income simulations: short and long run changes in the
number of passengers at different levels of per capital personal disposable income for
each of the four U .S. regions. ................................................................. ..................... 209

6-9 Jacksonville CSA-Income simulations: relationship between different levels of
personal disposable income and number of passengers for each of the four U.S.
reg io n s ................... .......................................................... ................ 2 1 1

6-10 Jacksonville CSA-Income simulations: short and long run changes in the number of
passengers at different levels of per capital personal disposable income for each of the
four U .S. regions. .......................................... ........................... .. 212

6-11 Fort Myers CSA-Income simulations: relationship between different levels of
personal disposable income and number of passengers for each of the four U.S.
reg io n s ................... .......................................................... ................ 2 14

6-12 Fort-Myers CSA-Income simulations: short and long run changes in the number of
passengers at different levels of per capital personal disposable income for each of the
four U .S. regions. .......................................... ............................ 215









6-13 Florida CSA-Airline ticket price simulations: A) Relationship between different
levels of airline ticket prices and number of passengers; and B) Short and long run
changes in the number of passengers at different levels of airline ticket prices for two
U .S. regions.............. .......... ........................... ..........................................2 18

6-14 South Florida CSA-Airline ticket price simulations: A) Relationship between
different levels of airline ticket prices and number of passengers; and B) Short and
long run changes in the number of passengers at different levels of airline ticket
prices for three U .S. regions. ................................................ ............................... 220

6-15 Tampa-St. Petersburg CSA-Airline ticket price simulations: A) Relationship between
different levels of airline ticket prices and number of passengers; and B) Short and
long run changes in the number of passengers at different levels of airline ticket
prices for tw o U .S. regions. ..................................................................... .................. 222

6-16 Jacksonville CSA-Airline ticket price simulations: A) Relationship between different
levels of airline ticket prices and number of passengers; and B) Short and long run
changes in the number of passengers at different levels of airline ticket prices for two
U .S. regions.............. .......... ........................... ..........................................224

6-17 Fort Myers CSA-Airline ticket price simulations: A) Relationship between different
levels of airline ticket prices and number of passengers; and B) Short and long run
changes in the number of passengers at different levels of airline ticket prices for the
N northeast region............ ... .................... ......... ................ ............. 225

6-18 Florida CSA-Terror simulations: comparison between the simulated results of
demand for air passengers from four U.S. regions in the presence and absence of the
9-11 terrorist attacks. .................................. .. .. ........ .. ............228

6-19 South Florida CSA-Terror simulations: comparison between the simulated results of
demand for air passengers from four U.S. regions in the presence and absence of the
9-11 terrorist attacks. .................................. .. .. ........ .. ............230

6-20 Orlando CSA-Terror simulations: comparison between the simulated results of
demand for air passengers from four U.S. regions in the presence and absence of the
9-11 terrorist attacks. .................................. .. .. ........ .. ............232

6-21 Tampa-St. Petersburg CSA-Terror simulations: comparison between the simulated
results of the demand for passengers from four U.S. regions in the presence and
absence of the 9-11 terrorist attacks....................................................................... ... 234

6-22 Jacksonville CSA-Terror simulations: comparison between the simulated results of
the demand for air passengers from four U.S. regions in the presence and absence of
the 9-11 terrorist attacks........... ............................................................ ....... .............. 236

6-23 Fort Myers CSA-Terror simulations: comparison between the simulated results of
the demand for passengers from four U.S. regions in the presence and absence of the
9-11 terrorist attacks. .................................. .. .. ........ .. ............238









6-24 Florida CSA-Hurricane simulations: short and long run changes in the number of
passengers from four U.S. regions in the absence of hurricanes in Florida.....................241

6-25 South Florida CSA-Hurricane simulations: short and long run changes in the number
of passengers from four U.S. regions in the absence of hurricanes in Florida. ...............243

6-26 Orlando CSA-Hurricane simulations: short run and long run changes in the number
of passengers from four U.S. regions in the absence of hurricanes in Florida. ...............245

6-27 Tampa-St. Petersburg CSA-Hurricane simulations: short and long run changes in the
number of passengers from four U.S. regions in the absence of hurricanes in Florida...247

6-28 Jacksonville CSA-Hurricane simulations: short and long run changes in the number
of passengers from four U.S. regions in the absence of hurricanes in Florida. ...............248

6-29 Fort Myers CSA-Hurricane simulations: short and long run changes in the number of
passengers from four U.S. regions in the absence of hurricanes in Florida.....................249

6-30 Florida CSA-Seasonality simulations: A) Monthly seasonal pattern of domestic
demand for air passengers; and B) Percentage change from the monthly average of
domestic demand for air passengers from four U.S. regions................. .....................251

6-31 South Florida CSA-Seasonality simulations: monthly seasonal pattern of domestic
demand for air passengers from four U.S. regions. ........................................................252

6-32 South Florida CSA-Seasonality simulations: percentage change from the monthly
average demand for passengers from four U.S. regions. ............................................253

6-33 Orlando CSA-Seasonality simulations: monthly seasonal pattern of domestic demand
for air passengers from four U.S. regions .......................................................................254

6-34 Orlando CSA-Seasonality simulations: percentage change from the monthly average
domestic demand for air passengers from four U.S. regions................. .....................255

6-35 Tampa-St. Petersburg CSA-Seasonality simulations: monthly seasonal pattern of
domestic demand for air passengers from four U.S. regions................. .....................256

6-36 Tampa-St. Petersburg CSA-Seasonality simulations: percentage change from the
monthly average domestic demand for air passengers from four U.S. regions...............257

6-37 Jacksonville CSA-Seasonality simulations: monthly seasonal pattern of domestic
demand for air passengers from four U.S. regions. ........................................................258

6-38 Jacksonville CSA-Seasonality simulations: percentage change from the monthly
average domestic demand for air passengers from four U.S. regions. ............................259

6-39 Fort Myers CSA-Seasonality simulations: monthly seasonal pattern of domestic
demand for air passengers from four U.S. regions. ........................................................260









6-40 Fort Myers CSA-Seasonality simulations: percentage change from the monthly
average domestic demand for air passengers from four U.S. regions. ............................261

6-41 Comparison of the demand response from each U.S. region traveling to Florida to a
three percent increase in income and airline ticket prices. ............................................263

6-42 Comparison of the demand response in each CSA-ORG pair to a three percent
increase in income e and airline ticket prices. ........................................ .....................264









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

DEMAND FOR AIR PASSENGER TRAFFIC AND ITS IMPACT ON THE TOURISM
INDUSTRY OF FLORIDA

By

Jose Cazanova

August 2008

Chair: Ronald W. Ward
Major: Food and Resource Economics

The airline industry plays a key role in the tourism industry, serving as a vital link between

consumers and the tourism industry. Since nearly one half of the tourists coming to Florida use

air transportation, having quantitative methods for both explaining and forecasting air passenger

traffic are vital to planning and infrastructure development for the tourist sector and the state in

general. Also, knowing how air travel demand responds to unexpected shocks, such as

hurricanes, provides guidelines for emergency planning, risk assessment, and prevention

management as well as impacts on the state's economy. The primary objective of this research

was to develop an understanding of the factors influencing domestic demand for airline travel to

Florida. Factors such as prices, income, terrorism, seasonality, storms, wildfires, and advertising

expenditures were included in the analysis. Estimation results from the partial adjustment model

indicates that overall air passengers respond immediately up to a certain extent to changes in

some demand drivers and that the full response to such change is realized in subsequent periods.

Also, results indicate that demand for air passenger travel has not fully recovered from the

terrorist attacks on September 11, 2001. Simulation analysis indicates that overall demand for air

passenger travel to Florida is more sensitive to changes in income and that wildfires have no

significant impact on demand for air passenger travel to Florida.









CHAPTER 1
INTRODUCTION

Tourism has multiple effects on various areas such as the economy, society, culture, and

environment of a particular area or country. It has been used as a tool for economic development

in both rich and impoverished nations. The economic impact of tourism has been widely

documented throughout the years. Tourism has been a viable source of income and employment

for individuals and a source of revenue for local, state, and national governments. According to

the World Travel and Tourism Council (WTTC), the tourism industry directly employed 72

million people worldwide, accounting for 3.8% of the world's gross domestic product (GDP) in

2005. In the United States, direct tourism employment totaled 5.7 million people generating USD

195.08 billion in compensation in 2005.1 Moreover, 923,064 people worked in Florida

representing nearly 16% of the national tourism workforce in that year.

In 2006 Florida's economy, with a real GDP of USD 609.9 billion2, ranked fourth in the

country behind California, New York, and Texas. Furthermore, if analyzed as an independent

state, Florida would be the 20th largest economy in the world above countries such as Australia,

Netherlands, and Russia. Tourism share of the state's GDP totaled approximately USD 37.29

billion for that year and is expected to increase in coming years. In fact, the World Travel

Tourism Council (WTTC) predicts that by the year 2010 international tourists will exceed one

billion. More and more destinations will be available to tourists and could represent a threat or

even an opportunity to Florida's tourism industry.

Despite the positive economic effects derived from tourism (e.g., urban regeneration,

foreign exchange), some negative effects can be attributed to the industry. Fyall and Garrod


1 Travel and Tourism Satellite Accounts. BEA, 2005
2 Chained dollars (2000)









(2005) stated that tourism expansion will bring considerable pressure on the already fragile

natural and cultural environments upon which tourism relies. New challenges arise on how to

manage growth given the limited resources in areas such as transportation network, energy,

water and waste management, and health facilities. Tourists bring traffic congestion, which

means more gas consumption due to delays on highways. Also, airline travelers contribute to

noise pollution created by planes affecting neighborhoods close to airports. Tourists also attract

crime to the area since thieves may perceive that travelers carry valuable items and are less

cautious and hence more vulnerable when vacationing. There is a cost associated to each of these

negative effects of tourism but the exceptional growth of the industry suggests that the benefits

far outweigh the costs.

Airline Transportation and Tourism

The tourism industry can be divided into five main sectors: the attraction sector,

accommodation sector, transportation sector, travel organizer sector, and destination organization

sector (Vanhove 2005). The transportation sector plays an important role in the tourism system

as it serves as a vital link between consumers and the tourism industry. Airlines, railways, bus

and coach operators, car rental operators, and shipping lines are included in the transportation

sector. Both the tourism industry and the transportation sector show a high level of

interdependence which is especially true when describing the airline transportation sector.

Personal mobility has increased substantially worldwide over the past decades and airline

travel continues to become an increasingly important transport mode (Dargay and Hanly 2001).

Airline transportation has contributed to such increase in mobility and, therefore, to the

economic growth of many countries around the world.

Airline transportation is the means used by millions of individuals to get to their touristic

destination. Tourists have access to more destinations today than 10 years ago, and their








spending has improved the economy of many countries. Africa, the fastest growing destination

worldwide, has experienced an expansion of their domestic economy and employment due to

tourism. An estimated 675,000 people directly employed in tourism are supported by overseas

visitors arriving by air, representing 20% of all tourism jobs in Africa (WTTC 2007).

According to the WTTC, 40% of international tourists now travel by air, up from 35% in

1990. Surveys from Visit Floridac indicate that 42.6 million tourists arrived by air in Florida

during 2006, accounting for 50.8% of that year's total visitors. Florida airline passenger traffic is

composed by domestic airline passenger traffic and international passenger airline traffic. Both

are measured in number of passengers traveling through the state's commercial airports.

Domestic airline passenger traffic in Florida totaled 56,059,319 people in 2006 which accounts

for 8.5% of total U.S. domestic airline passenger traffic, as shown in Figure 1-1.


12.0
10.0
8.0
6.0
4.0
2.0
0.0


Cn C1 C1
C ) .2 .| g C )
C1 -00



State

Figure 1-1. Share of U.S. domestic airline passenger traffic by state in 2006. Source: Bureau of
Transportation Statistics.
International airline passenger traffic coming to Florida totaled 10,499,493 passengers in

2006 which accounts for 13.7% of total international airline passenger traffic arriving in the

United States.


~LL


- 4 ii









Problem Statement

The tourism industry plays an important role in Florida's economy. Tourism taxable sales

totaled USD 32.96 billion in 2006 making the tourism industry a major source of revenue and the

largest employer in the state. Moreover, Florida is one of the premier tourist destinations for

domestic and international visitors alike. More than one half of the domestic visitors that traveled

to Florida used airline transportation (Visit Florida" 2006). Additionally, close to 10.5 million

international passengers arrived in Florida using airline transportation in 2006. In order for

Florida to continue to be a top destination in the world, an understanding of the factors that affect

demand is vital.

Accurate understanding of the number of passengers traveling to Florida is critical in order

to achieve proper development. Measures of tourism demand could be used as a guide that will

help state officials design a strategic plan to meet the long run needs of the state. Also, an

effective allocation of resources is critical for a strategic plan that aims to develop new

infrastructure and renovate and expand current infrastructure (e.g., highways, airports), secure

resources such as water and energy, prevent crime, and lessen the negative effects of noise and

air pollution and transformation of land use.

Research Objectives

The general objective of this study is to develop an understanding of the factors

influencing domestic demand for airline travel to Florida. In order to achieve this general

objective, three specific objectives were set and are outlined as follows:

Develop an econometric model that helps explain the driving forces of demand for
domestic airline travel to Florida and to five combined statistical areas (CSAs) in Florida

Examine the factors that contribute to increasing domestic airline passenger traffic to
Florida and to five destination CSAs in Florida









Use the results from the econometric model (i.e., partial adjustment model) to conduct
simulations on specific shocks to the system to determine their effect on the demand for
domestic airline passenger traffic.

Research Methodology

Passenger data used in this study were collected by the Bureau of Transportation Statistics

and are available online. Data present number of passengers from each state (domestic) and

country (international) traveling to their ultimate destination, Florida. Data also include month

and year of travel which could be used to measure any seasonal patterns in the airline passenger

traffic traveling to Florida.

Given the size of the research project, this work was limited to estimating an econometric

model for the domestic airline passenger traffic traveling to Florida. Nevertheless, data related to

international passengers traveling to Florida and freight and mail sent via commercial passenger

airlines to Florida were analyzed and some general descriptive statistics are illustrated in Chapter

3. Such descriptive statistics give a general overview of international passenger traffic that will

allow the reader have an understanding of the size and importance of the market. Furthermore,

freight and mail are included to provide a complete overview of the industry. Theoretically one

could expect that freight and mail help defray the cost of a flight incurred by airlines and it is

expected to be reflected in the airfare variable included in the model.

Chapter 4 presents the theoretical framework of the partial adjustment model and also the

empirical application to air passenger demand traveling to Florida. The model, which includes a

static and a dynamic component, is an attractive approach since it yields a speed of adjustment

coefficient that shows how fast passengers adjust to events that affect their decision to travel to

Florida. An empirical model was introduced to show how domestic demand for air passenger

traffic could be represented and estimated. Economic theory was used to build the static

component, and a spectral analysis was performed to identify the dynamic component of the









model. Preliminary diagnostics were conducted to determine whether the data included in the

model are stationary (augmented Dickey-Fuller test). Also, the Durbin h-statistic hypothesis test

was performed to identify if the errors had an auto regressive scheme of order 1. These

diagnostics helped in determining whether ordinary least squares (OLS) estimates are consistent

or whether other estimation procedures were needed.

Airline Passenger Traffic Partial Adjustment Model

The partial adjustment model developed by Nerlove (1958) introduces explanatory

variables with the argument that current demand is not only driven from its past values but also

by consumer responses to exogenous factors such as income, crime and terrorism, advertising,

and weather. This model has been used to model demand in areas such as tourism, finance, and

agriculture. For example, Flannery and Rangan (2006) specified a partial adjustment model to

identify target capital structures in firms. Also, Ward and James (1978) used the partial

adjustment framework to determine the effect of coupons on consumption of frozen orange juice.

The partial adjustment model attempts to identify any factors that influence the demand for

airline passenger traffic traveling to Florida. As described by Pollak (1970), past consumption

patterns are an important determinant of present consumption patterns. Adamowicz (1994)

suggested that consumers may be "learning by doing" when they visit a site, and thus repeated

visitation occurs.

The partial adjustment model is used to determine whether rigidity exists in the demand by

domestic air passengers traveling to Florida. Assume that there is a desired airline passenger

traffic demand and it is a function of airline ticket prices, personal disposable income, gross

domestic product by U.S. region and dummies for hurricanes, wildfires, and terror. If the partial

adjustment parameter (0) equals zero, there is no partial adjustment and demand for airline

passenger traffic is purely determined by its past values. If 0 equals one, demand for airline









passenger traffic is determined by some explanatory variables, X, and the ARIMA model is no

longer relevant.

It is hypothesized that the adjustment coefficient(0) of the demand for air passenger traffic

to Florida lies between zero and one and that it varies between origin (U.S. region) and

destination (CSA). Thus, separate equations were specified for domestic airline passenger

demand according to its originating U.S. region: Northeast, Midwest, South, or West. There were

a total of 20 equations estimated under the Seemingly Unrelated Regression (SUR) approach.

The SUR modeling accounts for any correlation across equations and should be the appropriate

approach. If any partial adjustment is present in airline passengers traveling to Florida,

individuals will adjust somewhat to changes in demand drivers in the short run but the ultimate

full effect will be realized in subsequent periods.

Finally, estimates of the partial adjustment models were used to simulate different shocks

such as hurricanes and wildfires and changes in prices and income. These simulations are helpful

when making recommendations for the tourist industry in an attempt to increase overall demand

for tourism in Florida.

Data and Scope

Passenger data used in this study were collected by the Bureau of Transportation Statistics

and are available online. Nearly 4.3 million observations were collected from the database T-100

Domestic3 Market and T-100 International Market between January 1990 and December 2006. A

market is defined by the first departure airport on a ticket and the ultimate arrival airport. The T-

100 Domestic Market database included all passengers enplaned at a U.S. origin airport and

deplaned at the ultimate U.S. destination airport. It also provides monthly statistics on freight and


3 Data include 50 States, District of Columbia, Puerto Rico, and other U.S. territories and possessions.









mail specified by their origin airport, city and state, carrier (airline) identification number,

destination airport, and miles between airports. Since the scope of the study only covers Florida,

all observations reporting Florida as destination state were then selected.

Due to the high volume of data collected, aggregation was performed in order to make the

data more manageable. Such aggregation was made at two levels: at the origin grouping states

into U.S. regions and at the destination grouping cities into 22 CSAs, both following U.S. Census

guidelines.

The T-100 International Market database includes all passengers enplaned at an

international origin airport and deplaned at the ultimate U.S. destination airport and all

passengers enplaned at a U.S. origin airport and deplaned at the ultimate international destination

airport. It also provides monthly statistics on freight and mail specified by their origin airport,

city and country, carrier (airline) identification number, destination airport, city and country, and

miles between airports. Again, all observations reporting Florida as ultimate destination state

were selected. International passenger data underwent a level of aggregation similar to that

performed on the domestic passenger data.

Data on airline ticket prices for domestic flights used in this study were collected by the

Bureau of Transportation Statistics (BTS) and are available online. The Origin and Destination

Survey (DB 1B) Market was used to collect one way and round trip airline ticket prices from

several domestic origins traveling to Florida. According to the BTS, the DB B-Market contains

(directional) origin and destination markets, which is a 10% sample of airline tickets from

reporting carriers. It includes items such as passengers, airfares, and distances for each

directional market, as well as information about whether the market was domestic or

international. Nearly 46.8 million observations reported on a quarterly basis were retrieved from









the DB1B-Market database between Quarter I, 1993 and Quarter IV, 2006. Description,

selection, and aggregation of airline ticket prices are detailed in Chapter 3.

Additionally, the following data were collected from various sources and included in this

study: annual gross domestic product by state and per capital disposable income by state, monthly

statistics on number of wildfires including total of acres burned across Florida, hurricane data,

average temperatures and rainfall from selected cities, brand advertising expenditures from

selected private firms, generic advertising expenditures, monthly average kerosene-type jet fuel

prices, annual population estimates by state, and annual crime statistics from Florida. This study

focused on the specification of a partial adjustment model for domestic airline passenger traffic

traveling to Florida between January 1995 and December 2006 only. Passenger data include all

passengers traveling to Florida without specifying the reason of travel. These passengers could

be residents coming back to Florida, individuals on business trips, or tourists. Therefore, all

inferences are restricted to total passengers rather than tourists.









CHAPTER 2
LITERATURE REVIEW

Over the past few decades, demand for tourism has become a focal point in several

empirical studies. Many industries depend on accurate forecasts of tourism demand. Unlike

manufacturing products that can be stored, tourism products such as airline seats and hotel rooms

cannot be stockpiled (Archer 1987). Any under-used tourism product has essentially no value

since it does not have alternative uses. Airlines, tour operators, and hotel chains incorporate these

forecasts into their everyday operational decision making, risk management, and strategic

planning. As Witt, Song, and Louvieris (2003) state, failure to predict major downturns or

upswings in tourism demand could have serious financial consequences.

Forecasting Tourism Demand

Forecasting literature regarding tourism demand is to a large extent dominated by a

relatively limited number of writers: Archer, Frechtling, Smeral, and Witt (Vanhove 2005).

Moreover, international tourism demand at a national level (due to data availability) dominates

the scope of most studies and only a few have focused on domestic demand (e.g., Witt,

Newbould, and Watkins 1992). In this study, these authors compared three models (exponential

smoothing, naive I, and naive II) to generate forecasts of visitor arrivals in Las Vegas, Nevada.

Mean absolute percentage error and root mean square percentage error were used to measure

forecasting accuracy of the three models. Also, data used in the study include monthly visitor

arrivals to Las Vegas from 1973 to 1986 reported by the Marketing Bulletin and the 1990 Ten

Year Summary. Results suggest that real improvements in forecast accuracy can be achieved

using the exponential smoothing technique, given the nature of their dataset. Both the

exponential smoothing and naive II models outperformed the naive I model in terms of mean

absolute percentage error. Also, these authors conclude that domestic tourism demand tends to









be less volatile than international tourism demand, since external influences such as exchange

rate fluctuations and international political events have much less impact on domestic tourism

than international tourism. Moreover, a weak U.S. dollar and political instability in foreign

countries may increase domestic demand because consumers would rather choose to travel to

cheaper and safer domestic destinations. They agree that the methods used to forecast

international demand may not be appropriate to estimate domestic demand. Hence, further

research is needed to determine the relative accuracy of various forecasting techniques applied to

estimate domestic tourism demand.

Studies on international tourism demand are found more frequently in the literature, with

an increased interest in developing countries. As mentioned in Chapter 1, tourism has become a

vital source of income and a major contributor in the balance of payments for developing

nations. Countries in Africa, Asia, and Latin America have greatly benefited from flows of

visitors traveling to their sites. In addition, researchers have used econometric forecasting models

to gain some insights into the forces driving tourism demand in these countries and also to

estimate future flows of visitors to these destinations.

The objective of these studies has been to forecast flows of tourists in order to plan for

necessary investments in infrastructure and labor and to determine the environmental and social

impact in their communities. Empirical studies have been conducted in Macau (Song and Witt

2006), Indonesia and Malaysia (Tan, McCahon, and Miller 2002), Laos (Phakdisoth and Kim

2007), Latin America (Eugenio-Martin, Martin Morales, and Scarpa 2004), and Africa (Naude

and A. Saayman 2004). Studies have also shown special interest in analyzing air travel flows to

Latin America, Southeast Asia, and Africa since most of the visitors use this mode of transport to









travel to these destinations. For instance, air travel in Europe competes with other modes of

transportation (i.e., car and train) because distances between origins and destinations are shorter.

Interestingly, their conclusions have notably varied from those made when analyzing

tourism demand in developed countries. Several studies have shown that the primary demand

driver of tourism in developed countries is income, whereas political and social stability and

accessibility drive demand for tourism in developing countries.

Phakdisoth and Kim (2007) present a static and a dynamic econometric model that

attempts to identify key determinants of inbound tourism in Laos. Their model includes both

demand and supply factors in a single equation model. Demand drivers such as gross national

income from countries that had bilateral trade with Laos; Laos GDP, exports, and imports; cost

of living in Laos; and distance between origin country and Laos are included in the analysis. In

addition, road infrastructure at Laos, communication infrastructure at Laos, and the rule-of-law

indicator published by the World Bank, were included as supply factors in the model. The

random effects and fixed effects model and ordinary least squares (OLS) were used as static

model specifications, while the partial adjustment model was used to specify the dynamic model.

These authors conclude that supply factors (i.e., roads, communication, and stability in Laos)

were more important than originating countries' income.

In developed countries, concerns that tourism is no longer a luxury good have prompted

more empirical studies to determine the effect of this trend. Smeral and Weber (2000) suggests

that there is an expectation that tourism will gradually lose the character of a luxury good and

that at least in some segments there will be a trend towards saturation. As a result, long-term

growth in tourism expenditures will level off. Proenca and Soukiazis (2005) conducted a study

that analyzed tourism demand in Portugal. They concluded that income is the most important









determinant of demand in their static model. These authors also specify a dynamic model which

identifies accommodation capacity as the most important supply determinant in tourism demand

in Portugal. Similarly to the previous study conducted in Laos, this paper specifies a random

effects and fixed effects model for the static model and a partial adjustment model for the

dynamic model. This study is also limited to the country level and includes four countries

identified as Portugal's largest tourism suppliers. Results suggest that some kind of inertia or

rigidity exists in tourism inflows. It is unclear though whether this is related to the maturation of

the markets. Nevertheless, the authors conclude that Portugal must develop new policies to

reduce the dependence from the United Kingdom, Spain, Germany, and France and explore

alternative markets.

Scope of Recent Tourism Demand Studies

Several studies with a domestic scope explore other issues related to air transportation but

that could affect tourism demand. For example, some studies focused on determining the impact

of airline's market power on optimal airport capacity (Zhang and Zhang 2006), analyzing

productivity and efficiency of hub and spoke models (Brueckner 2004), and estimating price,

income, and distance elasticities of air transportation demand (Bhadra and Kee 2008). These

studies, although not related directly to tourism demand, provide valuable insight since any

policies assessed to the transportation industry may have an effect on the tourism industry.

Recently, other concerns such as environmental issues (e.g., noise and air pollution), health

scares (e.g., influenza, SARS) and security incidents (e.g., 9-11 terrorist attacks) have been

subject to discussion and analysis related to air transportation and tourism. Lu and Morrell

(2001) analyzed the implications of an environment charge mechanism assessed to airlines,

specifically its impact on airline costs, airfares, and passenger demand. Since externalities such

as aircraft noise and engine emissions generate profound impacts on human beings and on the









environment, environmental charges are one of the economic instruments used to control them.

Government agencies have assessed these fees to commercial flights and proceeds are used to

improve the environment at airports and surrounding communities. The study concluded that

there is a need for an assessment of environmental charge mechanisms to encourage sustainable

development. Brons et al. (2002) state, however, that if airlines can charge all extra costs to

passengers without decreasing demand, the environment policy has no other effect than

increasing authorities' revenues and decreasing consumers' surplus.

Grais, Ellis, and Glass (2003) conducted a simulation analysis in which they explored

potential ramifications of the influenza pandemic strain of 1968-1969 as if it had resurfaced in

Hong Kong in 2000. They suggested that since air travel had increased since the late 1960s, the

disease could be spread faster (111 days earlier than it had been forecasted in 1968) and to more

people (176% greater than the 1958 pandemic).

Moreover, travelers sacrifice not only cash costs but also the opportunity of using the time

in an alternative activity (i.e., opportunity cost of travel). Studies by Blunk, Clark, and

McGibany (2006) and Ito and Lee (2005) have evaluated the economic impact of the 9-11

terrorist attacks on New York City, Washington D.C., and Pennsylvania. Both studies agreed that

the stricter security requirements implemented as a result of the terrorist attack have increased

the opportunity cost of travel.

Blunk, Clark, and McGibany (2006) used a modal choice model applied to communal

behavior to estimate air travel demand in terms of revenue passenger miles. They tested the

hypothesis that anything that permanently raises the opportunity cost of travel, holding benefits

constant, should reduce travel volume. Their study concluded that the 9-11 terrorist attacks









increased the opportunity cost of travel and that the detrimental impacts of the attacks were not

temporary.

In their study, Ito and Lee (2005) attempted to separate the effect of the 9-11 terrorist

attacks into temporary and ongoing components. Results suggested that negative impacts on both

quantity and price indicate that the 9-11 terrorist attacks resulted in a negative demand shift,

rather than a supply contraction. Such negative demand shift was estimated to be at 30% in its

transitory stage in addition to a 7.4% ongoing negative demand shift. They also agreed that such

a catastrophic event could require a long recovery process. Finally, they state that technological

innovations in security screening might eliminate some of the waiting time in airports, reducing

the hassle factor and making air travel more convenient, and hence, increase demand for air

travel.

Hodges and Mulkey (2001) also assessed the potential impact of the 9-11 terrorist attacks

on Florida's economy. They constructed a model using IMPLAN economic impact modeling

system to reflect linkages between industries, employees, institutions, and consumers. The

objective was to simulate both direct and indirect impacts triggered by the 9-11 terrorist attacks.

Results suggested that if there is a 10% decrease in overall tourism expenditures, the economic

impact in Florida would increase to USD 11.7 billion, including indirect effects (USD 1.3

billion) and induced effects (USD 3.54 billion). These authors concluded that the total economic

impacts on tourism are substantially greater than the direct value of visitor spending, due to the

multiplier effect of new money introduced to the state's economy. In addition, they argued that a

modest change in visitor spending can dramatically affect state government fiscal balances,

which is largely funded by sales tax revenues.









Econometric Techniques Applied to Tourism Demand Analysis

Several techniques have been used to model and forecast tourism demand. Econometric

literature presents techniques that range from pure qualitative studies such as surveys to rigorous

quantitative analyses (e.g., multiple regression models). Qualitative studies include the use of

surveys, Delphi models and, to a lesser extent, judgment-aided models. Within the quantitative

approach, studies can be divided into trip generation models, gravity models, time series

methods, and regression analysis. The following discussion will focus on quantitative techniques

used to estimate tourism demand.

Trip generation models are the stepping stone of a four-step transportation forecasting

process, followed by trip distribution, mode choice, and route assignment. Trip generation

models have been used to estimate the number of trips that each area will generate and attract.

There are two kinds of trip generation models: production models and attraction models. Trip

production models estimate the number of home-based trips to and from areas where trip makers

reside. Trip attraction models estimate the number of home-based trips to and from each area at

the non-home end of the trip. Trip production models are commonly used to estimate truck and

taxi trips, whereas trip attraction models are used to estimate tourism demand.

Gravity models are based on Newton's Law of Gravity with the idea that as the size of one

or both areas increases, movement between them will also increase. These models have been

used to predict the degree of interaction between two places. These studies attempt to determine

at which point consumers will prefer to travel to one area instead of the other. Explanatory

variables most frequently used include distance, time, and expenses. Applications of gravity

models have extended to international trade where it is used to estimate bilateral trade flows

between two areas. The model is based on the economic size of each area (e.g., gross domestic

product) and the distance between them. Since gravity models rely on distance as its primary









explanatory variable, they have been subject to criticism for their lack of theoretical foundation.

Sheldon and Var (1985) argue that gravity models can only be used to forecast number of

tourists, but not expenditures, occupancy rates, and other important variables. They also add that

origin zones are difficult to identify and that measures of travel time and cost are neither constant

nor accurate, since they can vary by means of travel or season.

Bhadra and Kee (2008) specify a gravity model to estimate airline travel demand in which

they express the explanatory variables in their log-linear form. Such variables include average

one way airfare between origin and destination, personal income and population at origin and

destination, and distance between origin and destination. Demand is expressed in number of

passengers between 192 metro areas in the United States. Metro areas are grouped into four

categories based on the level of traffic: super-thin, thin, thick, and super-thick markets.

Multicollinearity was mentioned as a potential problem because large populations tend to be

associated with higher levels of economic activities. A formal test was conducted and standard

errors were small and stable across sub-samples, but condition indices appear to be larger than

usually suggested. Results suggest that passenger flows between origin and destination markets

exhibit growth in recent years. In addition, demand is elastic with respect to airfares in thick

markets, supporting the hypothesis that there is high competition in these markets. Passenger

flows in semi-thick and super-thin markets were found to be distance-inelastic.

Several studies on international tourism demand focused only on forecasting tourism

demand using time series models where the sole objective is to predict future flows of tourists to

a particular country. Time series models, or non-causal models, assume that any variable can be

forecasted without taking into account any factor that influenced the level of the variable.

Exponential smoothing; sine-wave regression; and the Box-Jenkins family of time series models:









auto-regressive (AR), auto-regressive moving average (ARMA), auto-regressive integrated

moving average (ARIMA), and seasonal auto-regressive integrated moving average (SARIMA)

are some models specified to forecast tourism demand. Some of these models have superior

predicting ability (e.g., Box-Jenkins method) and are now widely used to forecast tourism

demand. In fact, the Box-Jenkins approach is widely applied in the airline industry and has been

labeled as the "airline model".

Time series models have been widely used to predict tourism demand because they tend to

perform better than regression models. However, a few empirical studies have concluded the

opposite. Several time series models and adaptations can be found in the literature. Most of the

focus has been to compare several time series models to determine which model generates the

lowest error magnitudes. Common performance measures used in these studies include mean

average percentage error (MAPE), root mean square error (MRSE), and mean absolute error

(MAE).

Oh and Morzuch (2005) present a comparison of eight time series models. Flight data from

monthly observations from July 1977 to July 1990 on travelers' arrivals at Singapore were used

in the study. Their study splits the data into two groups: a within-sample period from July 1977

to December 1988 and an out-of-sample period from January 1989 to July 1990. Each model was

used to generate data for a 3-month, 15-month and a 19-month horizon. Dickey-Fuller tests were

performed to test for stationarity. These authors then chose six performance criteria: one to

determine whether the estimate is unbiased and the remaining five, MAPE, MRSE, MAE,

Akaike's information criterion (AIC), Schwartz's Bayesian criterion (SBC), to quantify loss due

to forecast error. Results suggest that in the three-month horizon the ARIMA (3,1,0)

specification ranked first on four of the five performance statistics when applied to the post-









sample data, but its accuracy decreases as the forecast horizon increases. Also, within-sample

may not show any structural changes during the post-sample period and hence provide poor

forecasts. This study concluded that ARIMA models provide a reliable and consistent forecasting

performance across different time horizons, with the caveat that such models shed little light on

issues related to policy.

As mentioned earlier, not all empirical studies have agreed that time series models perform

better than econometric models. Fritz, Brandon, and Xander (1984) presented an econometric

model aimed to forecast air arrivals into Florida from domestic points of departure. Quarterly

data were collected from the State of Florida reported from 1960 to the fourth quarter of 1980.

The model includes variables describing economic characteristics such as disposable personal

income and the composite of leading and coincident indicators published by the Bureau of

Economic Analysis. In addition, these authors explain that all variables were lagged three

quarters behind to provide more timely forecasts. They used the Box-Jenkins method as a pure

time series model, an econometric model, and a combination of both. Results suggested that

econometric forecasts were considerably more accurate than the Box-Jenkins but that there is an

improvement in accuracy available with the combination of forecasts. Also, forecasting errors

increased as the time horizon increased. The study concluded that forecasting accuracy provides

important benefits to state and local government since their revenues derive from tourist

expenditures.

Determinants of Tourism Demand

Understanding causal relationships between variables proved to be vital in the design of

policy and planning strategies. Since pure time series models lack explanatory power, they are

used exclusively to forecast demand with no regard into determine which forces drive such

demand for tourism. Regression analysis, on the other hand, tries to fill the gap left by time series









models. Regression analysis (i.e., causal methods, econometric models) attempts to explain the

behavior of the dependent variable due to some explanatory variables, as well as to forecast

future levels of demand.

Under regression analysis, the literature offers several methods used to model tourism

demand. Single linear regression methods use one explanatory variable only and are the simplest

version found in the literature. Multiple regression methods which include more than one

explanatory variable in a single equation have also been used to model tourism demand. Most

frequently used explanatory variables include income, commercial ties between countries, price,

price substitutes, transportation cost, distance, travel time, exchange rate, promotions efforts,

population growth, supply capacities, business cycles, trend factors, qualitative factors, dummy

variables for special events, natural disasters, war, terrorism, and lagged dependent variables

(Vanhove 2005).

Air travel demand shares some of these determinants with tourism demand. Nielsen (2001)

classifies the determinants of air travel growth in two categories: drivers and impeders. He

identifies increased personal incomes with reduced real airfares as the major drivers of air travel

demand. Building up the aviation socio-technological system, technological innovation,

globalization, competition, incentives to the industry, economic growth policies, population

growth, and advertising are also mentioned as drivers of air travel. Impeders of air travel growth

include financial and time constraints, airport congestion, environmental policies, and alternative

lifestyles. He concluded that most air travel relates to globalization, changing geography, and

population dynamics.

Air travel has benefited from increased commercial relations between countries. Yet

environmental pressures are arising as countries increase their trade of products, services, and










human and financial capital and pose an important threat to air travel growth. Figure 2-1 presents

the physical, social, and political determinants of air travel growth as defined by Nielsen (2001).


Market Forces
Liberalization of markets
Competition between airlines
Competition between travel I
agencies
Marketing strategies

Aircraft Technology v
- Longer range
- Improved capacity
- Increased speed
- Reduced operation costs



Infrastructure
- Enlarged airport capacity
- Reduced rail capacity


Geography changing
- Population growth
- Migration
- Internationalization of family
structures J


Globalization
Globalization of market
Globalization of
companies
Globalization of political
system
More international
relations


Impeders -


Free time availability
- Work structures
- Holiday structures
- Rich and aging population
- People taking a year off


Economic Factors
- Reduced airfares
- Increased income
- Economic growth


Political Factors
S- Maintain growth
Maintain employment in
aviation socio-technical
system
Competition among
nations
Subsidies



Psychological Factors
-Individual needs, wants,
and desires





Social Factors
- Social norms and values
- Air travel as status maker
- Trends: further away,
deeper into the forest,
higher up in the mountains
- Experience other cultures
- Traveling cultures
- Escape from the "cage of
routine"


Figure 2-1. Determinants of air travel growth. Source: Nielsen 2001.

Log-linear specifications have been the primary functional form used in most analyses. It

has been shown that the log-linear specification tends to model the demand data better and it also

conveniently provides demand elasticities since the estimated parameters represent the

elasticities of the variables. These demand elasticities are then used to formulate policies and

examine how consumers respond to changes in demand variables (Uysal and El Roubi 1999).









Elasticities provide valuable information by quantifying how much the demand for a product or

service changes with a change in its determinants or drivers. For example, elasticities can show

how much a change in air ticket prices affects demand for air travel to a specific area.

Some authors have challenged the forecasting accuracy of econometric models and have

compared them to time series models. Most of them agreed that time series methods are more

accurate that econometric models. Nonetheless, authors have also stated that econometric models

play an important role in policy implementation and strategic planning. There is a need to

understand the determinants of demand and to assess the impact of any changes in these

determinants on tourism demand. It is also true, however, that there is a clear tradeoff between

forecasting accuracy and identification of causal relationships. The use of one model over the

other will depend on whether the interest is pure forecasts or identification of the determinants of

demand.

Martin and Witt (1989) evaluated seven quantitative forecasting methods, six time series

methods and one econometric forecasting method. This analysis used annual tourism flows data

at a country-to-country level (e.g., United Kingdom to Austria). They concluded that, although

its accuracy was poor compared to time series models, an econometric forecasting system does

allow exploration of the consequences of alternative future policies on tourism demand.

Econometric forecasting models, unlike time series models, provide helpful insight due to their

ability to identify and quantify the determining forces driving demand. Such insight represents a

great value at several time horizons. Witt and Witt (1995) argue that short-term forecasts are

required for scheduling and staffing needs, medium-term forecasts for planning tour operator

brochures, and long-term forecasts for investment in aircraft and infrastructure.









Witt, Song, and Louvieris (2003) reviewed past demand forecasting studies and compared

the performance of six econometric models and two univariate time series models on

international tourism demand in Denmark. These authors evaluated the forecasting performance

of four econometrics and two time series models, their ability to forecast direction of change, and

the unbiasedness of their estimates. Econometric models used in this study include a static

model; two error correction models: Wickens and Breusch procedure (WB) and Johansen

maximum likelihood method (JM); a reduced autoregressive distributed lag model (ADLM); a

time-varying parameter model (TVP); and an unrestricted vector autoregressive model (VAR).

The two univariate time series models were an ARIMA model based on the Box-Jenkins

procedure and a simple naive or no change model.

Results suggested that for a one-year horizon, the TVP and the reduced ADLM generated

the most accurate, unbiased forecasts when accuracy was defined in terms of error magnitude.

When forecasting unbiasedness is taken into account in conjunction with directional change

error, the static and TVP models outperformed the other four models in forecasting one-year-

ahead. Moreover, econometric models outperformed the naive model in terms of generating

accurate forecasts of directional change, which suggested the importance of taking structural

instability into consideration when generating short-term forecasts. The authors conclude that,

even though, econometric models require larger data, considerable user understanding, and more

expertise than univariate time series models, tourism researchers should give serious

consideration to using them to generate forecasts of international tourism demand.

Static versus Dynamic Regression Models

Other authors have combined both time series and econometric modeling into one general

pooled model. The study by Young (1972) compared three models: a time series model with a

partial adjustment specification, a cross-section model, and a combination of both. Young's third









model incorporates both temporal and contemporaneous interdependences in order to establish

relationships between macro time series parameters and micro cross-section parameters.

Independent variables included in all three models were permanent income, ticket price, and

journey time by the airplane. Results suggested that better results can be obtained by pooling

time series and cross-section data. In addition, Young was able to report short run and long run

elasticities of price, time, and income and concluded that short run elasticities were not very

elastic while long run elasticities were highly elastic. These conclusions are validated by

economic theory that shows that consumers and firms are able to adjust in the long run.

Under regression analysis, the literature also presents static and dynamic models that

attempt to model tourism demand. A particular challenge arises when using a static model, also

known as long-term model, to estimate tourism demand. Consumers tend to adjust better to price

and security signals in the long run than to sudden changes in costs or unforeseen security

incidents in the short run. This is particularly true in the tourism industry since consumers

typically book their flight and hotel reservations in advance limiting their ability to adapt

instantaneously. Therefore, dynamic models have become popular and are considered to estimate

demand more accurately because these models account for rigidities that could occur as

consumers adjust to changes in their demand drivers, especially in the short run. Differences

between static and dynamic models are fully explained in Chapter 4.

Forecast of tourism demand to Portugal using dynamic modeling was the main focus on an

empirical study by De Mello and Fortuna (2005). These authors argued that habit persistence,

adjustment costs, or imperfect information prevent consumers from automatically fully adjusting

every period. Therefore, an explicit dynamic structure is required to explain demand behavior

and to account for the short run adjustment process. They developed several dynamic stochastic









specifications including a partial adjustment model and a dynamic almost ideal demand system

(DAIDS) model. The study concluded that the partial adjustment model was consistent with the

postulates of economic theory and provided robust and empirically plausible estimates.

Challenges in Tourism Demand Analysis

Other studies have focused on the difficulties of collecting relevant data that could be used

to assess the real economic contribution of tourism-related activities to a particular area.

Problems arise because the national accounts classification system does not identify tourism as a

separate component of the GDP. The tourism satellite account (TSA) program aims to fill that

gap and facilitate the assessment of tourism's overall economic impact. According to the Office

of Travel and Tourism Industries web site, these data do not exist from any federal government

agency or private sector source. A prototype account is being established by the Bureau of

Economic Analysis in conjunction with the Tourism Industries office which will incorporate as

much data as is available to the "core" account consisting of the traditional industries for tourism.

Jones, Munday, and Roberts (2003) explore methodological difficulties in the construction

of a TSA at a regional level and the implications of deriving an effective tourism policy. They

mention that data from tourism surveys are seldom detailed enough to account for incidental

purchases made by tourists. Moreover, the oversimplification of the labor account presents a

problem in the construction of the TSA. Use of labor is highly uneven across space and time, and

even within a region. Therefore, these accounts do not illustrate the fact that the labor market in

the tourism industry is highly seasonal. They agree that the disaggregation of existing input-

output data to account for technical coefficients, local sourcing patterns, and labor use is needed.

They also realize, however, that this process may be cost-prohibited and that there is no real

evidence that the construction of the TSA is worth the effort.









Despite the overlapping worlds of tourism and transport, there has only been limited

progress evidenced in the literature (Lumsdon and Page 2004). Many of the shortcomings in

research have stemmed from the fact that transport and tourism data are inconsistent. Key

elements such as tourist or passenger, trip or tour are not clearly defined. Currently, efforts are

being made to standardize survey designs, procedures, and sampling. Also, a common challenge

faced by most studies has been to determine which dependent variable to use in order to

accurately estimate tourism demand. For example, there is no apparent consensus among

researchers on which dependent variable is the most suitable to measure tourism demand.

Although total tourist arrivals for specific regions or countries have been widely used, real

expenditures on tourism goods and services have also served as a measure of tourism demand.

The complexity of the industry also hampers the ability to disaggregate data to estimate pure

effects on demand.

In addition, more challenges arise because crises and disasters, which affect tourism

greatly when they occur, are impossible to forecast. Researchers face constraints in terms of

accurate and relevant data that are difficult to collect for some variables (Sheldon and Var 1985).

However, they have used several techniques and measures to improve the estimation of tourism

demand with the information available.

Multicollinearity has been mentioned as a potential problem in both time series and cross-

section estimates. For example, gravity models have encountered high correlation between

airfares and distance. This problem intensifies in time series models given that price and income

tend to be strongly associated with a time trend. Also, within the econometrics models,

multicollinearity is often present among explanatory variables, especially in income and cost of

travel variables when data are cross-sectional. Another limitation found in almost all models is









the forecast horizon. These models may only be applicable to short-term analysis, since estimates

change with time. Forecasting models tend to perform better on short-term horizons, while

forecasting error increases as the time horizon increases.

Lagged dependent variables create a problem in dynamic models. Studies conducted by

Phakdisoth and Kim (2007) and Proenca and Soukiazis (2005) encountered statistical problems

since there was a correlation between the dependent variable and the error term arising from the

lagged dependent variable used as an explanatory variable in the partial adjustment model.

Estimates are biased and inconsistent under ordinary least squares estimation (OLS) or general

least squares estimation (GLS), especially in smaller samples. Instrumental variables (IV)

estimation techniques were used to solve such problem. The idea behind IV is to identify

instruments that are highly correlated with the endogenous regressor but uncorrelated with the

error term.

Researchers have also dealt with the phenomenon of seasonality. Seasonality in tourism

demand has been well documented in the literature and has been regarded as a problem.

Seasonality can be identified as fluctuations of demand at tourist destinations. It affects tourism

suppliers, tourism employees, residents, and tourists alike. In the off-season, defined as low flow

of tourists to a particular area, suppliers struggle to fill capacity (e.g., hotel rooms, aircraft seats,

attraction tickets), employees have no access to sustainable long-term employment, tourists

encounter limited services (although can benefit due to discounts during this season), and

residents who own businesses in the area experience a sales decrease. Lundtorp (2001) discusses

four main reasons for the existence of seasonality: 1) weather (variation in temperature, rainfall,

snowfall, and daylight); 2) institutional patterns (school holidays, industrial holidays, and









calendar holidays); 3) destination characteristics (winter versus summer resorts); and 4)

marketing efforts (special events, conferences).

Accurate identification of seasonality and its drivers reduce uncertainty and therefore, risks

in tourism operations. It also could lead to a successful marketing strategy that will aim to

increase flow of tourists during the off-season. It helps the operations of tourism firms to plan

accordingly. For example, it is not uncommon to find some hotels at summer holiday

destinations (main season) closed during winter (off-season).

Kulendran and Wong (2005) present a time series model aimed to measure seasonal

variations in holiday, visit to friends and relatives (VFR), business, and total visits. Conventional

unit root tests were conducted to determine the nature of the seasonality and order of integration.

Two competing models were considered: an ARIMA model with first differences (ARIMA1)

and an ARIMA model with first and fourth differences (ARIMA14). These two models were

fitted to the inbound tourism quarterly data series of eight European countries and the United

States between the period of 1978 and 1998. Results suggested that seasonal variation varies

according to the type of visit. Holiday tourism was found to have the highest R2, which implies

that it has strong seasonal variation, followed by VFR and business visits. In addition, the

forecasting accuracy of the two models was measured using the MAPE. These authors concluded

that the forecasting performance of ARIMA1 and ARIMA14 depends on the nature of the

seasonal variation in tourist arrivals time series and that correct identification of seasonal

variation can improve the forecasting performance of the ARIMA model. The ARIMA1 provides

better forecasts for time series with less seasonal variation and ARIMA14 provides better

forecasts for time series with strong seasonal variation.









Recent Developments in Tourism Demand Analysis

Recent developments in econometric modeling and forecasting show that new and more

advanced models are being used in the tourism demand literature. Some of these models include

time varying parameter (TVP) estimation, almost ideal demand system (AIDS) models, the error

correction model (ECM), vector autoregressive (VAR) approach, and time series augmented

with explanatory variables such as structural time series models (STSM).

Li and Song (2006) describe some of these models. The first approach is related to the

TVP estimation approach. The TVP single-equation models have gained popularity due to their

flexibility. Elasticities derived from log-linear regressions are constant through time, a condition

quite restrictive and that often leads to failure of the dynamic analysis of tourism demand (Li,

Song, and Witt 2005). Parameters in the TVP single-equation model vary over time and allow

the model to identify any short run changes in consumer preferences. The second approach found

in the literature is related to the AIDS models with some refined specifications such as static

linear AIDS (LAIDS) and dynamic LAIDS. Dynamic LAIDS models are somewhat similar to

the partial adjustment model in that they account for consumer behavior in the short run.

Overall, AIDS models allow testing for other demand theories such as symmetry and

adding-up hypotheses. It also has a flexible functional form and does not impose a priori

restrictions on elasticities, satisfies the axioms of consumers without invoking to Engel curves,

and largely avoids the need for non-linear estimation. The authors suggest that, even though the

Rotterdam and translog models hold some of these features, only the AIDS model contains them

simultaneously and hence, it is more suitable for tourism demand analysis. The third approach

relates to a combination of the flexibility of TVP models and the multiple equation specification

from the LAIDS models. The authors also indicate that these models have only been applied to

annual data and have not modeled seasonality. They conclude that future research should focus









on seasonal patterns of tourism demand and that such seasonal components need to be

incorporated to these models. Moreover, Li, Song and Witt (2005) stated that investigation of the

forecasting performance of advanced econometric models in dealing with seasonality in tourism

demand and comparison of the abilities of alternative models to forecast tourism demand

changes are areas of interest and directions for further research.

Chapter Summary

This chapter presented a summary of the literature related to tourism demand analysis.

Most of tourism demand analyses have focused on forecasting rather than identifying the

determinants of demand. Such studies have concluded that time series analysis provides more

accurate forecasts of tourism demand. Nevertheless, researchers agree that econometric models

are vital for policy and strategic planning and risk management. However, recent approaches aim

to combine time series with causal models to improve accuracy of the estimates.

The complex interaction between the tourism and transportation industries has hindered the

ability to accurately estimate tourism demand. Data availability and the lack of a standardized

system of accounting between both industries have also been mentioned as a limitation. Most of

the studies use annual data and do not model seasonality, an intrinsic characteristic of tourism

demand.

This study develops a model that attempts to fill the gap by addressing those issues and

hence, contribute to the tourism research literature. A partial adjustment model is developed in

order to identify the demand drivers of air travel to Florida and its relationship to tourism. The

model incorporates seasonal components as well as economic, social, and weather factors related

to air travel demand. Chapter 4 presents a detailed derivation of the partial adjustment model.









CHAPTER 3
DESCRIPTIVE STATISTICS ON THE AIRLINE PASSENGER TRAFFIC TRAVELING TO
FLORIDA AND OTHER INDICATORS

This chapter consists of descriptive statistics for the passenger, freight, and mail traffic

coming to Florida by means of airline transportation, as well as economic and social indicators

related to four U.S. origins and Florida. It also includes weather statistics from Florida, such as

hurricanes, wildfires, and average temperatures and rainfall. Chapter 3 is organized into four

major sections, with a number of subsections in each section. The four major sections are defined

as a) airline passenger traffic, b) airline ticket prices, c) freight and mail transported by

commercial passenger airlines, and d) economic, social, and weather indicators.

The first section, airline passenger traffic, contains five subsections: a) description of the

collection and selection of airline passenger traffic data; b) total domestic airline traffic in terms

of number of passengers coming to Florida by U.S. region from 1990 to 2006; c) total domestic

airline traffic in terms of number of passengers coming to five Combined Statistical Areas

(CSAs) in Florida from 1990 to 2006; d) total international airline passenger traffic traveling to

Florida by world region from 1990 to 2006; and e) total international airline passenger traffic

traveling to three destination CSAs in Florida from 1990 to 2006.

The second section, airline ticket prices, contains six subsections: a) description of the

collection and selection of domestic airline ticket price data; b) one way and round trip airline

prices of domestic flights coming to Florida by U.S. region, from 1993 to 2006; c) one way and

round trip airline prices of domestic flights traveling to five destination CSAs in Florida by U.S.

region, from 1993 to 2006; d) description of the collection and selection of the international

airline ticket price data; e) round trip airline prices of international flights coming to Florida, by

world region from 1995 to 2006; and f) round trip airline prices of international flights coming to

three destination CSAs in Florida, by world region from 1995 to 2006.









The third section includes freight and mail transported by commercial passenger airlines.

An analysis by U.S. region and by destination CSA is performed to domestic freight and mail

traffic transported to Florida. An analysis by world region and by destination CSA is performed

to international freight and mail traffic transported to Florida. The fourth section, economic,

weather, and social indicators, contains five subsections: a) GDP by state; b) brand and generic

advertising expenditures; c) foreign exchange rate: Euro to U.S. dollar; d) historic oil prices; e)

hurricanes and wildfires affecting Florida; f) average temperatures in origin U.S. regions and

destination CSAs including rainfall in each of the five destination CSAs; and g) crime rates in

Florida.

Airline Passenger Traffic

Description, Selection, and Aggregation of Domestic Airline Passenger Traffic Data

Passenger data used in this study were collected by the Bureau of Transportation Statistics

and are available online. Nearly 4.3 million observations were retrieved from the database T-100

Domestic Market between January 1990 and May 2007 and T-100 International Market between

January 1990 and February 2007. A market is defined by the first departure airport on a ticket

and the ultimate arrival airport. The T-100 Domestic Market database includes all passengers

enplaned at a U.S. origin airport and deplaned at the ultimate U.S. destination airport. For

example, a flight from Rochester, New York to Miami, Florida with a stop in Atlanta, Georgia

identifies the origin state as New York and the ultimate destination state as Florida. Also, the

data set provides monthly statistics on freight and mail specified by their origin airport, city and

state, carrier (airline) identification number, destination airport, city and state, and miles between

airports. Since the scope of the study only covers Florida, all observations reporting Florida as

destination state were selected.









Due to the high volume of data collected, aggregation was performed in order to make the

data more manageable. Such aggregation was made at two levels: at the origin and destination.

At the origin level, origin states were aggregated according to the U.S. Census geographic

regions scheme as shown in Table 3-1. All domestic statistics from the 50 states and D.C. were

aggregated into four different geographical originating U.S. regions: Northeast, Midwest, South,

and West. Domestic statistics for Puerto Rico, U.S. Virgin Islands, and U.S. Pacific Trust

Territories and Possessions were aggregated into a category named "Other".

Table 3-1. Geographic region scheme as defined by the United States Census.
Northeast Midwest South West
1. Connecticut 1. Illinois 1. Alabama 1. Alaska
2. Maine 2. Indiana 2. Arkansas 2. Arizona
3. Massachusetts 3. Iowa 3. Delaware 3. California
4. New Hampshire 4. Kansas 4. D.C. 4. Colorado
5. New Jersey 5. Michigan 5. Georgia 5. Hawaii
6. New York 6. Minnesota 6. Kentucky 6. Idaho
7. Pennsylvania 7. Missouri 7. Louisiana 7. Montana
8. Rhode Island 8. Nebraska 8. Maryland 8. Nevada
9. Vermont 9. North Dakota 9. Mississippi 9. New Mexico
10. Ohio 10. North Carolina 10. Oregon
11. South Dakota 11. Oklahoma 11. Utah
12. Wisconsin 12. South Carolina 12. Washington
13. Tennessee 13. Wyoming
14. Texas
15. Virginia
16. West Virginia

At the destination level, all Florida airports within the same destination city were grouped.

For example, all passengers, freight, and mail destined to Orlando International Airport,

Kissimmee Gateway, Orlando Sanford International Airport, Daytona International Airport, New

Smyrna Beach Airport, and Bunnell Airport were aggregated to the Orlando CSA. Then,

destination cities were aggregated according to U.S. Census combined statistical areas (CSA)

assigned to Florida. For example, the South Florida CSA includes the cities of Miami, Fort

Lauderdale, and West Palm Beach. All Florida cities listed on the data were assigned to their









respective U.S. Census CSA. There were 22 CSAs in Florida. The top five destination CSAs in

terms of number of passengers received were South Florida, Orlando, Tampa-St. Petersburg,

Jacksonville, and Fort Myers. Destination CSA airports aggregated to each of the top five

destination CSAs are described in Appendix A.

Description, Selection, and Aggregation of International Airline Passenger Traffic Data

The T-100 International Market database includes all passengers enplaned at an

international origin airport and deplaned at the ultimate U.S. destination airport and all

passengers enplaned at a U.S. origin airport and deplaned at the ultimate international destination

airport. It also provides monthly statistics on freight and mail specified by their origin airport,

city and country, carrier (airline) identification number, destination airport, city and country and

miles between airports. Similar to the aggregation of the domestic data, only observations

reporting Florida as ultimate destination state were selected.

International passenger data underwent a level of aggregation similar to that performed on

the domestic passenger data. Aggregation of the international passenger data was made at two

levels: at the origin and at the destination. At the origin level, origin countries were aggregated

according to their respective origin geographical region and it was performed as follows. All

international statistics were aggregated into four world origin regions: Canada, Latin America,

Europe, and Other. Latin America region includes all countries in Central America, the

Caribbean, and South America. The Other region includes all countries in Asia, Africa, Oceania,

and the Middle East.

At the destination level, all Florida airports within the same destination city were

aggregated. For example, all international passengers, freight and mail destined to Orlando

International Airport, Kissimmee Gateway, Orlando Sanford International Airport, Daytona

International Airport, New Smyrna Beach Airport, and Bunnell Airport were aggregated to the









Orlando CSA. Then, destination cities were aggregated according to U.S. Census combined

statistical areas (CSA) assigned to Florida. For example, the South Florida CSA includes the

cities of Miami, Fort Lauderdale, and West Palm Beach. All Florida cities listed on the data were

assigned to their respective U.S. Census CSA. There were 22 CSAs in Florida. The top three

destination CSAs in terms of number of passengers received were South Florida, Orlando, and

Tampa-St. Petersburg.

Domestic and International Airline Passenger Traffic Traveling to Florida

Domestic airline passenger traffic dominates the commercial airline market traveling to

Florida. Approximately 82% of the total airline passenger traffic traveling to Florida has a

domestic origin, while 18% originates abroad. More than 701.6 million domestic airline

passengers, including intrastate traffic, have traveled to Florida from 1990 to 2006, with an

average of approximately 41.27 million per year over the 17-year period. In 2006, domestic

airline passenger traffic to Florida totaled 56,059,319 people which accounted for 7.61% of total

domestic airline passenger traffic and made Florida the third largest domestic market behind

California and Texas.

Intrastate traffic defined as flights originated at a Florida airport destined to another Florida

airport accounted for 8% of total domestic airline passenger traffic traveling to Florida during the

same time period. Intrastate traffic has increased 40% from 3.11 million passengers in 1990 to

4.35 million passengers in 2006. For the purpose of this study, intrastate traffic was not included

in the analysis. Therefore, hereafter, domestic airline passenger traffic traveling to Florida will

refer only to domestic flights originated out of the state of Florida destined to Florida.

Approximately 155.4 million international airline passengers traveled to Florida from 1990

to 2006, with an annual average of approximately 9.1 million during the 17-year period.

International airline passenger traffic coming to Florida totaled 10,499,493 passengers in 2006










which represents a 61% increase from 1990. More than 10.8 million international airline

passengers traveled to Florida in 2000, accounting for the highest level during the 17-year

period. Florida's share of international passenger traffic accounted for 13.74% of total

international airline passenger traffic arriving to the United States in 2006. Domestic airline

passengers traveled to Florida more frequently during March than any other month, while

international airline passengers did so in July. September registered the lowest level of airline

passenger traffic for both domestic and international flights. Figure 3-1 shows the seasonal

pattern of domestic and international airline passenger traffic traveling to Florida.


-4-Domestic --International
12.0%

1 I I I I I I I I I I I




1 .0% .....
S 11.0--------------------------I--------J-----------------------------I----------------


1 I I I I I I I I I
I I I I I I I I I I I
6.0% ----- I----- -- ---------------- -- -- -4-----------4------- -----


I I I I I I I I I I I
I I I I I I I I I I I










--International 8.9% 7.8% 8.8% 8.2% 7.5% 7.9% 10.1% 9.7% 7.1% 7.5% 7.7% 8.8%


Figure 3-1. Monthly seasonal pattern of domestic and international airline passenger traffic
traveling to Florida between 1990 and 2006. Source: Bureau of Transportation
Statistics.

Domestic airline passenger traffic by U.S. region

Domestic airline passenger traffic from the South region increased 71% from 12.48 million

in 1990 to 21.4 million in 2006. The South region also registers the largest share with 43% of the
5 .0 % ....









domestic airline passenger traffic traveling to Florida. The Northeast region (29%) ranks second

followed by the Midwest region (19%). The West region accounts for 6% of total domestic

airline passenger traffic and the "Other" region accounted for remaining 3%. Despite having the

smallest share of the four major U.S. regions, the West region experienced the greatest relative

growth (211%) in the 17-year period from 1.23 million airline passengers in 1990 to

approximately 3.82 million airline passengers in 2006. Figure 3-2 illustrates total airline

passenger traffic from five U.S. regions traveling to Florida from 1990 to 2006.

There have not been substantial changes in the shares between U.S. regions from 1990 to

2006. The South region has kept its dominance throughout the 17-year period but its share has

experienced a decrease from 44% in 1990 to 41% in 2006. Meanwhile the West region has

increased its share from 4% in 1990 to 7% in 2006. Figure 3-3 illustrates the share of domestic

airline passenger traffic from five U.S. regions traveling to Florida in 1990, 1998, and 2006.

Domestic airline passengers from every U.S. region traveled to Florida more frequently

during March than any other month. September registered the lowest level of airline passenger

traffic from all U.S. regions. High airline traffic levels in March could be attributed to

unfavorable weather in the other U.S. regions enticing passengers to travel to Florida. Likewise,

low airline traffic levels in September could be attributed to favorable weather in the other U.S.

regions dissuading passengers from traveling to Florida. Figure 3-4 shows the seasonal pattern of

domestic airline passenger traffic traveling to Florida.










1 Other West Midwest U Northeast U South


4LKIb
CC Om Cl Cfl 3t Lfl \
o~~ a a a a a a a
~~o~ a a a a a a a
Cl Cl Cl Cl Cl Cl C


Year
Figure 3-2. Florida CSA: total domestic airline passenger traffic from five U.S. regions between
1990 and 2006. Source: Bureau of Transportation Statistics.


U South U Northeast Midwest West Other


1990




Year 1998




2006


S18% 4% 2%




21% 6% 4%




17% 7% 3%
I__ -s__


Market Share in Percentage

Figure 3-3. Florida CSA: share of total domestic airline passenger traffic from five U.S. regions
in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics.


25.0


2 20.0
-
15.0
.C
-

10.0
-
S55.0

^ 0.0










x West Midwest -*-Northeast ---South
13%

11% ----i----------i+----------i---
%12
x I
9 12% .-------. -- --. --- -- ---- -------- --. -- ------.----. .------ ---. .


11N % 8.-- -------- 9.0 10 --- ------.1 9----4 4 84 -
1% ---- -* --* -- -^ >* *----- -- - - __.L -- -.- -



% ----- ------ -- ---- --- ------------------ -- ---- ----- -- -* ----------- ------
0-

6% ------ ---------- --- ------ ----- -------- ------- -------


Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec
X West 8.4% 8.0% 9.5% 8.5% 8.0% 8.5% 8.8% 8.5% 6.7% 7.8% 8.1% 9.2%
Midwest 8.5% 9.7% 11.7% 9.3% 7.6% 7.7% 7.7% 7.1% 5.6% 7.6% 8.1% 9.4%
--Northeast 8.2% 9.0% 10.1% 9.4% 7.8% 7.4% 8.4% 8.5% 6.0% 7.7% 8.4% 9.0%
-South 7.6% 8.1% 9.6% 8.6% 8.4% 8.6% 8.7% 8.2% 6.6% 8.2% 8.3% 9.0%
Figure 3-4. Florida CSA: monthly seasonal pattern of domestic airline passenger traffic from
four U.S. regions between 1990 and 2006. Source: Bureau of Transportation
Statistics.

Domestic airline passenger traffic by destination CSA

Nearly 648 million domestic airline passengers traveled to Florida from 1990 to 2006, with

an annual average of more than 38 million airline passengers during the 17-year period. Also,

Florida experienced growth in domestic airline passengers of 83% from 28.27 million in 1990 to

51.71 million in 2006. The destination CSAs of Orlando, South Florida, Tampa-St. Petersburg,

Jacksonville, and Fort Myers accounted for 92% of the total domestic airline passenger traffic

market traveling to Florida. The "Other" category includes the remaining 17 destination CSAs in

the state.

Moreover, the South Florida CSA received 38% of the total domestic passenger traffic,

followed by the Orlando CSA (28%), Tampa-St. Petersburg CSA (16%), Fort Myers CSA (6%),

and Jacksonville CSA (5%). The Jacksonville CSA experienced the largest relative growth









(130%) during the 17-year period, followed by the Fort Myers CSA (121%). The South Florida

CSA reported the smallest relative growth (59%) during the same time period, as shown in

Figure 3-5.


* Jacksonville Ft. Myers Other
Tampa-St. Petersburg U Orlando U South Florida


20.0


S 15.0


S10.0


5.0


0.0 ".




Year

Figure 3-5. Total domestic airline passenger traffic traveling to the destination CSAs in Florida
between 1990 and 2006. Source: Bureau of Transportation Statistics.

South Florida CSA domestic airline passenger traffic

Total domestic airline passenger traffic to the South Florida CSA increased 58% from

nearly 12.0 million in 1990 to approximately 19.0 million in 2006. The South region accounts for

the largest share with 38% of the domestic airline passenger traffic traveling to the South Florida

CSA. The Northeast region (36%) ranks second followed by the Midwest (15%) and West (7%)

regions. In spite of having the smallest share of the four major U.S. regions, the West region

experienced the greatest relative growth (177%) in the 17-year period from 0.61 million airline

passengers in 1990 to approximately 1.69 million airline passengers in 2006. The South region

experienced the second largest relative growth (57%) followed by the Northeast (51%) and










Midwest (42%) regions. Figure 3-6 illustrates total domestic airline passenger traffic from five

U.S. regions traveling to the South Florida CSA from 1990 to 2006.

The South region has kept the largest share (38%) during the 17-year period. Meanwhile

the West region has increased its share from 5% in 1990 to 9% in 2006, whereas the Northeast

region's share has declined from 39% in 1990 to 37% in 2006. Figure 3-7 illustrates the share of

domestic airline passenger traffic from five U.S. regions traveling to the South Florida CSA in

the years 1990, 1998, and 2006.

Domestic airline passengers from all U.S. regions traveled to the South Florida CSA more

frequently during March than any other month. September registered the lowest level of airline

passenger traffic from all U.S. regions. Figure 3-8 shows the seasonal pattern of domestic airline

passenger traffic traveling to the South Florida CSA.


SOther West Midwest U Northeast U South



8.0
a 7.0
o 6.0
5.0
4.0
S3.0
2.0



1 0 i m .. o z -
PI 1.0
0.0 ................




Year


Figure 3-6. South Florida CSA: total domestic airline passenger traffic from five U.S. regions
between 1990 and 2006. Source: Bureau of Transportation Statistics.











1 South U Northeast Midwest West Other


1990




Year 1998




2006


14a 5?o 4%




16% 6% 9%




13% 9% 4%


Market Share in Percentage

Figure 3-7. South Florida CSA: share of total domestic airline passenger traffic from five U.S.
regions in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics.


13.0%


11.0%


9.0%


7.0%


5.0%


3.0%


x West Midwest -*-Northeast South



------- ---- ------------ ----- I ----------------- ---------- --






-I I I I I I I
-------- L --- -- ------~---- -- - - - - - -


Mar


May


June


Aug


Sept


Nov


x West 9.1% 8.4% 9.9% 9.1% 8.4% 8.0% 8.5% 8.5% 6.2% 7.1% 7.9% 9.0%
Midwest 9.4% 10.4% 12.2% 9.6% 7.6% 7.0% 7.1% 6.8% 5.2% 6.9% 8.0% 9.8%
-A-Northeast 9.2% 9.6% 10.5% 9.7% 7.9% 6.8% 7.9% 8.1% 5.6% 7.2% 8.2% 9.3%
--South 8.5% 8.7% 10.1% 9.0% 8.7% 8.0% 8.2% 7.8% 6.2% 7.5% 8.1% 9.1%

Figure 3-8. South Florida CSA: monthly seasonal pattern of domestic airline passenger traffic
from four U.S. regions between 1990 and 2006. Source: Bureau of Transportation
Statistics.









Orlando CSA domestic airline passenger traffic

Nearly 187 million domestic airline passengers traveled to the Orlando CSA from 1990 to

2006, with an annual average of 11.0 million domestic airline passengers during the 17-year

period. Domestic airline passenger traffic to the Orlando CSA more than doubled from 7.45

million in 1990 to approximately 15.62 million in 2006. Among the major U.S. regions, the

South region accounted for the largest share with 37% of domestic airline passenger traffic

traveling to the Orlando CSA. The Northeast region (30%) ranked second followed by the

Midwest (22%) and West (8%) regions.

In spite of having the smallest share of the four major U.S. regions, the West region

experienced a three-fold increase in the number of airline passengers during the 17-year period

from 0.45 million in 1990 to approximately 1.37 million in 2006. The Northeast region

experienced the second largest relative growth (124%) followed by the Midwest (109%) and

South (78%) regions. Figure 3-9 illustrates total domestic airline passenger traffic from five U.S.

regions traveling to the Orlando CSA from 1990 to 2006.

Even though the South region registered the largest share (37%) in 2006, its share has

declined since 1990 when it accounted for 42% of total domestic airline passengers traveling to

the Orlando CSA. On the other hand, two other major U.S. regions have experienced growth in

their shares. The Midwest region has increased its share from 29% in 1990 to 31% in 2006.

Figure 3-10 illustrates the share of domestic airline passenger traffic from five U.S. regions

traveling to the Orlando CSA in the years 1990, 1998, and 2006.

Domestic airline passengers from four major U.S. regions traveled to the Orlando CSA

more frequently during March than any other month. September registered the lowest level of

airline passenger traffic from all U.S. regions, as illustrated in Figure 3-11.










Other West Midwest 0 Northeast U South


r 6.0

5.0


C 3.0

: 2.0

n 1.0

0.0 --n -- n -- \ C- o a n .t




Year


Figure 3-9. Orlando CSA: total domestic airline passenger traffic from five U.S. regions between
1990 and 2006. Source: Bureau of Transportation Statistics.


U South U Northeast Midwest West I Other


1990




Year 1998




2006


m 21% 6% 2%




24% 8% 2%



21% 9% 3%
-!------- __S


Market Share in Percentage

Figure 3-10. Orlando CSA: share of total domestic airline passenger traffic from five U.S.
regions in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics.










x West Midwest --Northeast --South
11%
Q) I I I I I I I I I I I
I I I I I I I I I I I

I I I I I I I I I I I
idwest 7.810% 8.9----------- ------8.1 8.7 8.5 7 6.1 8.3 8.0 8.6---------







-Northeast% ------ 8.3-- 9---- 9--- 8. ------- 8.9 9.1-- -----6.-- 8.1-------8.3-- 8.2
8 % ----- - - ------- -. - -.----- -------- --------
7% --------------------------------------- -- ----








Midwest 7.8% 8.9% 10.5% 8.9% 8.1% 8.7% 8.5% 7.7% 6.1% 8.3% 8.0% 8.6%


--ESouth 7.4% 8.1% 9.5% 8.6% 8.6% 8.9% 8.9% 8.1% 6.6% 8.5% 8.2% 8.5%


Figure 3-11. Orlando CSA: monthly seasonal pattern of domestic airline passenger traffic from
four U.S. regions between 1990 and 2006. Source: Bureau of Transportation
Statistics.

Tampa-St. Petersburg CSA domestic airline passenger traffic

The Tampa-St. Petersburg CSA accounts for 16% of the total domestic airline passenger

traffic traveling to Florida making it the third largest destination market in the state. Its airline

passenger traffic more than doubled from 4.07 million in 1990 to approximately 8.25 million in

2006. The South region held the largest share with 43% of the domestic airline passenger traffic

traveling to the Tampa-St. Petersburg CSA. The Northeast region (26%) ranked second followed

by the Midwest (24%) and West (6%) regions.

In spite of having the smallest share of the four U.S. regions, the West region experienced

the greatest relative growth (357%) in the 17-year period from 0.15 million airline passengers in

1990 to approximately 0.69 million airline passengers in 2006. The Northeast region experienced










the second largest relative growth (113%) followed by the Midwest (85%) and South (84%)

regions as illustrated in Figure 3-12.

Both the South and Midwest regions have experienced a decrease in the share of domestic

airline passenger traffic to the Tampa-St. Petersburg CSA between 1990 and 2006. In contrast,

the Northeast and West regions have increased their share during the same time period. Figure 3-

13 shows the evolution of the share of domestic airline passenger traffic traveling to the Tampa-

St. Petersburg CSA.

Domestic airline passengers from every U.S. region traveled to the Tampa-St. Petersburg

CSA more frequently during March than any other month. September registered the lowest level

of airline passenger traffic from the four major U.S. regions. Figure 3-14 shows the seasonal

pattern of domestic airline passenger traffic traveling to the Tampa-St. Petersburg CSA.


S Other West Midwest Northeast U South



4.0
S 3.5
3 3.0
2.5
2.0
1.5
1.0
P 0.5






Year

Figure 3-12. Tampa-St. Petersburg CSA: total airline passenger traffic from five U.S. regions
between 1990 and 2006. Source: Bureau of Transportation Statistics.










1 South U Northeast Midwest West Other


1990




Year 1998




2006


I -----) ----




2'O o 5 o 1%

-2% 8


21% 8% 1%


Market Share in Percentage

Figure 3-13. Tampa-St. Petersburg CSA: share of total domestic airline passenger traffic from
five U.S. regions in 1990, 1998, and 2006. Source: Bureau of Transportation
Statistics.


12%

11%

10%

9%

8%

7%

6%

5%


x West Midwest ---Northeast --South


I- i ----- T ------ i ------ ----- ------i ----- ----------- ------

-- ----------- ------ I---------- ------ ----- I------ ------

i ---- ----- r --..------ ------ i---- T ----------------- T ------ iiiA-i
.. "--- r---- T- -1-- T---- ----- r----

I ---- I I------------ I I I I I I I I------------
i i i i i i i i i i


Mar


June


Aug


x West 8% 8% 10% 8% 8% 8% 9% 8% 7% 8% 8% 9%
Midwest 8% 9% 12% 9% 8% 8% 8% 7% 6% 8% 8% 9%
---Northeast 8% 9% 10% 9% 8% 8% 9% 9% 6% 8% 8% 9%
--South 7% 8% 10% 9% 8% 9% 9% 8% 7% 8% 8% 9%
Figure 3-14. Tampa-St. Petersburg CSA: monthly seasonal pattern of domestic airline passenger
traffic from four U.S. regions between 1990 and 2006. Source: Bureau of
Transportation Statistics.









Jacksonville CSA domestic airline passenger traffic

Total domestic airline passenger traffic traveling to the Jacksonville CSA increased 130%

from 1.16 million in 1990 to approximately 2.66 million in 2006. The South region has the

largest share with almost 71% of the total domestic airline passenger traffic traveling to the

Jacksonville CSA. The Northeast (14%) and Midwest (14%) regions ranked second followed by

the West region (<1%). Figure 3-15 shows total domestic airline passenger traffic from each U.S.

region traveling to the Jacksonville CSA from 1990 to 2006.

The South region has kept the largest market share despite experiencing a reduction from

73% to 69% market share from 1990 to 2006. In contrast, share from the Midwest region has

increased since 1990. Also, the Northeast region recorded an increase from 16% to 17% in its

market share of domestic airline passengers traveling to the Jacksonville CSA during the same

time period. Figure 3-16 shows the evolution of the share of domestic airline passenger traffic

traveling to the Jacksonville CSA since 1990.

Total domestic airline passenger traffic from the South, Northeast, and Midwest regions

traveled more frequently to the Jacksonville CSA during March, while passengers from the West

region did so in July. January experienced the lowest levels of domestic airline passenger traffic

originating from the South, Northeast, and Midwest regions. The West region traveled less

frequently during September to the Jacksonville CSA than any other month. Figure 3-17

illustrates monthly seasonal patterns of domestic airline passenger traffic traveling to the

Jacksonville CSA from 1990 to 2006.











1 Other West Midwest U Northeast U South


W, 2.U


1.5
E

. 1.0


S 0.5

P 0.0


0z C m >n ^o 0 3 o -m Cl Cm ~ t
~ z z z 00 00 000
~ z z z 00 00 000
--------------------------------C Cl(l C(Cl ( C


Year


Figure 3-15. Jacksonville CSA: total domestic airline passenger traffic from five U.S. regions
between 1990 and 2006. Source: Bureau of Transportation Statistics.


1 South U Northeast Midwest West


1990




Year 1998




2006


11% 1%


16% 0%




12% 2%


Market Share in Percentage


Figure 3-16. Jacksonville CSA: share of total domestic airline passenger traffic from four U.S.
regions in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics.










x West Midwest -*-Northeast --South
11.0% i i ,
11.0%


I I I I I I I I I I I

9.0% ------- I I I I I----------------------------

8.0% ---- ---- --- ----- ---- ----- -------- --- ---- ---- ----

4^ 7.0% 4' -3^ ^ -- -- -- - -- -- -- -- ---- -- - --- -___ -_____ ______-

4^ 6.0% -------- ------------------- --- ------------------- ----------.----- ------------

5.0
7.0% .------------ ..----------.--.-----.--.--.----------






X West 7% 7% 9% 8% 9% 9% 10% 10% 7% 8% 7% 9%
Midwest 6% 7% 10% 9% 8% 9% 9% 9% 7% 9% 8% 8%
-*-Northeast 6% 7% 9% 9% 8% 8% 9% 9% 7% 9% 9% 9%
---South 7% 7% 9% 9% 9% 9% 9% 9% 7% 8% 8% 9%
Figure 3-17. Jacksonville CSA: monthly seasonal pattern of domestic airline passenger traffic
from four U.S. regions between 1990 and 2006. Source: Bureau of Transportation
Statistics.

Fort Myers CSA domestic airline passenger traffic

Total domestic airline passenger traffic traveling to the Fort Myers CSA increased 121%

from 1.57 million in 1990 to approximately 3.47 million in 2006. The South region recorded the

largest share with 38% of the total domestic airline passenger traffic traveling to the Fort Myers

CSA. The Midwest region (35%) ranked second, followed by the Northeast (26%) and West

(< %) regions. Four major U.S. regions experienced growth in terms of number of airline

passengers traveling to Fort Myers CSA during the 17-year period. The West region registered

the largest increase at 606%, followed by the Northeast (164%), Midwest (158%), and South

(67%) regions. Figure 3-18 shows total domestic airline passenger traffic from five U.S. regions

traveling to the Fort Myers CSA from 1990 to 2006.









The South region experienced the largest drop of market share of total domestic airline

passenger traffic traveling to the Fort Myers CSA, decreasing from 45% in 1990 to 34% in 2006.

By 2006, the Northeast, Midwest, and South regions held one third of the total share each. The

Midwest region's share increased from 28% to 33% during the same time period. The Northeast

region also increased its share from 27% in 1990 to 32% in 2006. Figure 3-19 shows the

evolution of the share of domestic airline passenger traffic traveling to the Fort Myers CSA since

1990. Total domestic airline passenger traffic from the South, Northeast, and Midwest regions

traveled more frequently to the Fort Myers CSA during March, while passengers from the West

region did so in December. September experienced the lowest levels of domestic airline

passenger traffic originating from four major U.S. regions. Figure 3-20 illustrates monthly

seasonal patterns of domestic airline passenger traffic traveling to the Fort Myers CSA from

1990 to 2006.


1.4
. -1.2
1.0
S08
S0.6
S0.4
0.2
P a nn


LI


Year

Figure 3-18. Fort Myers CSA: total domestic airline passenger traffic from five U.S. regions
between 1990 and 2006. Source: Bureau of Transportation Statistics.


Er Other West 0 Northeast Midwest 0 South











1 South U Northeast Midwest West


1990




Year 1998




2006


28% 0%


33% 1%


Market Share in Percentage

Figure 3-19. Fort Myers CSA: share of total domestic airline passenger traffic from four U.S.
regions in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics.


18.0%

16.0%

14.0%

12.0%

10.0%

8.0%

6.0%

4.0%

2.0%

0.0%


x West Midwest -*-Northeast South


-i- --------- ---- -----------------.---. --
x
----- ----------x---|------------------------|------.------------ --- ------
I-- I I I I I I I I

------ --'------- -- --------------





------------------| -- -- -------------------- -- ------------ -----T
x
------ -------- ---- ---------------L_-- -11-_ ------ l_--- -


May


June


Sept


X West 12% 9% 14% 11% 6% 5% 5% 4% 3% 5% 11% 15%
Midwest 10% 12% 16% 11% 6% 6% 5% 5% 4% 7% 8% 10%
---Northeast 9% 11% 13% 11% 7% 6% 7% 7% 5% 7% 9% 9%
-South 9% 10% 12% 10% 8% 7% 7% 7% 5% 8% 8% 9%

Figure 3-20. Fort Myers CSA: monthly seasonal pattern of domestic airline passenger traffic
from four U.S. regions between 1990 and 2006. Source: Bureau of Transportation
Statistics.









International airline passenger traffic by world region

The Latin America region accounted for the largest share with 65% of the total

international airline passenger traffic traveling to Florida from 1990 to 2006, followed by Europe

(23%) and Canada (11%). Also, Latin America has experienced the largest increase (79%) in

airline passengers since 1990, followed by Europe (63%). In contrast, airline passenger traffic

from Canada has dropped 5% from 1990 to 2006. In 2006, the Latin America and Europe region

almost matched its previous highest level of airline passenger traffic reached in 2000, while

Canada failed to improve its highest airline passenger level (approximately 1.14 million airline

passengers) reached in 1990. Figure 3-21 shows total international airline passenger traffic from

four world regions traveling to Florida from 1990 to 2006.

Latin America increased its share of the international airline passenger traffic traveling to

Florida from 61% in 1990 to 68% in 2006. Meanwhile, Canada's share dropped 7% during the

same time period. Europe has kept its share at 23% during the 17-year period as presented in

Figure 3-22.

International airline passengers from Canada traveled to Florida more frequently during

March, while September registered the lowest level. High airline traffic levels in March could be

attributed to unfavorable weather in Canada enticing airline passengers to travel to Florida.

Likewise, low airline traffic levels in September could be attributed to favorable weather in

Canada dissuading airline passengers from travel to Florida. The Latin America and Europe

regions registered the highest airline passenger traffic during July. Airline passenger levels from

Europe traveled less in January, while September registered the lowest level of airline passengers

traveling from Latin America as shown in Figure 3-23.









I Other 0 Canada Europe U Latin America


Qfl


S 6.0

4.0
t-

S 2.0

S ( 0.0


-_,-7 1 I--7WI


Year

Figure 3-21. Florida CSA: total international airline passenger traffic from four world regions
between 1990 and 2006. Source: Bureau of Transportation Statistics.


U Latin America Europe I Canada U Other


1990


22%


18% 0%


Year 1998



2006


l0Uo 1%



10' o 0%


Market Share in Percentage


Figure 3-22. Florida CSA: share of total international airline passenger traffic from four world
regions in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics.


~IIIIIIIIII~ ~L


I











n-Other ~- Canada Europe --Latin America


16%



10% --- C---- -144---i0---- ------ --------- F---- ----%----
I I I I I I I I I I I





08% 7%- 7 ----- ------ -----
I I I I I I
= I I I I I I I I


Cn3 14% ------14%----- -1---------- ------------ --1------5-- --4
I I I I I I I I I I
I I I I I I I I I



10% ---------------------------------








2006% 7%,nlv goa ome94ioai 7%snesdrn 8% 1%1 00%7-
I I I I I I I









--Latin America 10% 8% 18% 18% 17% 18% 110% 110% 7% 7% 17% 1 9%







Figure 3-23. Florida CSA: monthly seasonal pattern of total international airline passenger traffic
from four world regions between 1990 and 2006. Source: Bureau of Transportation



International airline passenger traffic by destination CSA

More than 155 million international airline passengers traveled to Florida from 1990 to



















Petersburg CSA (3%). The South Florida CSA experienced the largest relative growth (93%)


during the 17-year period, followed by the Orlando CSA (63%). The Tampa-St. Petersburg CSA
reported a 45% decline in number of international airline passengers during the same time
I I I I I I I I I I I



'-- Other 8 8 '70 6 80 '70 '70 '70 bo l3o 90






Figure 3-23. Florida CSA: monthly seasonal pattern of total international airline passenger traffic










2006, with an annual average of approximately 9.14 million airline passengers during the 17-


year period. Also, Florida experienced growth in international airline passenger traffic of 610


from 6.53 million in 1990 to 10.5 million in 2006. The destination CSAs of South Florida,







international passenger traffic, followed by the Orlando CSA (150), and the Tampa-St.


Petersburg CSA (3o). The South Florida CSA experienced the largest relative growth (930%)


during the 17-year period, followed by the Orlando CSA (63%). The Tampa-St. Petersburg CSA


reported a 45% decline in number of international airline passengers during the same time









period. Figure 3-24 presents total international airline passenger traffic traveling to the top three

CSAs in Florida between 1990 and 2006.


10.0
S 8.0

.= 6.0

o 4.0

a 2.0

0.0


Tampa U Orlando U South Florida











AA1 11111aIMI **iiiiiiaAii i
| | | | | | | | | | | | |


z Cl m ~~ t- 0 0 l C ~
~~ z z z z z z0
~~ z z z z z z0
Cl Cl Cl Cl Cl Cl C


Year


Figure 3-24 Total international airline passenger traffic traveling to the top three destination
CSAs in Florida between 1990 and 2006. Source: Bureau of Transportation Statistics.

South Florida CSA international airline passenger traffic

International airline passenger traffic to the South Florida CSA increased 63% from 5.21

million in 1990 to nearly 8.46 million in 2006. The Latin America region accounts for the largest

share with 77% of the international airline passenger traffic traveling to the South Florida CSA.

The Europe region (15%) ranks second, followed by Canada (7%). Also, Latin America recorded

the largest relative growth (80%) in the 17-year period from 3.76 million in 1990 to

approximately 6.77 million airline passengers in 2006. The Europe region experienced the

second largest relative growth at 35%. Figure 3-25 illustrates total international airline passenger

traffic traveling to the South Florida CSA from 1990 to 2006.










Latin America increased its share of international airline passenger traffic traveling to the

South Florida CSA from 72% in 1990 to 80% in 2006. Meanwhile, Canada's share dropped from

12% to 7% during the same time period. The Europe region's share decreased from 16% to 13%

during the 17-year period. Figure 3-26 illustrates the evolution of the share of international

airline passenger traffic from four world regions traveling to the South Florida CSA from 1990

to 2006.

International airline passengers from Canada traveled to the South Florida CSA more

frequently during March, while September registered the lowest level. Latin America and Europe

registered the highest levels of airline passenger traffic during July. Airline passenger from

Europe traveled less frequently in May, while September registered the lowest level of airline

passengers traveling from Latin America. Figure 3-27 shows monthly seasonal pattern of

international airline passenger traffic traveling to the South Florida CSA.


1 Other M Canada Europe E Latin America


7,000
6,000
5,000
4,000
3,000
2,000
1,000
0


z Cl CI D ~


z -l~rI\
z z z z z
z z z z z
- \Cl Cl Cl ClCl Cl Cl


Year

Figure 3-25. South Florida CSA: total international airline passenger traffic from four world
regions between 1990 and 2006. Source: Bureau of Transportation Statistics.


1 r7 I I I I






dodo ANdl NUMMENNOWN a I


/











1 Latin America Europe N Canada 0 Other


1990




Year 1998




2006


16% 12% 0%




14% 1%




13% 7% 0%


Market Share in Percentage

Figure 3-26. South Florida CSA: share of total international airline passenger traffic from four
world regions in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics.


15%


13%


11%


9%


7%


5%


3%


--Other --Canada Europe -4-Latin America


I I I I I I I I I I I
I I I I I I I I I I I
I I I I I I I I I I I
I I I I I I I I I I I
I I I I I I I I I I
L----- ----- --- --- .---- ----- L----- L----- L----- L----- ------

.--.-----. --- ----- ---- ----- ------


___ _____ L_---L---___--- ---- ------ ----- ------
I I I I I I I I I I I



I I I I I I I I I I I


July Aug


Sept


Nov Dec


----Other 8.3% 7.0% 8.5% 7.9% 7.7% 8.4% 9.4% 9.2% 7.3% 8.2% 8.2% 9.8%
---Canada 12.4% 12.9% 13.1% 9.4% 5.2% 4.2% 5.6% 5.6% 4.1% 5.6% 9.7% 12.1%
Europe 7.8% 7.8% 9.1% 8.4% 7.2% 7.6% 10.1% 8.9% 7.3% 8.9% 8.2% 8.8%
----LatinAmerica 9.4% 7.4% 8.3% 7.9% 7.6% 8.2% 10.3% 10.2% 7.0% 7.1% 7.6% 8.8%


Figure 3-27. South Florida CSA: monthly seasonal pattern of total international airline passenger
from four world regions between 1990 and 2006. Source: Bureau of Transportation
Statistics.









Orlando CSA international airline passenger traffic

International airline passenger traffic to the Orlando CSA almost doubled from 0.86

million in 1990 to nearly 1.66 million in 2006. The Europe region dominated the Orlando CSA

market with a 68% share of total international airline passenger traffic. The Canada region (16%)

ranks second, followed by Latin America (15%). Also, Latin America recorded the largest

relative growth (152%) in the 17-year period from 0.10 million in 1990 to approximately 0.26

million airline passengers in 2006. The Europe region experienced the second largest relative

growth at 92% followed by Canada (64%). Figure 3-28 illustrates total international airline

passenger traffic traveling to the Orlando CSA from 1990 to 2006.

The Europe region recorded a market share of 65% of total international airline passenger

traffic traveling to the Orlando CSA in 1990. This U.S. region showed an increase of 70% in

1998, but it decreased to 65% by 2006. In contrast, Latin America increased its share from 12%

in 1990 to 16% in 2006. Meanwhile, Canada's share decreased from 22% to 19% during the

same time period. Figure 3-29 illustrates the evolution of the share of international airline

passenger traffic from four world regions traveling to the Orlando CSA from 1990 to 2006.

International airline passengers from Europe traveled to the Orlando CSA more frequently

during July, while January registered the lowest level. The Latin America region registered the

highest level of airline passenger traffic during July, while Canada did so in March. Airline

passenger from Latin America and Canada traveled less frequently in September as shown in

Figure 3-30.









1 Other U Latin America I Canada Europe


1,200
1,000
800
600
400

200 U
0


i L


: ih i [h -


Year
Figure 3-28. Orlando CSA: total international airline passenger traffic from four world regions
between 1990 and 2006. Source: Bureau of Transportation Statistics.


Europe 0 Canada U Latin America U Other


1990



Year 1998



2006


69%


66%


1300


19%


Market Share in Percentage

Figure 3-29. Orlando CSA: share of total international airline passenger traffic from four world
regions in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics.












I Canada Europe --Latin America


18%
1 I I I I I--- I ----I ----I-- I I I-------
I I I I I I I I I I I
S16% --------------------------- ------------------------------------
I I I I I I I I I I I



( 10% -.-- --- ----- r -----. --- -r--- ....r............... .
I I I I I I I I I I I




1% -------- ------------- L -- I-----------I-----.-----.---
I I I I I I I I I I I
I I I I I I I I I I I



S12% ----- L --- -----L--
I I I I I I I I I I I








2% ----- ----- ----- ------- ----- ------ __ -----_____ _----_____ _----_____

0% ----- ----- -----,
ed 6% ,,-----L ----L ---- ___, .....


-*-Canada 12% 14% 13.9% 10.8% 5.8% 4.7% 6.2% 6.2% 3.9% 5.5% 7.8% 9.6%
Europe 5% 5% 6.1% 7.2% 9.9% 10.1% 12.3% 11.8% 10.9% 10.6% 5.8% 5.6%
--*LatinAmerica 9.05% 6.56% 7.2% 7.1% 6.4% 8.5% 15.4% 11.0% 6.0% 6.2% 6.0% 10.6%

Figure 3-30. Orlando CSA: monthly seasonal pattern of total international airline passenger
traffic from four world regions between 1990 and 2006. Source: Bureau of
Transportation Statistics.

Tampa-St. Petersburg CSA international airline passenger traffic

International airline passenger traffic to the Tampa-St. Petersburg CSA decreased from

0.43 million in 1990 to nearly 0.24 million in 2006. The Canada region dominated the Tampa-St.

Petersburg CSA market with 60% share of total international airline passenger traffic. The Latin

America region (21%) ranks second, followed by Europe (21%). Europe is the only region that

recorded a positive relative growth (78%) in the 17-year period. Both Canada and Latin America

experienced a decrease of 62% and 45% respectively. Figure 3-31 illustrates total international

airline passenger traffic traveling to the Tampa-St. Petersburg CSA from 1990 to 2006.

Canada has experienced a drastic decrease in its share of international airline passenger

traffic traveling to the Tampa-St. Petersburg CSA from 70% in 1990 to 47% in 2006. In contrast,

Europe increased its share from 10% in 1990 to 32% in 2006. Meanwhile, Latin America










increased its 1990 share of approximately 20% to 21% in 2006. Figure 3-32 illustrates the

evolution of the share of international airline passenger traffic from the world regions traveling

to the Tampa-St. Petersburg CSA from 1990 to 2006.

International airline passengers from Europe and Canada traveled to the Tampa-St.

Petersburg CSA more frequently during March, while passengers from Latin America did so in

July. Airline passengers from Canada and Latin America traveled less frequently in September,

while airline passengers from Europe did so in January. Figure 3-33 shows monthly seasonal

patterns of international airline passenger traffic traveling to the Tampa-St. Petersburg CSA from

1990 to 2006.


1 Other U Latin America Europe E Canada


300
.2 250
S 2200
150




0 1 1__ _1
J a100*
50





Year

Figure 3-31. Tampa-St. Petersburg CSA: total international airline passenger traffic from four
world regions between 1990 and 2006. Source: Bureau of Transportation Statistics.










Canada Europe U Latin America 0 Other


1990 uo 10%0




Year 1998 0%




2006 47o 32%,0%



Market Share in Percentage

Figure 3-32. Tampa-St. Petersburg CSA: share of total international airline passenger traffic
from four world regions in 1990, 1998, and 2006. Source: Bureau of Transportation
Statistics.


Canada Europe -*-Latin America
18% 0
I I I I I I I I I I I
S 1 6 % . ..- .. .- -- s .. ... . '-- . . . .- . .

1% -- ------- --------- ----- ------ ----- -L--------------------
*a 14% ..-----.- --------- -..---- L L L ------ L ------ L ------. L -----

12% --- --------- -- -----r-----r----- ----- r r ----- r ---r----

10% ..-----.-- --..---- ... -~'- .- ..----- .-.----- ----- -----

8% _L --L--- L ---- ---- -- -L- ----L---



08%


Jan Feb MarAprMayJueJuy-----Aug Sep----t----- OctNovDec------- ------------
Jma Feb Mar Apr May June July Aug Sept Oct Nov Dec


-*-Canada 11.3% 14.0% 15.8% 12.2% 5.9% 3.7% 4.4% 4.6% 3.5% 6.0% 9.1% 9.5%
Europe 6.4% 7.3% 10.9% 8.6% 7.2% 8.0% 10.1% 8.7% 8.1% 9.4% 7.8% 7.6%
-A-LatinAmerica 8.6% 7.7% 9.2% 8.7% 8.9% 9.8% 10.5% 9.5% 5.3% 6.8% 7.8% 7.3%
Figure 3-33. Tampa-St. Petersburg CSA: monthly seasonal pattern of total international airline
passenger traffic from four world regions between 1990 and 2006. Source: Bureau of
Transportation Statistics.









Airline Ticket Prices

Description, Selection, and Aggregation of Airline Ticket Price Data for Domestic Flights

Data on airline ticket prices1 for domestic flights used in this study were collected by the

Bureau of Transportation Statistics (BTS) and are available online. The Origin and Destination

Survey (DB 1B) Market was used to collect one way and round trip airline ticket prices from

several domestic origins traveling to Florida. According to the BTS, the DBlB-Market contains

(directional) origin and destination markets, which is a 10% sample of airline tickets from

reporting carriers. It includes items such as passengers, airfares, and distances for each

directional market, as well as information about whether the market was domestic or

international. Nearly 46.8 million observations reported in a quarterly basis were retrieved from

the DB1B-Market database between Quarter I, 1993 and Quarter IV, 2006. A market is defined

by the first departure airport on a ticket and the ultimate arrival airport.

The DB1B-Market database includes a market ID that identifies the itinerary of a

passenger and it is unique to each itinerary. If the market ID appears once in the data, it

represents that the itinerary was a one way trip. If the market ID appears more than once, the

itinerary was a round trip and may include multiple stops. For example, if a passenger traveled

from New York City to Jacksonville then to Miami and finally back to New York City, the

database will include the itinerary's market ID three times to represent each leg of the trip: (1)

New York City to Jacksonville, (2) Jacksonville to Miami, (3) Miami to New York City. For

simplification, all observations with more than two market IDs were dropped from the analysis.

That is, only itineraries involving two legs, origin to destination and destination to origin, were

selected to calculate average round trip airline ticket prices. For market IDs appearing twice in



1 The terms airline ticket prices and airfare are equivalent in this study and both are used interchangeably.









the database, the first entry represents the first leg of the itinerary and the second entry represents

the second (or returning) leg. This information is useful to determine the origin and the ultimate

destination of a flight. Also, itineraries with one market ID were selected to calculate one way

airline ticket prices.

Within the DB1B-Market database, only those itineraries that included a city in Florida

were selected. Then, a few airports were selected to represent the average airline ticket price of

each U.S. region to follow the airline passenger data scheme. Itineraries from La Guardia

International Airport and John F. Kennedy International Airport from New York City and Logan

International Airport from Boston were selected as a proxy for the average airline ticket price for

the Northeast region. Midway Airport and Chicago O'Hare International Airport from Chicago

and Lambert-Saint Louis International Airport from Saint Louis were selected to calculate the

average airline ticket price for the Midwest region.

Hartsfield-Jackson Atlanta International Airport from Atlanta, Dallas/Fort Worth

International Airport and Fort Worth Alliance from Dallas, and George Bush Intercontinental

Airport and William P. Hobby Airport in Houston were selected to calculate the average airline

ticket price for the South region. Finally, itineraries from Los Angeles International Airport in

Los Angeles, Denver International Airport from Denver, and Seattle-Tacoma International

Airport in Seattle were selected to calculate the average airline ticket price for the West region.

Then, destination airports in Florida chosen for the airline ticket price analysis are the

following: Miami International Airport for the South Florida CSA, Orlando International Airport

for the Orlando CSA, Tampa International Airport for the Tampa-St. Petersburg CSA,

Jacksonville International Airport for the Jacksonville CSA, and the Southwest Florida Regional









Airport for the Fort Myers CSA. Also, an average of the airline ticket prices of the top five

destination CSAs was calculated to represent the average airline ticket price for the Florida CSA.

Once the airports from each origin region and destination CSA were chosen, averages were

calculated for each origin region traveling to each destination CSA. In other words, there were

two average airline ticket prices: (1) one way price, and (2) round trip price from each

originating region traveling to each destination CSA for every quarter from 1993 to 2006. For

example, all ticket price entries from New York and Boston to Miami were used to calculate the

airline ticket price from the Northeast region to the South Florida CSA for every quarter from

1993 to 2006.

Description, Selection, and Aggregation of Airline Ticket Price Data for International
Flights

Data on ticket prices for international flights were estimated using the passenger yield

reported by the Air Transport Association Group (ATAG). According to the ATAG web site,

"passenger yield" is the average price someone pays to fly one mile (excluding government taxes

and fees, which often constitute a substantial portion of an airline ticket).

Passenger yield data, which ATAG reports monthly by geographic region, use reports from

seven U.S. airlines and results are not adjusted for inflation or trip length. Data are based on

100% of scheduled service and reflects all "revenue" passengers, including those redeeming

frequent flyer miles (USD 0 airfare) for award travel. Passenger yield data are often used as one

indicator of recent U.S. airline market/pricing trends. The ATAG reports the passenger yields of

four different regions: (1) Domestic (U.S.), (2) Atlantic (Europe and Africa), (3) Latin (Latin

America), and (4) Pacific (Asia).

The passenger yield was then multiplied by the average distance from its originating world

region. Average distance was calculated using the T-100 International Market database which









reports the miles between airports. For example, in order to get an average distance between

Europe and South Florida CSA, an average distance from all the flights originated from Europe

to Miami International Airport in Miami was calculated. Then, this average distance was

multiplied by two to denote a round trip airline ticket and finally the passenger yield reported for

Atlantic flights was multiplied by the average distance to get the average airline ticket price of a

round trip flight from Europe to the South Florida CSA. Similar calculations were performed for

the Canada region and the Latin America region. Note that the passenger yield for domestic

flights was used to estimate an average ticket price of a round trip flight from Canada to Florida.

The following sections discuss nominal domestic airline ticket prices by U.S. region and nominal

international airline ticket prices by world region traveling to Florida and each destination CSA.

Domestic Airline Ticket Prices by U.S. Region

One way airline ticket prices on flights to Florida have decreased since 1993. Over the 15-

year period the Midwest region has experienced the largest decrease in one way airline ticket

prices (16%) while the South region records the smallest decrease at 1%. Passengers from the

West region paid more for one way airline tickets to Florida than any other region. This fact

could explain why this U.S. region recorded the lowest level of airline traffic among U.S. regions

traveling to Florida as discussed in the previous section. Distance could be attributed to the

higher prices in the West region. Note that the year 2000 recorded the highest price levels on one

way airline tickets on three of the four U.S. regions. Since 2001 prices have been decreasing

until 2005 where one way airline ticket prices started to increase. Higher oil prices could be

responsible for this recent upward trend. Figure 3-34 shows average one way trip airline ticket

prices from each U.S. region traveling to Florida from 1993 to 2006.










SNortheast Midwest U South West


600

500

400

300

200

100






Year

Figure 3-34. Florida CSA: average one way airline ticket prices for domestic flights from four
U.S. regions between 1993 and 2006. Source: Bureau of Transportation Statistics.

Round trip airline ticket prices to Florida have decreased 2% overall since 1993. But

airline passengers traveling from the Northeast region paid 3% more for round trip airline tickets

in 2006 compared to 1993. In contrast, the other three U.S. regions have experienced round trip

airline ticket price decreases over the same time period. The South region records the largest

decrease in round trip airline ticket prices (8%) while the West region records the smallest

decrease at 1%. As with one way airline ticket prices, airline passengers from the West region

paid more for round trip airline tickets to Florida than any other region. This fact could explain

why this U.S. region recorded the lowest level of airline traffic among U.S. regions traveling to

Florida. Distance could be attributed to the higher prices in the West region. Figure 3-35 shows

average round trip airline ticket prices from each U.S. region traveling to Florida from 1993 to

2006.
2006.










1 Northeast Midwest U South West


450
400
350
S 300
250
:. 200
150
100
50 0





Year

Figure 3-35. Florida CSA: average round trip airline ticket prices for domestic flights from four
U.S. regions between 1993 and 2006. Source: Bureau of Transportation Statistics.

Since 1993 both one way and round trip airline tickets from each U.S. region reported

higher prices (USD 353 and USD 366 respectively) during the first quarter of the calendar year

compared to the other three quarters. On the other hand, the third quarter reports the lowest

prices on one way (USD 319) and round trip airline tickets (USD 330) to Florida. Comparing

across U.S. regions, the Northeast region paid less for one way and round trip airline tickets

while the West region paid the most in all quarters. Also note that airline passengers from the

West and South regions paid more (16% and 2% respectively) for a one way airline ticket than

for a round trip airline ticket to Florida compared to the other two U.S. regions. Figure 3-36

illustrates the seasonality of average one way (OW) and round trip (RT) airline ticket prices

traveling to Florida from four U.S. regions during the period between 1993 and 2006.










I Northeast Midwest U All Regions U South West


600

500

400

S300 -

o 200

S 100 0

0

o o 9 o


Quarter
Figure 3-36. Florida CSA: quarterly seasonal pattern of average one way (OW) and round trip
(RT) airline ticket prices for domestic flights from four U.S. regions between 1993
and 2006. Source: Bureau of Transportation Statistics.

Domestic Airline Ticket Prices by Destination CSA

The national average price of a one way airline ticket to the South Florida CSA was the

most expensive (USD 378) among the five destination CSAs in Florida during the period

between 1993 and 2006. The South Florida CSA was also the only destination CSA that

experienced an increase of 4% in airline ticket prices during the same time period. The Orlando

CSA reported the cheapest ticket price (USD 312) during the 14-year period. One way airline

ticket prices to the Tampa-St. Petersburg CSA experienced the largest decrease (14%) from USD

350 in 1993 to USD 303 in 2006. The Jacksonville CSA ranked second with a 12% decrease,

followed by the Orlando CSA (10%), and the Fort Myers CSA (4%).

The average price of a round trip airline ticket traveling from the U.S. to the South Florida

CSA was the most expensive (USD 385) among the top five destination CSAs in Florida during









the period between 1993 and 2006. Both the South Florida CSA and the Fort Myers CSA

experienced an increase of 4% in their airline ticket prices during the same time period. The

Orlando CSA and the Fort Myers CSA reported the cheapest ticket price (USD 333) during the

14-year period. Round trip airline ticket prices to the Jacksonville CSA experienced the largest

decrease (10%) from USD 403 in 1993 to USD 363 in 2006. The Tampa-St. Petersburg CSA

ranked second with an 8% decrease, followed by the Orlando CSA (2%).

An individual analysis of airline ticket prices for each of the five destination CSAs is

presented next. It includes a discussion of one way and round trip airline ticket prices for

domestic flights originating from four U.S. regions traveling to a specific destination CSA.

South Florida CSA domestic airline ticket prices

One way airline ticket prices on flights to the South Florida CSA have increased 4% from

USD 339 in 1993 to USD 352 in 2006. Despite the overall increase, the average price in 2006

was lower than the average price over the 14-year period (USD 378). The highest price was

reported in 2000 when the one way airline ticket price rose to USD 430. Then, prices

experienced a steady decline from 2001 to 2005. The downward trend halted in 2006 when

prices increased again. Over the 14-year period the West region has experienced the largest

increase in one way airline ticket prices (15%) from USD 444 in 1993 to USD 511 in 2006. In

contrast, the Midwest region has experienced the largest decrease (9%) from USD 261 in 1993 to

USD 237 in 2006. Figure 3-37 shows average one way airline ticket prices from each U.S. region

traveling to the South Florida CSA from 1993 to 2006.










1 Northeast Midwest U South West


700
600
o 500
m 400
.3 300
200
100
0




Year

Figure 3-37. South Florida CSA: average one way airline ticket prices for domestic flights from
four U.S. regions between 1993 and 2006. Source: Bureau of Transportation
Statistics.

Round trip airline ticket prices to the South Florida CSA have increased 4% overall since

1993. Airline passengers traveling from the West region paid 11% more for round trip airline

tickets in 2006 compared to 1993. Similarly, the Northeast region experienced a 10% increase in

round trip airline ticket prices to the South Florida CSA. In contrast, the Midwest and South

regions have experienced round trip airline ticket price decreases (3% and 4% respectively) over

the same time period. As with one way airline ticket prices, airline passengers from the West

region paid more for round trip airline tickets to the South Florida CSA than any other region.

All four U.S. regions recorded the highest price in 2000. Figure 3-38 shows average round trip

airline ticket prices from each U.S. region traveling to the South Florida CSA from 1993 to 2006.










1 Northeast Midwest U South West


600 -

500

400

300

200

100

0




Year

Figure 3-38. South Florida CSA: average round trip air ticket prices for domestic flights from
four U.S. regions between 1993 and 2006. Source: Bureau of Transportation
Statistics.

Since 1993 both one way and round trip airline tickets from each U.S. region reported

higher prices (USD 401 and USD 403 respectively) during the first quarter of the calendar year

compared to the other three quarters. On the other hand, the third quarter reports the lowest

prices on one way (USD 362) and round trip airline tickets (USD 368) to the South Florida CSA.

Comparing across U.S. regions, the Northeast region paid less for its one way and round trip

airline tickets while the West region paid the most in all quarters. Also note that airline

passengers from the West and South regions paid more (16% and 4% respectively) for a one way

airline ticket than for a round trip airline ticket to the South Florida CSA compared to the other

two U.S. regions. Figure 3-39 illustrates the seasonality of average one way (OW) and round trip

(RT) airline ticket prices traveling to the South Florida CSA from four U.S. regions during the

period between 1993 and 2006.










SNortheast Midwest U All Regions U South West


600

500

400

300

S200

100






Quarter
Figure 3-39. South Florida CSA: quarterly seasonal pattern of average one way (OW) and round
trip (RT) airline ticket prices for domestic flights from four U.S. regions between
1993 and 2006. Source: Bureau of Transportation Statistics.

Orlando CSA domestic airline ticket prices

One way airline ticket prices on flights to the Orlando CSA decreased 10% from USD 330

in 1993 to USD 296 in 2006. The highest price was reported in 2000 when the one way airline

ticket price rose to USD 347. Then, price decreases followed until the year 2005 when prices

increased again. Over the 14-year period the Midwest region experienced the largest decrease in

one way airline ticket prices (16%) from USD 260 in 1993 to USD 218 in 2006. The Northeast

region ranks second with a 14% decrease in one way airline ticket prices, followed by the South

(9%) and West (6%) regions. Still, the West region paid more for one way airline tickets than

any other region. Figure 3-40 illustrates average one way airline ticket prices from each U.S.

region traveling to the Orlando CSA during the period between 1993 and 2006.







SNortheast Midwest U South West


600
500
400
300


200
100 -
0


IIIIIIIIIImmlI
OC miilZ-liii
z zn zo z zo
z z z z z ^0 0
Cl l l l C C00 0 0 0


Year
Figure 3-40. Orlando CSA: average one way airline ticket prices for domestic flights from four
U.S. regions between 1993 and 2006. Source: Bureau of Transportation Statistics.
Similarly to one way airline ticket prices, round trip airline ticket prices to the Orlando
CSA have decreased 2% overall since 1993. Airline passengers traveling from the South region
paid 6% less for round trip airline tickets in 2006 compared to 1993, accounting for the largest
decrease among U.S. regions. Similarly, the West region experienced a 5% decrease in round trip
airline ticket prices to the Orlando CSA. An opposite situation was experienced by the other two
U.S. regions. The Midwest and Northeast regions experienced price increases in their round trip
airline tickets (2% and 1% respectively) over the same time period. Airline passengers from the
West region paid more for round trip airline tickets to the Orlando CSA than any other region.
All four U.S. regions recorded the highest price in 2000. Figure 3-41 illustrates average round
trip airline ticket prices from each U.S. region traveling to the Orlando CSA from 1993 to 2006.


~i~ii
IIC~










1 Northeast Midwest 0 South West


450
400
350
300
250
: 200
150
100
50





Year

Figure 3-41. Orlando CSA: average round trip air ticket prices for domestic flights from four
U.S. regions between 1993 and 2006. Source: Bureau of Transportation Statistics.

One way and round trip airline tickets from any U.S. region reported higher prices (USD

368 and USD 353 respectively) during the first quarter of the calendar year compared to the

other three quarters. On the other hand, the third quarter reports the lowest prices on one way

(USD 331) and round trip airline tickets (USD 314) to the Orlando CSA. Comparing across U.S.

regions, the Northeast region paid less for one way and round trip airline ticket prices while the

West region paid the most in all quarters. Also note that airline passengers from the West region

paid more (19%) for a one way airline ticket than for a round trip airline ticket to the Orlando

CSA compared to the other three U.S. regions. Figure 3-42 illustrates the seasonality of average

one way (OW) and round trip (RT) airline ticket prices traveling to the Orlando CSA from four

U.S. regions during the period between 1993 and 2006.










SNortheast Midwest U All Regions U South West


500

400

.E 300

c 200

100

0

o 6 o 6o


Quarter
Figure 3-42. Orlando CSA: quarterly seasonal pattern of average one way (OW) and round trip
(RT) airline ticket prices for domestic flights from four U.S. regions between 1993
and 2006. Source: Bureau of Transportation Statistics.

Tampa-St. Petersburg CSA domestic airline ticket prices

One way airline ticket prices on flights to the Tampa-St. Petersburg CSA decreased 14%

from USD 350 in 1993 to USD 303 in 2006. The highest price was reported in 2000 when the

average one way airline ticket price totaled USD 364. Price decreases followed until the year

2005 when prices increased again but did not reach the levels of the year 2000. Over the 14-year

period the Midwest region experienced the largest decrease in one way airline ticket prices

(23%). The Northeast region ranks second with a 19% decrease in one way airline ticket prices,

followed by the South (11%) and West (6%) regions. The West region paid more for one way

airline tickets than any other region. Figure 3-43 illustrates average one way airline ticket prices

from each U.S. region traveling to the Tampa-St. Petersburg CSA during the period between

1993 and 2006.










SNortheast Midwest U South West


600

500 -

400

300

200

100

0






Figure 3-43. Tampa-St. Petersburg CSA: average one way airline ticket prices for domestic
flights from four U.S. regions between 1993 and 2006. Source: Bureau of
Transportation Statistics.

Similarly to one way airline ticket prices, round trip airline ticket prices to the Tampa-St.

Petersburg CSA have experienced a decrease (8%) since 1993. Airline passengers traveling from

the Northeast region paid 15% less for round trip airline tickets in 2006 compared to 1993,

accounting for the largest decrease among U.S. regions. Similarly, the West and South regions

experienced a 7% decrease in their round trip airline ticket prices to the Tampa-St. Petersburg

CSA, followed by the Midwest region where price decreased 5% over the same time period.

Airline passengers from the West region paid more for round trip airline tickets to the Tampa-St.

Petersburg CSA than any other region. Figure 3-44 shows average round trip airline ticket prices

from each U.S. region traveling to the Tampa-St. Petersburg CSA from 1993 to 2006.










1 Northeast Midwest U South West


450
400
350
300












Year
250
S200
a 150
100
50
0




Year

Figure 3-44. Tampa-St. Petersburg CSA: average round trip air ticket prices for domestic flights
from four U.S. regions between 1993 and 2006. Source: Bureau of Transportation
Statistics.

One way and round trip airline tickets from any U.S. region reported higher prices (USD

336 and USD 351 respectively) during the first quarter of the calendar year compared to the

other three quarters. On the other hand, the third quarter reports the lowest prices on one way

(USD 312) and round trip airline tickets (USD 324) to the Tampa-St. Petersburg CSA.

Comparing across U.S. regions, the Northeast region paid less for its one way and round trip

airline tickets while the West region paid the most in all quarters. Also note that airline

passengers from the West region paid more (21%) for a one way airline ticket than for a round

trip airline ticket to the Tampa-St. Petersburg CSA compared to the other three U.S. regions.

Figure 3-45 shows the seasonality of average one way (OW) and round trip (RT) airline ticket

prices traveling to the Tampa-St. Petersburg CSA from four U.S. regions during the period

between 1993 and 2006.










1 Northeast Midwest U All Regions U South West


500

400

300 -

200 -

100



o o o o
0 0 0


Quarter
Figure 3-45. Tampa-St. Petersburg CSA: quarterly seasonal pattern of average one way (OW)
and round trip (RT) airline ticket prices for domestic flights from four U.S. regions
between 1993 and 2006. Source: Bureau of Transportation Statistics.

Jacksonville CSA domestic airline ticket prices

One way airline ticket prices on flights to the Jacksonville CSA decreased 12% during the

period between 1993 and 2006. The highest price was reported in 1998 when the average one

way airline ticket price totaled USD 371. Over the 14-year period the Midwest region

experienced the largest decrease in one way airline ticket prices (20%). The Northeast region

ranks second with a 16% decrease in one way airline ticket prices, followed by the West region

(10%) and South (5%) regions. On average the West region paid more for one way airline tickets

than any other region during the 14-year period. The Northeast region reported the lowest one

way airline ticket price during the same time period. Figure 3-46 illustrates average one way

airline ticket prices from each U.S. region traveling to the Jacksonville CSA during the period

between 1993 and 2006.










1 Northeast Midwest U South West


600

500

400

300

200

100

0
1 0 0 0 Z C 00



Year

Figure 3-46. Jacksonville CSA: average one way airline ticket prices for domestic flights from
four U.S. regions between 1993 and 2006. Source: Bureau of Transportation
Statistics.

Similarly to one way airline ticket prices, round trip airline ticket prices to the Jacksonville

CSA have experienced a decrease (10%) since 1993. The highest round trip prices were reported

in 1993 when the average ticket price averaged USD 403 for a round trip flight to the

Jacksonville CSA. Airline passengers traveling from the Midwest region paid 13% less for round

trip airline tickets in 2006 compared to 1993, accounting for the largest decrease among U.S.

regions. Similarly, the Northeast region experienced an 11% decrease in its round trip airline

ticket prices to the Jacksonville CSA, followed by the West region with a 10% decrease and the

South region with a 6% decrease over the same time period. Airline passengers from the West

region paid more for round trip airline tickets to the Jacksonville CSA than any other U.S. region

as illustrated in Figure 3-47.










1 Northeast Midwest U South West


500

4 400

m 300

200

100






Year

Figure 3-47. Jacksonville CSA: average round trip air ticket prices for domestic flights from four
U.S. regions between 1993 and 2006. Source: Bureau of Transportation Statistics.

On flights to the Jacksonville CSA, one way and round trip airline tickets from any U.S.

region reported higher prices (USD 352 and USD 369 respectively) during the first quarter of the

calendar year compared to the other three quarters. On the other hand, the third quarter reports

the lowest prices on one way (USD 329) and round trip airline tickets (USD 341) to the

Jacksonville CSA. Comparing across U.S. regions, the Northeast region paid less for its one way

and round trip airline tickets while the West region paid the most in all quarters. Also note that

airline passengers from the West region paid more (21%) for a one way airline ticket than for a

round trip airline ticket to the Jacksonville CSA compared to the other three U.S. regions. Figure

3-48 shows the seasonality of average one way (OW) and round trip (RT) airline ticket prices

traveling to the Jacksonville CSA from four U.S. regions during the period between 1993 and

2006.










1 Northeast Midwest U All Regions U South West


= 600

500

400

300

a 200




0-' -i
100 0


o o 'o o 6


Quarter
Figure 3-48. Jacksonville CSA: quarterly seasonal pattern of average one way (OW) and round
trip (RT) airline ticket prices for domestic flights from four U.S. regions between
1993 and 2006. Source: Bureau of Transportation Statistics.

Fort Myers CSA domestic airline ticket prices

One way airline ticket prices on flights to the Fort Myers CSA decreased 4% during the

period between 1993 and 2006. The highest price was reported in 2000 when the average one

way airline ticket price totaled USD 380. Over the 14-year period the Northeast and Midwest

regions experienced a decrease in one way airline ticket prices (18% and 11%, respectively). In

contrast, the West region reported a 4% increase in one way airline ticket prices, followed by the

South region with a 1% increase. On average, the West region paid more for one way airline

tickets than any other region during the 14-year period, while the Northeast region reported the

lowest one way airline ticket price during the same time period. Figure 3-49 illustrates average

one way airline ticket prices from each U.S. region traveling to the Fort Myers CSA during the

period between 1993 and 2006.










Northeast Midwest U South West



600

500

400

300

200

100

0
1 0 0 0 Z C 00
0 '- 0 0",l ",l ",l ",l 0",l 0l 0 0,l


Year

Figure 3-49. Fort Myers CSA: average one way airline ticket prices for domestic flights from
four U.S. regions between 1993 and 2006. Source: Bureau of Transportation
Statistics.

Unlike one way airline ticket prices, round trip airline ticket prices to the Fort Myers CSA

have increased 4% since 1993. The highest round trip prices were reported in 2000 when the

average ticket price averaged USD 347 for a round trip flight to the Fort Myers CSA. Airline

passengers traveling from the West region paid 8% more for round trip airline tickets in 2006

compared to 1993, accounting for the largest increase among U.S. regions. Similarly, the

Midwest region experienced a 7% increase in round trip airline ticket prices to the Fort Myers

CSA. The Northeast region was the only U.S. region that experienced a decrease in round trip

airline ticket prices (1%), while round trip airline ticket prices from the South region in 2006

were the same as the prices reported in 1993. Airline passengers from the West region paid more

for round trip airline tickets to the Fort Myers CSA than any other region. Figure 3-50 illustrates

average round trip airline ticket prices from each U.S. region traveling to the Fort Myers CSA

from 1993 to 2006.










SNortheast Midwest U South West


450 r
400 j -
350
300
250
200
150
100
50

0 .. .. .. ......



Year

Figure 3-50. Fort Myers CSA: average round trip air ticket prices for domestic flights from four
U.S. regions between 1993 and 2006. Source: Bureau of Transportation Statistics.

Unlike the other four destination CSAs, the Fort Myers CSA reported higher prices in one

way and round trip airline tickets from any U.S. region during the third quarter of the calendar

year compared to the other three quarters. On the other hand, the first quarter reported the lowest

prices on one way (USD 297) and round trip airline tickets (USD 303) to the Fort Myers CSA.

Comparing across U.S. regions, the Northeast region paid less for one way and round trip airline

tickets while the West region paid the most in all quarters. Also note that airline passengers from

the West region paid on average 19% more for a one way airline ticket than for a round trip

airline ticket to the Fort Myers CSA, followed by the South region with 7%. Figure 3-51 shows

the seasonality of average one way (OW) and round trip (RT) airline ticket prices traveling to the

Fort Myers CSA from four U.S. regions during the period between 1993 and 2006.










SNortheast Midwest U All Regions U South West


t-I
600

y, 500

= 400

300

200
100






Quarter
Figure 3-51. Fort Myers CSA: quarterly seasonal pattern of average one way (OW) and round
trip (RT) airline ticket prices for domestic flights from four U.S. regions between
1993 and 2006. Source: Bureau of Transportation Statistics.

International Airline Round Trip Airline Ticket Prices by World Region

Annual average prices of round trip airline tickets for international flights to Florida have

increased 4% overall since 1995. But airline passengers traveling from Latin America paid 15%

less for round trip airline ticket in 2006 (USD 342) compared to 1995 (USD 402). Also, round

trip airline ticket prices from Canada have experienced a decrease of 4% from USD 360 in 1995

to USD 346 in 2006. In contrast, Europe reported a round trip airline ticket price increase of 15%

during the same time period. Figure 3-52 illustrates average round trip airline ticket prices from

three world regions from 1995 to 2006.

During the period between 1995 and 2006, round trip airline ticket prices from Canada

reported an annual average price of USD 354. February recorded the highest price equivalent to

6% above the annual average price, while August and December registered the lowest round trip










airline ticket price (USD 338 or 5% below the annual average price). Airline passengers from

Latin America paid more for round trip airline tickets during December (USD 384), equivalent to

4% above the annual average price (USD 370). September and October recorded the lowest price

from Latin America. Europe registered higher prices in June (USD 971), a 7% increase from the

annual average price. Europe paid the lowest price of the year in March (USD 850), equivalent to

a 6% decrease from the annual average. Figure 3-53 illustrates the seasonality of the relative

change of round trip airline ticket prices traveling to Florida from three world regions from 1995

to 2006.


1 Canada 0 Latin America Europe


1,200
1,000
800
600
400

200
0


Year
Figure 3-52. Florida CSA: average international round trip airline ticket prices from three world
regions between 1995 and 2006. Source: Bureau of Transportation Statistics.











- Canada -*-Latin America Europe


6% -i------i------------ ----i---- i---------- ----------------



S2% ----a -- ------ ----- ----:------:------:-- ------ ; -----:-------
8%
I I I I I I I I I I I



0% I I I I



-8%
I I I I I I I I I I I
I I I I II I I I
40 I I I I I I I I I I I


-2% ------- ------ ---- ----- ----------- ---

4 % ------- ----.^-- --- --- -F ----- r ----- r -- --- ---- --7 -- --.----- T----- T'--



S-* 0% -1 I I i -4% 4% -1%




EuropCa 1% % -6% -5% 0% 7% 6.% 1-% 4% 10% -2% -5%
-Europe -6% I- I- I I I ..... .. __ _.l= 6_ ___ 4. -...2.5 .




Figure 3-53. Florida CSA: monthly seasonal pattern of relative change from the average round
trip airline ticket prices from three world regions between 1995 and 2006. Source:
Bureau of Transportation Statistics.

International Airline Round Trip Airline Ticket Prices by Destination CSA

The top three destination CSAs (South Florida CSA, Orlando CSA, and Tampa-St.

Petersburg CSA) in terms of number of international airline passengers were selected to analyze

international airline round trip airline ticket prices from 1995 to 2006. Round trip airline tickets

from an international origin traveling to the South Florida CSA were the most expensive (USD

518) among all three destination CSAs during the 11-year period. Also, the South Florida CSA

experienced an increase of 5% in airline ticket prices from USD 569 in 1995 to USD 598 in

2006. The Tampa-St. Petersburg CSA reported the cheapest international airline ticket (USD

486). Round trip airline ticket prices to the Orlando CSA reported the largest decrease (5%) from

USD 582 in 1995 to USD 555 in 2006. The Tampa-St. Petersburg CSA ranked second with a 4%

decrease.









An individual analysis of international airline ticket prices for each of the three destination

CSAs is presented next. It includes a discussion of round trip airline ticket prices on international

flights originating from three world regions (Canada, Latin America, and Europe) traveling to a

specific destination CSA.

South Florida CSA international airline ticket prices

Annual average prices of round trip airline tickets for international flights to the South

Florida CSA have increased 5% overall since 1995. But airline passengers traveling from Latin

America paid 14% less for round trip airline tickets in 2006 (USD 341) compared to 1995 (USD

396). Also, round trip airline ticket prices from Canada have experienced a 6% decrease from

USD 372 in 1995 to USD 348 in 2006. In contrast, Europe reported a round trip airline ticket

price increase of 18% during the same time period. Figure 3-54 illustrates average round trip

airline ticket prices to the South Florida CSA from three world regions from 1995 to 2006.


* Canada U Latin America Europe


1,200
1,000
S 800 0
600
3 400
S 200







Year
Figure 3-54. South Florida CSA: average international round trip airline ticket prices from three
world regions between 1995 and 2006. Source: Bureau of Transportation Statistics.










During the period between 1995 and 2006, the round trip airline ticket price from Canada

reported an annual average price of USD 366. February recorded the highest price equivalent of

a 4% increase from the annual average price, while August registered the lowest round trip

airline ticket price (USD 355 or 5% below the annual average price). Airline passengers from

Latin America paid more for round trip airline tickets in March (USD 380), equivalent to 4%

above the annual average price. September recorded the lowest price for tickets from Latin

America. Europe registered higher prices during June (USD 1,012), a 7% increase from the

annual average price. Europe paid the lowest price of the year in March (USD 878). Figure 3-55

shows the seasonality of the relative change of round trip airline ticket prices traveling to the

South Florida CSA from three world regions between 1995 and 2006.


-*-Canada -*-Latin America Europe


10%

8%

6%

s 4%
2 2%

S0%

-2%

-4%
-6%

-8%


---- -------------------------------- --
I I I I I I
I I I I I I


----~-----'*---------?----* ---*-"-T---*
----- ----,---r--

Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec
---Canada 2% 4% 3% 1% -1% 0% -4% -5% 1% 3% 2% -4%
--Latin America 2% 1% 4% 0% -1% 0% -1% -2% -3% -3% -1% 3%
Europe 0% -1% -7% -5% 0% 7% 7% 1% 5% 1% -2% -6%

Figure 3-55. South Florida CSA: monthly seasonal pattern of relative change from the average
round trip airline ticket prices from three world regions between 1995 and 2006.
Source: Bureau of Transportation Statistics.










Orlando CSA international airline ticket prices

Annual average prices of round trip airline tickets for international flights to the Orlando

CSA decreased 5% overall during the period between 1995 and 2006. Airline passengers

traveling from Latin America paid 43% less for round trip airline tickets in 2006 (USD 294)

compared to 1995 (USD 514). Also, round trip airline ticket prices from Canada have

experienced a decrease of 1% during the same time period. In contrast, Europe reported a round

trip airline ticket price increase of 17% from USD 869 to USD 1,015 in 2006. Average round trip

airline ticket prices to the Orlando CSA from three world regions from 1995 to 2006 are

presented in Figure 3-56.


I Canada 1 Latin America Europe




1,200

S1,000 -o i__ -

S800

600

400

200

T) \0 (^ 00 t- OC Z -t- CI o
0 0'






Year
Figure 3-56. Orlando CSA: average international round trip airline ticket prices from three world
regions between 1995 and 2006. Source: Bureau of Transportation Statistics.

Between 1995 and 2006, the round trip airline ticket price from Europe reported an annual

average price of USD 867. June recorded the highest price equivalent to a 7% increase from the

annual average price, while March registered the lowest average round trip airline ticket price









(USD 809 or 7% below the annual average price). Airline passengers from Latin America paid

more for round trip airline tickets in February (USD 491), equivalent to 15% above than the

annual average price. September recorded the lowest average price from Latin America. Canada

registered higher prices during February (USD 365), an 8% increase from the annual average

price of USD 337. Canada paid the lowest price of the year in August (USD 317). Monthly

seasonal patterns of the relative change of round trip airline ticket prices traveling to the Orlando

CSA from three world regions between 1995 and 2006 are shown in Figure 3-57.

SCanada --Latin America Europe

20%

15% --------- ------ ------------ ----- ----- ----- ------ --------.-----


0%


I I I I I I I I I I I
I I I I I I I I I I I
S 10% -- -----4- 4 ------ ----- '-- 4 ----- .--- .. ..---- 1..- --------. .






Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec
Canada -1% 8% 5% 1% 2% 1% -3% -6% 0% -1% -3% -4%
Latin America 2% 15% 7% 1% 4% 4% -1% -7% -11% -10% -1% -2%
Europe 1% 0% -7% -5% 0% 7% 6% 1% 4% 1% -2% -5%

Figure 3-57. Orlando CSA: monthly seasonal pattern of relative change from the average round
trip airline ticket prices from three world regions between 1995 and 2006. Source:
Bureau of Transportation Statistics.
10% ..._ J _- _-- - - --- _--- -----... .. .. S ^ ----------- .. .. "

5% "---- --- --- --- --- --- --- --- --- ---- --- ---. ,... L".. ... .. ..... .J















Tampa-St. Petersburg CSA international airline ticket prices

Annual average prices of round trip airline tickets for international flights to the Tampa-St.

Petersburg CSA decreased 3% overall during the period between 1995 and 2006. Airline

passengers traveling from Latin America paid 8% less for round trip airline tickets in 2005 (USD









344) compared to 1995 (USD 373). Also, round trip airline ticket prices from Canada have

experienced a decrease of 4% between 1995 and 2006. In contrast, Europe reported a round trip

airline ticket price increase of 12% from USD 932 to USD 1,041 in 2006. Average round trip

airline ticket prices to the Tampa-St. Petersburg CSA from three world regions from 1995 to

2006 are shown in Figure 3-58. Note that average round trip prices for Latin America were not

available in 2006.


SCanada U Latin America Europe



1,200
| 1,000 1 _
Q 800
^ 600

400


-" l Cl C C l Cl C l
p 20 200







Year
Figure 3-58. Tampa-St. Petersburg CSA: average international round trip airline ticket prices
from three world regions between 1995 and 2006 Source: Bureau of Transportation
Statistics.

Between 1995 and 2006, round trip airline tickets from Europe reported an annual average

price of USD 926. June recorded the highest price equivalent to a 7% increase from the annual

average price, while March registered the lowest average round trip airline ticket price (USD 864

or 7% below the annual average price). Airline passengers from Latin America paid more for

round trip airline tickets in July (USD 514), equivalent to 54% above the annual average price

(USD 333). September recorded the lowest average price from Latin America. Canada registered










higher prices during April (USD 352), a 7% increase from the annual average price of USD 329.

Canada paid the lowest price of the year in July (USD 301). Monthly seasonal patterns of the

relative change of round trip airline ticket prices traveling to the Tampa-St. Petersburg CSA from

three world regions between 1995 and 2006 is presented in Figure 3-59. Seasonal values for

Latin America were not available for January and February.


Canada -*-Latin America Europe

a 60%
00% ----I I-- ---- I -------------

4 0% -.------ r------ ------ ------ ------ ----.--A--r ----- r ----- r ----- r----- r------
I I I I I I I I I I I


3 40% --------L----------------- -----J ----- ---- ---- L --------- L ------L-----

S 30% -------- ----- ------------- -- t----- t------------. ----


5200---- ------------L-----------------L-L---






-20%
Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec
--Canada 0% 6% 4% 7% 6% 1% -8% 5% 0% -3% -5% -3%
-6-Latin America 0 0 -11% -2% -9% 0% 54% -14% -15% -8% 15% -10%
Europe 0% 1% -7% -6% 0% 7% 6% 1% 4% 1% -2% -6%
Figure 3-59. Tampa-St. Petersburg CSA: monthly seasonal pattern of the relative change from






the average round trip airline ticket prices from three world regions between 1995 and
2006. Source: Bureau of Transportation Statistics.

Freight and Mail Transported via Commercial Passenger Airlines to Florida

Domestic freight transportation2 to the Florida CSA accounted for 3% of the total freight
I I I I I I I I
0 I I II I I I I I I
I I I I I I I I I

























traffic transported via commercial passenger airlines in the United States. Domestic freight to the

Florida CSA via commercial passenger airlines has increased nearly 486% from 178.5 million
I I I I I I I I I I
I i I I i I I I I I i





























2 Includes freight transported by commercial passenger airlines only. Freight transported by cargo airlines (e.g.,
FEDEXg) is not included.
FEDEXR) is not included.









pounds in 1990 to 1.045 billion pounds in 2006. Approximately 63% of the more than 6.44

billion pounds of freight were transported from the South region to the Florida CSA during this

time period. The Midwest region ranked second with a 15% share, followed by the Northeast

(13%) and West (9%) regions.

The South and Midwest regions experienced the largest increases in total domestic freight

transported to the Florida CSA (782% and 647% respectively). Meanwhile, the Northeast region

almost doubled its freight sent to the Florida CSA, while the West region experienced a 60%

increase during the same time period. Figure 3-60 shows total domestic freight transported to the

Florida CSA between 1990 and 2006.

Similar to domestic airline passengers, domestic freight from three of the four U.S. regions

was transported more frequently during March than any other month between 1990 and 2006.

The Northeast, Midwest, and South regions transported approximately 10% of total freight

during that month. The West region recorded its highest level during December with 10% of its

total annual freight transported to the Florida CSA.

July registered the lowest level of domestic freight transported from the Midwest,

Northeast, and West regions, while December registered the lowest level for freight originated

from the South region. Figure 3-61 presents the seasonal pattern of total freight transported by

commercial passenger airlines to the Florida CSA per month, by U.S. region from 1990 to 2006.











1 Other West 0 Northeast Midwest U South


800,000
700,000
600,000
o 500,000
H 400,000
300,000
S200,000
L- 100,000
0




Year

Figure 3-60. Florida CSA: total domestic freight transported from four U.S. regions using
commercial passenger airlines between 1990 and 2006. Source: Bureau of
Transportation Statistics.

x West Midwest -*-Northeast South
10%

I .I _/I
1 I I I I I I I I I I I



% ---------------------L-------------L--

I I I I ----------- ----- ------






Midwest 8% 8% 10% 9% 9% 7% 7% 8% 8% 9% 9% 9





Northeast 9 9 9 9 9 7 7-------- 8------------- 7----- 9 8--- 9-----------------------------
w 7% -------- L .. -------- --. -. ------------ .. ----- ------------ .... ------------ L





Midwest 8% 8% 10% 9% 9% 7% 7% 8% 8% 9% 9% 9%


---South 8% 8% 10% 9% 9% 8% 8% 8% 8% 8% 8% 7%

Figure 3-61. Florida CSA: monthly seasonal pattern of total freight transported from four U.S.
regions using commercial passenger airlines between 1990 and 2006. Source: Bureau
of Transportation Statistics.










The South Florida CSA received nearly 4.0 billion pounds of domestic freight transported

to the Florida CSA equivalent to 58% of the total share. The Orlando CSA ranked second with

23%, followed by the Tampa-St. Petersburg CSA with 10% of the total domestic freight

transported to the Florida CSA. The top five destination CSAs experienced large increases in

domestic freight traffic of which Jacksonville recorded the largest increase (814%) during the

17-year period. Total domestic freight transported to each of the six destination CSAs in Florida

is presented in Figure 3-62.

Other Fort Myers U Jacksonville
Tampa-St. Petersburg U Orlando U South Florida


600,000

500,000

400,000

300,000

200,000

100,000


0z Cl Cm ~ T 0 3 Z m-i Ci m~ -
~ z z z 00 00 000
~ z z z 00 00 000
--------------------------------(l1 (l1C(C 1 ( C


Year

Figure 3-62. Total domestic freight transported to six destination CSAs in Florida using
commercial passenger airlines between 1990 and 2006. Source: Bureau of
Transportation Statistics.

Unlike domestic freight transportation, domestic mail transportation3 to the Florida CSA

via commercial passenger airlines has decreased approximately 58% during the period between

1990 and 2006. Approximately 42% of the more than 2.45 billion pounds of mail sent to the


3 Includes mail transported by commercial passenger airlines only. Mail transported by cargo airlines (e.g.,
FEDEX) is not included.


r r r










Florida CSA originated from the South region during this time period. The Northeast region

ranked second with a 25% share, followed by the Midwest (22%) and West (10%) regions.

The South and Midwest regions experienced the largest decreases in total domestic mail

transported to the Florida CSA (74% and 66% respectively). Mail sent from the Northeast region

decreased 43% from 43.4 million pounds in 1990 to 24.6 million pounds in 2006, while the West

region experienced a 14% decrease during the same time period. Figure 3-63 presents total

domestic mail transported via commercial passenger airlines to the Florida CSA between 1990

and 2006.


1 Other West Midwest U Northeast U South


QI

100,000

o 80,000

S 60,000

40,000

20,000
0O ^---------------------__----------------------------------------











Year
Figure 3-63. Florida CSA: total domestic mail transported from four U.S. regions using
commercial passenger airlines between 1990 and 2006. Source: Bureau of
Transportation Statistics.

Domestic mail via commercial passenger airlines from the four major U.S. regions was

transported more frequently during December than any other month between 1990 and 2006. The

Northeast, Midwest, and South regions transported approximately 11% of total mail during that

month. The West region recorded its highest level during December with 11.5% of its total











annual mail transported to the Florida CSA. September registered the lowest level of domestic

mail transported from four major U.S. regions. Figure 3-64 illustrates the monthly seasonal

pattern of domestic mail transported to the Florida CSA from 1990 to 2006.


12%


11%


10%


9%


8%


7%


6%


x West Midwest --Northeast --South


x
I I I I I I I I I I I
I I I I I I I I I I I
I I I I I I I I I I I
1 ----- ------- L ----- 1 ----- i ------L----- ----- T/----
I I I I I I I I I I I
I I I I I I I I I I I
I I I I I I I I I I I

I I I I I I I I I I I
--------1------1--------


Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec
x West 9% 8% 9% 8% 8% 8% 7% 8% 7% 8% 8% 12%
Midwest 10% 9% 9% 9% 8% 7% 7% 7% 7% 8% 8% 10%
--Northeast 10% 9% 10% 9% 8% 7% 7% 7% 7% 8% 8% 11%
South 10% 8% 9% 8% 8% 8% 8% 8% 7% 8% 8% 11%

Figure 3-64. Florida CSA: monthly seasonal pattern of total mail transported from four U.S.
regions using commercial passenger airlines between 1990 and 2006. Source: Bureau
of Transportation Statistics.

The South Florida CSA received nearly 1.11 billion pounds of domestic mail which

represented 45% of total mail transported to the Florida CSA. The Tampa-St. Petersburg CSA

ranked second with 22%, followed by the Orlando CSA (18%), Jacksonville CSA (13%), and

Fort Myers CSA (1%). Four of the top five destination CSAs experienced decreases in domestic

mail traffic of which Jacksonville recorded the largest drop (89%) during the 17-year period. In

contrast, mail transported to the Fort Myers CSA increased 15% during the same time period.

Total domestic mail transported to each of the six destination CSAs in Florida is shown in Figure

3-65.









r. Other
Orlando


Fort Myers U Jacksonville
I Tampa-St. Petersburg U South Florida


100,000

80,000

60,000

40,000

20,000


I -I


0 !




Year
Figure 3-65. Total domestic mail transported to six destination CSAs in Florida using
commercial passenger airlines between 1990 and 2006. Source: Bureau of
Transportation Statistics.

International freight transportation to the Florida CSA using commercial passenger airlines

has increased approximately 206% during the period between 1990 and 2006. Nearly 88% of the

more than 23.72 billion pounds of freight were transported from Latin America to the Florida

CSA during this time period. Europe ranked second with a 10% share. Canada's share was less

than half a percent. Europe experienced an increase in total freight transported to the Florida

CSA from 52.88 million pounds in 1990 to 164.95 million pounds in 2006. Latin America also

grew 196% during the 17-year period, as presented in Figure 3-66. International freight from

Canada and Latin America was transported more frequently during February than any other

month between the year 1990 and 2006. Europe recorded its highest levels during August when

it transported 13% of its total annual freight to the Florida CSA. Europe registered the lowest

level of international freight transported (7.2%) in January, as shown in Figure 3-67.















S Canada U Other Europe U Latin America


2000


1500


1000


500


0


iT 7 r r r H i
z l c~~ .~ \C t~- OC ~ Z Cl C
I I z z z z z zI
03 z z z z z z~
Cl Cl Cl Cl Cl Cl C


Year

Figure 3-66. Florida CSA: total international freight transported from three world regions using
commercial passenger airlines between 1990 and 2006. Source: Bureau of
Transportation Statistics.


14%
13%
12%
11%
10%
9%
8%
7%
6%
5%
4%


*Canada Europe Latin America

I------ I ----- I I I I I I I------ I I------

-- --- ----------------- -------- ------ ------ ------


II
- - - -- -- .- --- -- -- - 4.- .- I - -
I I I I I I I I I I I
I I I I I I I I I I I

I-I I I -I -- =- -


-- ---- --_------------ ------ ---------- --------
-- ----------------- -- ---- ---------- ---- --- -- -----------
----- --------L- ----- -- ---------- ----- --- -----



i"--- L----- TL-- l-f- l- i I' T 'I f -I- --- I i -I'--


Mar


May


June


Sept


Canada 9% 11% 11% 9% 8% 7% 6% 6% 5% 8% 11% 11%
--Europe 7% 8% 8% 7% 8% 8% 8% 13% 8% 9% 9% 8%

Latin America 9% 9% 9% 9% 8% 7% 8% 8% 8% 9% 8% 9%

Figure 3-67. Florida CSA: monthly seasonal pattern of total freight transported from three world
regions using commercial passenger airlines between 1990 and 2006. Source: Bureau
of Transportation Statistics.


Between 1990 and 2006, the South Florida CSA received approximately 23.0 billion


pounds of international freight which represents 97% of total international freight transported to










the Florida CSA. The Orlando CSA ranked second with less than 3%. The other three destination

CSAs accounted for less than a quarter percent during the same time period. The South Florida

CSA reported an increase of 203% in international freight traffic. Total international freight

transported to each of the six destination CSAs in Florida is shown in Figure 3-68.

r" Other Fort Myers U Jacksonville
Orlando U Tampa-St. Petersburg U South Florida


25 2,000

E 1,500


o 1,000
a,


I IuIu1urn1u1n1r1u1u


0
00C t- 0 Z C


Year


Figure 3-68. Total international freight transported to six destination CSAs in Florida using
commercial passenger airlines between 1990 and 2006. Source: Bureau of
Transportation Statistics.

Unlike international freight transportation, international mail transported to the Florida

CSA using commercial passenger airlines has decreased approximately 65% during the period

between 1990 and 2006. In 1990 more than 16 million pounds of mail were transported to the

Florida CSA but by 2006 the total had decreased to only 5.53 million pounds. Nearly 54% of the

more than 109 million pounds of mail was transported from Latin America to the Florida CSA

during this time period. Europe ranked second with a 35% share, followed by Canada with a

share of 11%. Both Latin America and Europe experienced a decrease of approximately 75% in

total mail transported to the Florida CSA. In contrast, Canada has reported large gains in mail


|


On


r
a
oa 5(
o
L









transported to the Florida CSA from 0.1 million pounds in 1990 to more than 1.4 million pounds

in 2006. Total international mail transported to the Florida CSA from four world regions

between 1990 and 2006 is illustrated in Figure 3-69.


SCanada 0 Other Europe E Latin America


: 14.0
12.0
a 10.0
8 80
6.0 -




o --- --- ---------- -- -'-- '-- C r C C



Year


Figure 3-69. Florida CSA: total international mail transported from four world regions using
commercial passenger airlines between 1990 and 2006. Source: Bureau of
Transportation Statistics.

International mail from Canada and Europe was transported more frequently during

December than any other month between the year 1990 and 2006. Latin America recorded its

highest levels during the month of May when it transported approximately 12% of total annual

mail to the Florida CSA. Europe registered the lowest level of international mail transported

(7.2%) in September. Canada recorded its lowest levels during July (6.3%), while Latin America

did so in October (6.2%). Figure 3-70 illustrates the monthly seasonal pattern of domestic mail

transported to the Florida CSA from 1990 to 2006.











-1 -Canada --Europe Latin America


14%
13% -i----i---------- -------------i--------------i--------i
13% ------------- ---- ---- ----------------------------------------- p


11%_l_/ ------- ^- - ------- ------ ------ ------ ------ -----.- ----




8% ---- --- -- -- 4-


% ------- ----- ----------- ---- ---------------- ------------------------

6H % - - ------- ------ ------- - -. - - .
12% .0--------.-. .-- -------------..---- -.- -
I I I I I I I I I


% ------------------- ------ ------------------------------ --- -----










4%
Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec
I I I I I I I I I
o I I I I I I I I I I







-Canada 10.7% 10.7% 8.9% 7.5% 6.8% 6.4% 6.3% 6.5% 6.3% 7.0% 10.0% 13.0%
I I I I I I I I I
0) 90---L-- --I I---------- I-- ------~--- -----
I I I I I I I I I



%----r--r--r----------,-----------.--------- --------r----
I I I I I I I





Canada 10.7t ,10.70o8.90 750 6.8-, 6.4' 6.30 6.50 6.3, 7. 10.O -13.O0o
-*-Europe 7.9% 7.4% 8.5% 7.9% 8.5% 8.5% 8.3% 7.3% 7.3% 7.6% 9.0% 12.0%
Latin America 8.9% 8.9% 9.1% 8.8% 12.2% 9.5% 8.4% 6.8% 6.4% 6.2% 6.6% 8.4%

Figure 3-70. Florida CSA: monthly seasonal pattern of total mail transported from three world
regions using commercial passenger airlines between 1990 and 2006. Source: Bureau
of Transportation Statistics.

Between 1990 and 2006, the South Florida CSA received approximately 106.2 million


pounds of international mail which represents 97% of total international mail transported to the


Florida CSA. The Orlando CSA ranked second with 2%. The other three destination CSAs


accounted for less than a percent during the same time period. All destination CSAs reported


decreases in total international mail transported since 1990. The South Florida CSA reported a


decrease of 64% from 15.7 million pounds in 1990 to 5.6 million pounds in 2006. Total


international mail transported to each of the six destination CSAs in Florida is presented in


Figure 3-71.










E Other Fort Myers Jacksonville
Orlando U Tampa-St. Petersburg U South Florida


100,000

80,000

60,000

E 40,000

20,000







Year

Figure 3-71. Total international mail transported to six destination CSAs in Florida using
commercial passenger airlines between 1990 and 2006. Source: Bureau of
Transportation Statistics.

Economic, Social, and Weather Indicators

Additionally, the following statistics were included in this study: annual gross domestic

product, per capital personal disposable income, and population collected from every state by the

U.S. Bureau of Economic Analysis; brand advertising expenditures from selected private firms

and generic advertising expenditures from Florida government entities collected by TNS Media

Intelligence; monthly statistics on average monthly prices of kerosene-type jet fuel collected by

the U.S. Department of Energy; and annual crime statistics collected by Florida's Department of

Law Enforcement.

Gross Domestic Product, Personal Disposable Income, and Population

Annual gross domestic product (GDP), per capital personal disposable income, and

population estimates for each state and the District of Columbia were retrieved from the online









database provided by Bureau of Economic Analysis through its Regional Economic Analysis

Division. Retrieved data covered the period between 1990 and 2006. States were then aggregated

according to the U.S. Census geographic regions scheme as shown previously in Table 3-1.

Annual GDP, PDI-PC and population totals calculated for each U.S. region are discussed next.

In 2006, GDP for the four U.S. regions was reported at USD 12.58 trillion, which

represents a 126% increase from 1990 when GDP totaled USD 5.57 trillion. The Northeast

region accounted for the largest share among the four U.S. regions, with nearly 30% of total

GDP. The Midwest region followed with 26.68%. The South (21.90%) and West (21.88%)

regions ranked third and fourth, respectively. Among all U.S. regions, the South region

experienced the largest increase (150%) in GDP during the 17-year period. The West region

ranked second with a 142% increase, followed by the Midwest (116%) and Northeast (107%)

regions. Total GDP by U.S. region from 1990 to 2006 is presented in Figure 3-72.

During the period between 1990 and 2006, per capital personal disposable income (PC-

PDI) averaged USD 23,195 for the four U.S. regions. Also, the PDI-PC, reported at USD 31,568

in 2006, increased 90% during the same time period. Among the four U.S. regions, the Northeast

region is the only region with a PDI-PC above the national average. Its PDI-PC amounted to

USD 25,746, equivalent to 11% above the national average. In regards to PDI-PC growth, the

South region experienced the largest increase (95%) during the 17-year period. The Midwest

region ranked second with a 91% increase in its PDI-PC, followed by the West (88%) and

Northeast (86%) regions. Average PDI-PC for each U.S. region from 1990 to 2006 is presented

in Figure 3-73.

Total population for four U.S. regions was estimated at 282.29 million in 2006, which

represents a 17% increase since 1990 when the population was estimated at 240.94 million. The










Midwest region accounted for the largest population share among the four U.S. regions, with

nearly 28%. The Northeast region followed with 27%. The South (23%) and West (22%) regions

ranked third and fourth, respectively. Among all U.S. regions, the West region experienced the

largest increase (31%) in population from approximately 49.03 million in 1990 to 64.02 million

in 2006. The South region ranked second with a 25% increase, followed by the Midwest (11%)

and Northeast (8%) regions. Figure 3-74 presents population estimates from each U.S. region

from 1990 to 2006.


1 South West Midwest U Northeast


4,000
3,500
3,000
2,500
2,000
1,500
1,000
500
0


Year

Figure 3-72. Annual gross domestic product from four U.S. regions between 1990 and 2006.
Source: Bureau of Economic Analysis-Regional Economic Analysis Division.











1 South West Midwest U Northeast


35,000
30,000
S25,000
20,000
15,000
10,000
5,000
0





Year

Figure 3-73. Annual per capital personal disposable income from four U.S. regions between 1990
and 2006. Source: U.S. Department of Commerce-Bureau of Economic Analysis.


West South U Northeast U Midwest


80
. 70
S60
S 50
40
30
1 20
20
a 10
0


Year

Figure 3-74. Annual population estimates from four U.S. regions between 1990 and 2006.
Source: U.S. Department of Commerce-Bureau of Economic Analysis.










Brand and Generic Advertising Expenditures

Description, selection, and aggregation of advertising expenditures data

Data used to analyze brand and generic advertising expenditures were collected by (Taylor

Nelson Sofres) TNS Media Intelligence and are available online through subscription. Brand

advertising expenditure data used in this research include the period between 1995 and 2006,

while generic advertising expenditure data cover the period between 2002 and 2006.

Tourism promotion accounts of four major tourist companies were selected to analyze

brand advertising efforts conducted to attract visitors to Florida. Busch Entertainment

Corporations, Walt Disney Company, Carnival Cruise Lines, and NBC-Universalc have

tourist attractions in Florida, as well as in other locations in the United States and abroad. These

companies have several accounts related to the tourism promotion efforts conducted in the

United States.

For the purpose of this research, these accounts were divided into three major categories:

Florida, Combined, and Non-Florida. The "Florida" category denotes all accounts specifically

stating that their budget was committed to advertise its Florida-based attractions only. For

example, the "Florida" category of Walt Disney Company" includes advertising expenditures

related to attractions in Florida such as Walt Disney World, Disney Cruises, and Disney

Hotels. The "Combined" category includes accounts stating their budget was used to advertise

attractions in Florida and elsewhere. For example, the "Combined" category of Walt Disney

Company includes those accounts related to general advertising expenditures for attractions in

Florida and California. Finally, all other accounts that advertise attractions outside the state of

Florida were included in the "Non-Florida" category. For example, the "Non-Florida" category









of Walt Disney Companyc includes only those accounts related to advertising expenditures for

attractions located outside of Florida, such as Disneyland in California.

As for generic advertising expenditures, tourism promotion accounts from Florida cities,

Florida counties, and the state of Florida were selected from the year 2002 to 2006. The "City"

category includes all tourism promotion efforts conducted by cities (e.g., Orlando, Miami,

Tampa) in Florida to promote their respective city. The "County" category includes all tourism

promotion efforts conducted by counties (e.g., Dade, Hillsborough, Alachua) in Florida to

promote their respective county. Finally, the "State" category includes all accounts in which

promotional efforts were conducted to promote the state of Florida as a whole.

Brand advertising expenditures

More than USD 2.63 billion were spent by four major tourist companies with businesses in

Florida during the period between 1995 and 2006. On average, these four companies spent

approximately USD 220 million per year during the 12-year period. Tourism promotion efforts

to advertise their attractions in Florida accounted for 44% and the combined effort accounted for

53% of total advertising expenditures during the 12-year period.

Walt Disney Companyc spent more dollars (USD 1.29 billion) in advertising than the other

three companies, equivalent to 49% of total brand advertising expenditures. Carnival Cruise

LinesC ranked second (25%), followed by Busch Entertainment Groupc (18%), and NBC-

Universalc (9%). Walt Disney Companyc spent nearly USD 994.94 million to promote

attractions in Florida, which represented approximately 77% of total advertising expenditures

during the 12-year period. NBC-Universalc was a distant second spending USD 128.7 million in

efforts to promote attractions in Florida. Still, it represented more than one half of its advertising

expenditure budget. Note that some companies advertise their Florida and non-Florida attractions

in the same advertising spot. For example, Busch Entertainment GroupC usually combined the










three water parks in the same advertising spot. In fact, 91% of the advertising expenditures of

Busch Entertainment Groupc fall under the "Combined" category. Note that all accounts related

to tourism promotions from Carnival Cruise Linesc were all categorized as "Combined". Total

brand advertising expenditures spent by these four companies are presented in Figure 3-75.


a Florida '- Combined a Non-Florida


1,200

1,000

0 800

600






Bush
Buh NBC-Universal
Entertainment
Inc.
Group


Walt Disney Carnival Cruise
Company Lines


H Florida 29,415,200 128,684,900 994,937,900
u Combined 422,623,600 88,127,000 234,317,100 649,163,700
a Non-Florida 13,802,800 9,746,100 68,341,500

Figure 3-75. Total brand advertising expenditures from four tourism-related companies in
Florida between 1995 and 2006. Source: TNS Media Intelligence.

During the period between 1995 and 2006 Busch Entertainment GroupC, NBC-Universalc,

and Walt Disney Company" made most of their advertising efforts to promote their attractions in

Florida during the first five months of the calendar year. NBC-Universalc spent two-thirds of its

total advertising budget to promote Florida attractions between January and May, while Busch

Entertainment Group and Walt Disney Company spent approximately 62% of their budget

during those five months. Advertising expenditures decreased later in the calendar year. For

example, Busch Entertainment GroupC allotted 1% of its advertising budget to November, while

NBC-Universalc assigned 3%. July, August, and September registered the lowest levels of










advertising expenditures by Walt Disney CompanyC. Figure 3-76 illustrates monthly seasonal

patterns of total brand adverting expenditures of Busch Entertainment GroupC, NBC Universalc,

and Walt Disney Companyc from 1995 to 2006.


S-0-Busch E. Group NBC-Universal -- Walt Disney Co.
18% .
18%
I I I I I I I I I I I
I I I I I I I I I I I


S1 ------- ---- ----- ------------ --



-- u 16% 14-o-5-- 1 -----7 --------- --------------- -----1 2
0. 12 % ------------- ------ L--- ----- 0^ -1 ----- ----- -J --Q---J -----.- I ----
% -- -



6_ % --------- ------------------- --- --- -------------------.,-----
S % ------- ------------------ -------------- --- -


Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec
-0- Busch E. Group 15% 14% 15% 12% 7% 8% 12% 6% 5% 4% 1% 2%
NBC-Universal 9% 15% 14% 16% 14% 6% 5% 5% 5% 5% 3% 4%
-0- Walt Disney Co. 14% 13% 12% 11% 11% 4% 3% 3% 3% 8% 9% 8%

Figure 3-76. Florida attractions: monthly seasonal pattern of total brand adverting expenditures
from three tourism-related companies between 1995 and 2006. Source: TNS Media
Intelligence.

During the period between 1995 and 2006 Busch Entertainment GroupC, NBC-UniversalC,

and Carnival Cruise Lines made most of their advertising efforts to promote their combined

attractions in Florida and elsewhere during the first six months of the calendar year. Carnival

Cruise Lines spent 65% of its advertising budget to promote cruise lines between January and

June and advertised most heavily in January (16%) and February (12%).

NBC-UniversalC and Busch Entertainment Group showed similar patterns in terms of

advertising expenditures throughout the calendar year between 1995 and 2006. NBC-UniversalC

spent 65% of its total advertising budget to jointly promote Florida and Non-Florida attractions

between January and June. It spent more heavily in April (19%) and March (16%). Busch










Entertainment Group spent more advertising dollars in June (17%) than any other month,

followed by May (15%) and April (13%). In total, it spent almost 70% of its budget during the

first semester. Advertising expenditures decreased later in the calendar year for both companies.

Busch Entertainment GroupC allotted 3% of its advertising budget to the each of the last four

months of the year, while NBC-UniversalC assigned no more than 7%.

Despite spending more than 55% of its budget during the first semester, Walt Disney

CompanyC also spent a large amount of dollars to jointly advertise attractions in Florida and

California between September and November (33%). Walt Disney CompanyC also spent heavily

during January (17% of its advertising budget), more than the NBC-UniversalC and Busch

Entertainment GroupC combined. Advertising expenditures were reduced during December

(3%), July (4%), and June (5%) as shown in Figure 3-77.

-0- Busch E. Group -A-NBC-UniversalO
-0-Walt Disney Co. -0Carnival CruisesC
20% I
18%
18% -------------- ---- ------------------------------- ----------- ------ ----


S14% - \^- --- -- - - - t_- - - - - - -


s 00%/ ------- i------ ^-----s ~- --- --- ------ ^-------- f--T. ------ s---.------


6% ----- ---- -- ----------- --
0 1% -------------L ----- -------- ---------- ------- -------- -----
Source: TNS Media Intelligence.

2% +-----;--------------- ---------;-----;----- -;------ r------ ------ ----- -----
0%
Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec

Month
Figure 3-77. Florida and non-Florida attractions: monthly seasonal pattern of total brand
adverting expenditures from three tourism-related companies between 1995 and 2006.
Source: TNS Media Intelligence.









Generic advertising expenditures

Cities, counties, and the state of Florida spent more than USD 339 million in promotional

efforts during the period between 2002 and 2006. On average these three entities combined spent

approximately USD 67.84 million per year during the 5-year period. Tourism promotion efforts

conducted by the state to advertise state attractions accounted for 63% of total generic

advertising expenditures. City governments ranked second with 15%, followed by counties with

15% of total generic advertising expenditures during the 5-year period.

Since 2002 generic advertising expenditures from the state increased 134%, while the

cities and counties experienced a decrease in expenditures. Cities' expenditures in generic

advertising decreased 23% and advertising expenditures from counties decreased 6% during the

5-year period. Total generic advertising expenditures by government entities in Florida from

2002 to 2006 are presented in Figure 3-78.

County City State



60.0
S50.0
0 40.0
S30.0
20.0
lo.o


Cl Cl l rCl Cl

Year
Figure 3-78. Total generic advertising expenditures spent by government entities from Florida
between 2002 and 2006. Source: TNS Media Intelligence.

Advertising efforts conducted by city, county, and state government in Florida during the

first six months of the calendar year accounted for 67% of total advertising expenditures. City

governments throughout Florida spent most heavily in May, 18% of their advertising budget, to














promote their city attractions during May. April ranked second with 12%, followed by March



(11%). Advertising expenditures were reduced during December (3%) and September (4%).



County governments in Florida spent more heavily in April (22%), March (19%), and May



(16%) to advertise their county attractions. In total, they spent almost 70% of their budget during



the first semester. Advertising expenditures decreased later in the calendar year, when county



governments assigned less than 4% of their advertising budget to each of the last three months of



the year. Finally, the state government spent more heavily during May (14% of its advertising



budget). The state government assigned similar amounts of advertising dollars (approximately



6% to each month) during the second semester of the calendar year. Figure 3-79 presents



monthly seasonal patterns of total generic advertising expenditures for each of the three



government entities between 2002 and 2006.


I County -h- City State


I I I I
I I I I
I I I I
I I I I
I I I I

I I I I
I I I I
I I II
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I
I .. I
...... *"I
_ I I I
/ I I I
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I


Mar


Apr


I I I I I I I
I I I I I I I
I I I I I I I
I I I I I I I
I I I I I I I
I I I I I I I
- ------------ -----I- 4 ----- -.
I I I I I I I
I I I I I I I
I I I I I I I
I I I I I I I
.I I I I I I I
I I I I I I I

.-----J----L---- 1----- J------- L- .
I I I I I I I
I I I I I I
I I I I I I
I I I I I I I
I I I I I I
I I I I I I I

I I I I I I I
I I I I I I
I I I I I I
I I I I I
I I I I
I I I
I 'J. I 1""
I I ',. I I, "" I % I

I I I I II I


May


June


County 3% 3% 19% 22% 16% 7% 6% 6% 10% 4% 3% 2%

r- City 8% 10% 11% 12% 18% 10% 7% 5% 4% 6% 6% 3%

State 9% 8% 7% 10% 14% 12% 6% 6% 6% 6% 8% 6%


Figure 3-79. Monthly seasonal pattern of total generic adverting expenditures spent by city,

county, and state government to promote Florida between 2002 and 2006. Source:

TNS Media Intelligence.


25%


20%o




15%




10%




5%


VI
B

-cs

pa
*a o.





'Sr


uc "
*_a
&>

o










Foreign Exchange Rate: Euro to U.S. Dollar

Foreign exchange rate data were retrieved from the Federal Reserve through its Statistical

Release published on its Web site. Monthly rates from January 1990 to December 2006 were

obtained for the Euro, official currency of the European Union. This foreign exchange rate, Euro

to U.S. dollar, was used to describe economic conditions in Europe and its relationship to airline

passenger travel to Florida. An exchange rate level over 1.00 denotes a weak U.S. dollar and

airline tickets to Florida are more affordable to Europeans. The U.S. dollar has weakened 2%

compare to the Euro () since 1990. The U.S. dollar performed better in the period between 1997

and 2003 when it was stronger than the Euro, but it has grown weaker since 2004. Figure 3-80

presents annual air passenger traffic from Europe to Florida and the annual exchange rate

average (Euro to USD) from 1990 to 2007. Note that an exchange rate level over 1.00 denotes a

weak U.S. dollar.


3,000 Europe Exchange rate 1.50
1.40
8 2,500 1.30

0- 1.20
2,000 -
S1.10
1,500 1.00
S0.90
o 1,000 080
.0.80
e .0.70
E 500
0.60
0 I I 0.50
0 -r- --l -r -r -r- -I- -r 0.50



Year
Figure 3-80. Annual air passenger traffic traveling from Europe to Florida and annual exchange
rate (Euro to USD) between 1990 and 2006. Source: Bureau of Transportation
Statistics and Federal Reserve.











Historic Jet Fuel Prices

Data on monthly kerosene-type jet fuel prices for three different stations, New York


Harbor, Gulf Coast, and Los Angeles in the United States; and two international stations,


Rotterdam and Singapore, were retrieved from U.S. Department of Energy-Energy Information


Administration and are available online. Retrieved data covered the period between 1990 and


2006. Kerosene-type jet fuel prices have increased 106% during the period between 1990 and


2006. Over the 17-year period jet fuel prices have followed a similarly increasing pattern in the


three U.S. stations and the two international stations. Kerosene-type jet fuel prices in the New


York Harbor and the Gulf Coast stations have increased 114%, while prices in Los Angeles


recorded a 93% increase. Similarly, kerosene-type jet fuel prices in the two international stations


have increased. Figure 3-81 shows the annual average price of kerosene-type jet fuel in five


stations worldwide from 1990 to 2006.


-s-New York Harbor --U.S. Gulf Coast -A-Los Angeles
--Rotterdam S inga pore


S200.00 ---i-- ------- ---------------- -i-i- I-- -i- -
250.00 -....


1 200.00 ----- ---- ---------------------------- --- -------------- ---





S 150.00 --- --- -- -- - ---- -- .- --- ---------.---.w- -.-
I I I I I I I I I I I I I I
I I I I I I I I I I I I I I I I
I I I I I I I I I I I I I I I I
I I I I I I I I I II I I I I I I
SI I I I I I I I I I I I I I I
I I I I I I I I I I I I I I I


0.00 ---'------------- -'---'-7-'- F. .
I I I I I I I I I I I I I I I
I I I I I I I I I I I I I I I I
I I I I I I I I I I I I I I I I


I I I I I I I I I I I I I I



0.00 -- -
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
-*-New York Harbor 91 64 59 56 53 52 66 59 43 52 91 74 71 87 119 171 195
U.S. Gulf Coast 90 61 57 53 49 49 61 56 40 50 85 72 69 83 115 171 192
SLos Angeles 106 63 60 60 54 57 66 63 45 58 94 77 73 89 128 174 203
Rotterdam 97 67 59 55 50 51 64 58 41 52 88 74 70 86 120 169 194



Figure 3-81. Historic average kerosene-type jet fuel prices from five worldwide locations
between 1990 and 2006. Source: Department of Energy.
between 1990 and 2006. Source: Department of Energy.










Between 1990 and 2006, kerosene-type jet fuel prices reported an annual average price of

USD 0.83 per gallon. October recorded the highest price equivalent to an 11% increase from the

annual average price for all five locations, while January registered the lowest jet fuel price

(USD 0.83 or 7% below the annual average price). Monthly seasonal patterns of the relative

change of kerosene-type jet fuel prices from five locations worldwide between 1990 and 2006

are presented in Figure 3-82.

-*-New York Harbor ---U.S. Gulf Coast --Los Angeles
---Rotterdam Singapore
15%





0%


-5% L -- I I- ---------- --- -- --- ---- -----


-10% r r -- --
Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec
--New York Harbor -6% -6% -6% -4% -3% -6% -1% 3% 8% 10% 6% 3%
-U.S. Gulf Coast -7% -8% -7% 4% -3% -5% 0% 4% 9% 13% 6% 2%
----Los Angeles -8% -7% -6% -2% -3% -3% -1% 5% 7% 9% 7% 2%
--Rotterdam -8% -7% -6% -4% -4% -5% 0% 3% 8% 11% 8% 4%
Singapore -6% -6% -7% -4% -4% -6% -1% 3% 7% 11% 8% 5%

Figure 3-82. Monthly seasonal pattern of historic average kerosene-type jet fuel prices from five
worldwide locations between 1990 and 2006. Source: Department of Energy.

Hurricanes and Wildfires Affecting Florida

Hurricane data were retrieved from the National Oceanic and Atmospheric Administration

(NOAA) and the National Weather Service-Archive of Hurricane Seasons which are available

online. Only tropical storms that affected Florida were selected. Tropical storms were

categorized from 1 to 7 to denote strength in wind speed. Categories 1-5 were assigned to each









storm following the Saffir-Simpson Hurricane Scale, category 6 for a tropical storm, and

category 7 for a tropical depression.

Forty-nine storms affected Florida during the period between 1990 and 2006, which

represents an annual average of less than three storms during the 17-year period. More than one

half of the storms (57%) were categorized as tropical storms and less than 30% of the storms

have been categorized as hurricanes. The year 2005 was the busiest in terms of number of storms

affecting Florida. Five tropical storms, two category 3 hurricanes, and one category 1 hurricane

affected the peninsula this year. Figure 3-83 presents total number of storms that affected Florida

from 1990 to 2006.

Cat 1 Cat 2 Cat 3 Cat 4 Cat 5 Trop Dep Trop Storm






5

4

e 3


0 2




Year
Figure 3-83. Number of tropical storms (by category) affecting Florida between 1990 and 2006.
Source: National Oceanic and Atmospheric Administration.

Data on the number of wildfires including total of acres burned across Florida were used in

this study as reported by the Florida Department of Agriculture and Consumer Services, Division

of Forestry. Daily data retrieved from January 1990 to December 2006, were then aggregated

into monthly data. Also, wildfires were grouped into a specific category according to the size of









the wildfire (number of acres burned) and is described as follows: category 1 includes all

wildfires that bum between 0.10 and 0.29 acres; category 2 represents wildfires where 0.30 to

9.99 acres were burned; category 3 represents 10.00 to 99.99 acres burned; category 4 represents

100.00 to 299.99 acres burned; category 5 represents 300.00 to 999.99 acres burned; category 6

represents 1,000.00 to 4,999.99 acres burned; and category 7 represents 5,000.00 or more acres

burned.

More than 73,000 wildfires have affected Florida burning nearly 2.95 million acres during

the period between 1990 and 2006. The number of wildfires decreased 26% from 6,526 total

wildfires in 1990 to 4,802 total wildfires in 2006. But the number of acres burned has not

decreased as much as the number of wildfires. Despite the decrease in the number of wildfires,

the acres burned have only decreased 9%. Approximately 60% of the wildfires fell in the

category 2, followed by category 1 (23%), and category 3 (18%). In other words, more than 97%

of all wildfires were no larger than 100 acres. More acres were burned in 1998 than any other

year during the 17-year period. Nearly 5,000 wildfires were responsible for burning more than

0.5 million acres in 1998.

Note that the year 2005 registered 2,263 total wildfires (only 3% of total) that burned less

than 1% of total acres burned during the 17-year period. Coincidentally, it was the busiest year in

terms of storms affecting Florida as described previously. Figure 3-84 presents total number of

wildfires by category affecting Florida from 1990 to 2006. Most of the wildfires (86%) occurred

during the first six months of the calendar year. Also, the first six months registered

approximately 72% of the total acres burned. May recorded the largest number of wildfires with

10,739 that burned more than 643,000 acres, followed by February (13%) with 373,500 acres

and March (13%) with 264,600 acres. June was the most devastating in terms of number of acres









burned. More than one in every four acres burned (nearly 824,000 acres) during the 17-year

period were recorded during June. In contrast, October registered the fewest number of acres

burned (less than 1%) during the same time period. Figure 3-85 shows the percentage of

wildfires and acreage burned in Florida from 1990 to 2006.

Trop Dep Trop Storm Cat 5 Cat 4 Cat 3 E Cat 1 M Cat 2


/


4,000
3,000
2,000
1,000


mm mm_ mm m mmm_ rnr


U I I
T C^ CO C^ 00 O~ O O
Year
Figure 3-84. Number of wildfires (by category) affecting Florida between 1990 and 2006.
Source: Florida Department of Agriculture and Consumer Services-Division of
Forestry.

I Acres Burned m Number of Wildfires
30% -


20%

10% -

0%-


a h= 0 > o
5 wA ) W%4 0-A
1 r ( 0017


Month

Figure 3-85. Percentage of wildfires and acreage burned in Florida (by month) between 1990 and
2006. Source: Florida Department of Agriculture and Consumer Services-Division of
Forestry.


4


m


ii









Average Temperatures in Origin Regions and Destination CSAs

Weather data including monthly maximum, minimum and average temperatures,

precipitation, and snowfall from different cities in each U.S. region and Florida were retrieved

from the National Climatic Data Center, a NOAA division. Weather data from the cities of New

York (Northeast region), Chicago (Midwest region), Atlanta (South region), and Los Angeles

(West region) were obtained and used as a general representation of the temperatures in each

U.S. region. Also, weather data from Miami (South Florida CSA), Orlando (Orlando CSA),

Tampa (Tampa-St. Petersburg CSA), Jacksonville (Jacksonville CSA), and Fort Myers (Fort

Myers CSA) were collected. Data from these five destination CSAs composed a Florida average.

Florida recorded an annual average temperature of 73 degrees Fahrenheit during the period

between 1990 and 2006. The period from May to September recorded the hottest average

temperatures that ranged from 78 to 81 degrees Fahrenheit. The hottest months on record were

July and August (average of 83 degrees Fahrenheit). The coldest period was between December

and February (average range between 62 and 64 degrees Fahrenheit) with January (average of 60

degrees Fahrenheit) being reported as the coldest month. Across U.S. regions, the Midwest

region reported the coldest temperatures (average of 29 degrees Fahrenheit in January) during

winter, followed by the Northeast region (average of 35 degrees Fahrenheit in January). The

West region was four to six degrees cooler than Florida during winter, and reported the smallest

fluctuations in temperatures when compared to the temperatures recorded in Florida.

Also, all U.S. regions but the West region reported similar temperatures to Florida during

the summer months of July and August. The Northeast region recorded an average temperature

of 78 degrees Fahrenheit during these two months, only three degrees cooler than Florida. Also,

the Midwest region recorded average temperatures in the rage of 74 to 76 degrees Fahrenheit,

while the South region reported average temperatures in the 80s during these two months. In









contrast, the West region was, on average, 13 degrees Fahrenheit cooler than Florida during

these summer months. The difference between the average temperature in Florida and each of

the four U.S. regions are shown in Figure 3-86.

I + Northeast Midwest -A-South West


35 I I


25 ------------- ---I I ------I I-------------------------------
S 20 ------------ --------- ---------- ------- ------ _----------- ----



15 --- -------- I -
10 --------------- -S--------- ------- -------- ----- 1----
U C 5- -.----------f- --- -----r-- --YC- C-L----- ------ -------
0
I I I I I I I I I I I

SJan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month
Figure 3-86. Difference between average temperatures in Florida and each U.S. region between
1990 and 2006. Source: National Climatic Data Center.

Precipitation in Florida

Florida recorded an annual rainfall of 55.3 inches during the period between 1990 and

2006. The year 1997 was the wettest (64.4 inches), followed by 1994 (62.9 inches) and 2005

(61.3 inches) during the 17-year period, while 1990 reported the lowest rainfall levels at 39.6

inches. The South Florida CSA recorded the highest precipitation levels during the same period

(65.8 inches), followed by the Fort Myers CSA (57.9), Jacksonville CSA (54.0 inches), Orlando

CSA (51.8), and Tampa-St. Petersburg CSA (46.9 inches). The period between June and

September recorded the highest precipitation levels during the calendar year. Average rainfall in

these months ranged from 7.6 to 8.8 inches. The driest month was November with an average

rainfall of 2.2 inches. The wettest month in each of the top five destination CSAs was June with

an average rainfall range of 7.1 to 10.6 inches. Figure 3-87 presents average rainfall by month in

each of the six destination CSAs between 1990 and 2006.


145


I










---- Tampa-St. Petersburg Orlando A Jacksonville
-0-Ft. Myers -- South Florida 4CSA Average

I I I I I I I I I I I


------l----------- ------- -----
i I L -_ I .- I



--------T--- --- --- ------- -- ---Y ------
,A, __ .,,.. L ,_ ,
,, \ '


Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
----Tampa-St. Petersburg 2.3 2.9 2.4 2.4 1.6 7.1 7.2 8.0 6.5 2.4 1.3 2.9
Orlando 2.1 2.3 3.3 2.7 3.3 8.1 7.6 7.3 5.7 3.8 1.9 2.8
A Jacksonville 3.3 2.8 3.7 2.7 2.4 7.6 7.3 6.1 8.2 4.7 1.9 2.9
-Ft. Myers 2.3 1.8 2.2 2.4 2.6 10.6 9.6 10.7 9.2 2.5 2.0 1.7
-- South Florida 1.7 2.1 2.7 3.1 5.5 10.4 6.3 9.3 10.9 8.1 3.7 2.3
-*-CSAAverage 2.3 2.4 2.8 2.7 3.1 8.8 7.6 8.3 8.1 4.3 2.2 2.5

Figure 3-87. Monthly average rainfall in each of the five destination CSAs in Florida between
1990 and 2006. Source: National Climatic Data Center.

Crime Rates in Florida

Crime data were collected from the Florida Law Enforcement Agency Uniform Crime

Reports and are available online. The crime index includes the following categories: violent

crimes of murder, forcible sexual offenses, robbery and aggravated assault; and the property

crimes of burglary, larceny-theft, and motor vehicle theft that were reported to Florida law

enforcement. The crime index is defined as the number of crimes reported divided by the

population and multiplied by 100,000.

Since 1990, Florida's crime index has decreased by 45%. All crime categories included in

the crime index have decreased during the 17-year period, especially the ones most frequently

committed against tourists: burglary of holiday homes, vehicle theft, and robbery. Burglary has

experienced the largest decrease (38%) from 275,104 total burglaries in 1990 to 170,133 in 2006.

In addition, robbery has decreased 37% from 54,015 robberies in 1990 to 34,123 in 2006.









Similarly, motor vehicle theft has experienced a 25% decrease during the same time period.

Figure 8-88 presents the crime index per 100,000 in Florida from 1990 to 2006.



9,000

8,000

7,000

p 6,000

5,000

4,000 .


Year

Figure 3-88. Annual crime index in Florida between 1990 and 2006. Source: Florida Law
Enforcement Agency Uniform Crime Reports.

Chapter Summary

This chapter presented descriptive statistics for the domestic and international passenger,

freight, and mail traffic traveling to Florida by means of airline transportation. All passenger,

mail, and freight statistics were presented according to the particular travel destination CSA.

Overall, the South region presented the highest levels of air passenger traffic, while the West

region exhibited the lowest levels of passengers traveling to Florida. The top three destination

CSAs were South Florida, Orlando, and Tampa-St. Petersburg, respectively. Regarding

international passengers traveling to Florida, there were more passengers traveling from Latin

America than any other region. Passengers from Europe and Canada followed. The South Florida

CSA received more passengers from Latin America than any other international origin, while the

Orlando CSA received more from Europe and the Tampa-St. Petersburg from Canada.









Airline ticket prices from the West region were more expensive than from any other region

traveling to Florida. The Northeast region yielded the cheapest airline tickets. Airline ticket

prices from Europe were more expensive, while airline tickets from Latin America were the

cheapest among international origins.

The chapter also included economic indicators, as well as average temperatures and

precipitation related to the origin U.S. regions. It also presented weather statistics on average

temperatures, precipitation, and hurricanes and wildfires affecting Florida. Oil prices, advertising

expenditures, and crime rate statistics from Florida were also shown in this chapter. Advertising

expenditure statistics are highly seasonal. Most of the generic and brand advertising expenditures

occurred in the first quarter of every year.









CHAPTER 4
THEORETICAL FRAMEWORK

Motivation

Demand for domestic air passenger traffic to Florida can be modeled in several ways.

Chapter 2 discussed several approaches currently used by econometricians. Such approaches

range from simple auto-regressive models to more complicated specifications such as time-

varying parameter models. Most of these studies had used annual data to estimate demand and

the scope had been limited to forecasting international demand for tourism. Nevertheless, studies

by Phakdisoth and Kim (2007) and Proenca and Soukiazis (2005) used a partial adjustment

model framework to identify drivers of demand for tourism in Laos and Portugal, respectively.

Conversely, this study developed a partial adjustment framework to model domestic demand as

opposed to international, used monthly data instead of annual data, and more importantly,

expanded the dynamic1 structure of the partial adjustment model from one to three lagged

dependent variables.

In order to motivate the partial adjustment model framework, first let a model be

constructed where the number of passengers (Pt) is defined as a function of its past

values (Pt-j) as shown in Equation 4-1. Time series literature identifies auto-regressive

integrated moving average (ARIMA) models as a popular approach to model and forecast this

type of specification:

Pt = f(Pt-j). (4-1)

Since the ARIMA model lacks a base of economic theory critical for policy implications,

an extension can be made to add variables that can explain the behavior of Pt. Number of


1 The term dynamic refers to lagged values of the dependent variable included as right-hand side variables to model
demand response to economic stimuli as explained by W.H. Greene in Econometric Analysis, Chapter 19.









passengers is now defined in Equation 4-2 as a function of its past values and other explanatory

variables:

Pt = f(Pt ,Xp). (4-2)

Explanatory variables in XP attempt to account for the levels of Pt while still recognizing

that there remain patterns that cannot be directly explained. Hence, inclusion of Pt-j is still

expected to be of major importance and represents the dynamic component of the model.

As consumers, in general, do not immediately adjust to changes in their demand

determinants, appropriate dynamic systems are essential, before plausible behavioral hypothesis

can be tested (De Mello and Fortuna 2005). These authors also agreed that suitable dynamic

generalizations of demand systems are a rare feature in empirical studies. This study, by

specifying a partial adjustment model, attempts to address this issue as a contribution to the

research literature.

Nerlove (1972) used adjustment costs example to explain why consumers do not adapt

instantaneously to changes in prices and how dynamics of behavior is a more suitable approach

to analyzing short-term phenomena. The static or long run framework does not allow prices to

change. It assumes that prices are the same through time and hence, consumer behavior rests in a

state of equilibrium. This assumption is very restrictive and may only be useful when applying a

comparative statics analysis.

Introduction of dynamics is necessary to the economic framework to improve the

forecasting accuracy of models. Vanhove (2005) argues that lagged dependent variables are

often used to take into account a time lag in the relationship between a dependent variable and an

independent factor. Inclusion of lagged dependent variables in tourism demand functions allows

for habit persistence and supply rigidities. Sources of these dynamics within the demand for









airline transportation to Florida can be attributed to habits of consumption created through time

and by events in previous periods that affect consumer behavior.

Habit, which is unobservable, suggests that consumers' current purchases are influenced

by past purchases. In the case of air travel to Florida, habit can be a result of factors such as

ownership of real estate or time shares in the state, institutional structure such calendar holidays

that attract visitors, sense of security and safety, and a supply of unique products (e.g., beaches,

attractions). In addition, consumers return to Florida to reduce search costs related to finding

another place to travel. Consumers are also already familiarized with the environment and know

the services provided which creates a sense of comfort and security.

Consumers face difficulties in order to respond to events immediately. Adjustment costs

and imperfect information cause this consumers' rigid response. The system in place provides

little or no flexibility to consumers when hurricanes, wildfires, and terrorist attacks occur.

Rescheduling or cancellation of bookings will most likely carry a transaction cost. While some

consumers will plan ahead and buy insurance to reduce their risk, they still incur the cost of

insurance.

Such habits and events prevent consumers to adjusting from one period to another and this

is where a dynamic structure is needed to explain demand behavior. Lagged explanatory

variables that take into account such behavior should be included in the model. By introducing a

lagged dependent variable as an explanatory variable, the model attempts to capture any

persistence effect or rigidity of tourist behavior. The underlying assumption is that the period

which is required for full adjustment exceeds the interval of observation (Nerlove 1958).

Other authors have used the same principles of the partial adjustment model to include

habit persistence in the estimation. Chamera and Deadman (1992) used a similar framework to









explain how inertial factors prevent an immediate movement to a new desired consumption level.

They also argue that this framework can be extended to aggregate consumption theories which

also lead to estimating forms similar to the ones presented in the next section.

The Partial Adjustment Model

Partial adjustment models include short run dynamics that static models fail to address.

The partial adjustment model has two components, a static component and a dynamic

component. The static component, shown in Equation 4-3, states that desired amount of the

dependent variable (y*) is determined by some explanatory variables(X):

y 0 = ao + Plxt + Ut. (4-3)

The dynamic component, shown in Equation 4-4, can be explained as follows. The

difference between the current level (Yt) and the previous level (yt-1) is a portion of the

difference between the desired level (y*) and the previous level (yt-1).

Yt Yt-1 = O(y; Yt-i). (4-4)

The coefficient 0 is the speed of adjustment coefficient and can take values between 0 and

1. If 0 0, then lim yt = Yt-1 which implies a slow speed of adjustment. Consumers take
0-0

more than one period to fully adjust to short run changes in their demand drivers. If 08

1, then lim yt = yt which implies a high speed of adjustment. In other words, consumers adjust

immediately to any short run changes in their demand drivers. If consumers adjust fairly quickly

to short run changes in their demand drivers, the adjustment coefficient (0) becomes irrelevant

and therefore, the dynamic specification expressed in Equation 4-4 is not necessary. The

following equation is obtained when substituting Equation 4-3 into Equation 4-4:









Yt Yt-1 = 0 [(o + Pixt + ut) -yt-1]

Yt = 0 [(o + Pixt + ut) -yt-1] + yt-1

Yt = ao + Ol1xt + Out Oyt-1 + Yt-1

yt = Oao + Ollxt + (1- O)yt_- + et. (4-5)

The coefficient (f/) in Equation 4-3 can be interpreted as the long run relationship between

the dependent variable Y and the explanatory variable X. Note that this long run relationship

cannot be directly estimated since the equilibrium quantity demanded (y*) cannot be observed.

An unobservable quantity demanded may arise due to frequent changes in prices and income.

However, the partial adjustment model specification provides a useful way to derive estimates of

the long run relationship. Equation 4-5 shows that, under the partial adjustment model

framework, the long run coefficient (f/) is multiplied by the speed of adjustment coefficient (0).

Since there are two coefficients (0, p) associated with X, a transformation is needed to estimate

Equation 4-5. The estimating equation is presented in Equation 4-6. Note that the new estimable

coefficients (7r) are now interpreted as the short run relationship between the dependent variable

Y and the explanatory variable X:

Yt = no + 1lXt + 7T2Yt-1 + ut, (4-6)

where o0 = 0a07, nl = 0/, and T2 = (1- 0).

There is no adjustment if 72 = 1 (i.e., 0 = 0) and the dependent variable at current period

is a function of its past values which is a function of the X/ related to that past period. Any

changes in the explanatory variables in that past period are still affecting the dependent variable

in the current period. Consumers have not adjusted to any changes in the determinants from the

past period yet and hence, equilibrium has not been achieved. In this case, the dynamic

specification is relevant in order to have a correct interpretation of the coefficients.









The opposite extreme would be if 7T2 = 0. The speed of adjustment (0 = 1) implies

that limso_ yt = y* Therefore, the dependent variable is solely determined by the explanatory

variables X at the current period. The dynamic component of the model specified in Equation 4-5

is no longer relevant because consumers fully and instantaneously adjust to any changes in the

determinants and consumers are able to achieve equilibrium within the same time period. In

other words, short run elasticities are equivalent to long run elasticities and equilibrium is

achieved immediately.

Domestic Air Passenger Traffic Partial Adjustment Model

The domestic air passenger traffic partial adjustment model (DAP-PAM) identifies factors

that influence demand for air passenger traffic traveling to Florida. As described by Pollak

(1970), past consumption patterns are an important determinant of present consumption patterns.

Adamowicz (1994) suggested that consumers may be "learning by doing" when they visit a site,

and thus repeated visitation occurs (habits). The DAP-PAM will determine if a habit formation

pattern exists among passengers traveling to Florida.

Assume that there is a desired domestic air passenger traffic demand (DAP*) and it is a

function of some explanatory variables(X). These explanatory variables, presented in Equation

4-7, include airline ticket prices, personal disposable income, average temperature, and average

precipitation. Wildfires, advertising expenditures, and dummies for storms, monthly seasonality,

and terrorism (i.e., 9-11 terrorist attacks) are also included in X.

S FAREt, PCDIt,TEMPt, PCPt, FIREt,ADVt,
HCATLOWt, HCATMEDt, HCATHGHt,
DAPt* = f(X) = f TER2t ,TER3t ,TER4t ,TERSt, TER6t, TER7t, (4-7)
MTH2t, MTH3t, MTH4t, MTH5t, MTH6t, MTH7t ,
SMTH8t,MTH9t ,MTH10t,MTH11t,MTH12t

Then, the desired demand for air passenger traffic can be given a non-linear specification

where its explicit form can be expressed as









DAP* = expP0+E FAREt1 PCDIt/2 TEMPt3 PCP/t4 FIREt5 *

expf6HCATLOWt expf7HCATMEDt expfsHCATHGHt ADVtf *

exp j=2 8+jTERi expk=2= f14+kMTHk. (4-8)

The adjustment process is assumed as:

0
( DAPt, DAPt

j=2,42,,12 DAP j=2,4,12 DAPt)


where 0< 0 <1 and Y w = 1. (4-9)
j=2,4,12

Taking the log-linear form of Equation 4-9, the result is


nDAPt 4 wnDAPt-j = InDAP* InDAPt-j
j =2,4,12 j =2,4,12


InDAPt = 0 InDAPt wjlnDAPt-j + wjInDAPt-
j =2,4,12 j=2,4,12


InDAPt = 0lnDAPt* 08 wjInDAPt-, + wjlnDAPt-j
j=2,4,12 =2,4,12


InDAPt = 6lnDAPt + (1 0) ( ,4 w2 InDAPt-j
j=2,4,12


where 0 < 0 < 1 and Y = 1. (4-10)
j=2,4,12

The first component in Equation 4-10 represents the static component of the DAP-PAM

and the second component refers to the dynamic component of the model. The static component

was constructed using economic theory and included variables illustrated in Equation 4-7.









Variables included in the dynamic component resulted from a spectral analysis performed to the

domestic passenger data. Note that a restriction on the weights has been imposed where the sum

of the weights must add up to one. The restriction was imposed in order to identify the elasticity

of adjustment coefficient(0), otherwise no inference can be made about the level of response

realized in the current period versus past periods. A detailed explanation of all variables included

in the model is presented in the following sections.

Construction of Empirical Domestic Air Passenger Model

It is hypothesized that the elasticity of adjustment coefficient (0) varies across destination

CSA and origin region (CSA-ORG) pair. Therefore, separate demand equations have been

specified for each CSA-ORG pair. Each equation includes a dependent variable, passenger traffic

per 100,000 persons, on the left-hand side and a constant term and twenty-nine variables on the

right-hand side. The right-hand side includes 26 variables representing the static component of

the model and three variables representing the dynamic component of the model. The static

component includes a CPI-adjusted air ticket price specific to the CSA-ORG pair, income,

temperature index, and rainfall index specific to the origin region, a fire index and three storm

dummy variables specific to all Florida, advertising expenditures, six 9-11 terrorist attack

dummy variables, and eleven monthly dummy variables. The dynamic component includes

dependent variable lagged two, four, and twelve months as determined by the spectral analysis

performed on the domestic air passenger data. The spectral analysis will be discussed later in the

chapter.

The dependent variable, DAP, represents number of passengers traveling from a particular

origin region (ORG) to a specific destination (CSA) adjusted by population (100,000 habitants)

of the ORG. Population estimates reported annually by the U.S. Census were used to estimate









monthly population values. Refer to Chapter 3 for an explanation of the methodology used in the

selection and aggregation of airline passenger traffic for each CSA-ORG pair.

One major limitation of the variable is that it does not provide information about the

passenger. The variable DAP fails to identify the passenger as a person traveling for leisure or

business or a resident returning home. It only provides information about the passenger's

itinerary. Recall that every itinerary has two legs: first leg and returning leg. Still, the variable

could give valuable insights on travel patterns. According to Visit Florida", more than one half

of domestic visitors travel to Florida via commercial flights. Since the objective of the study is to

identify general patterns of air travel to Florida, the variable was used to model such patterns.

Description of Variables in the Static Component of the DAP-PAM

The variable FARE represents air ticket prices of each CSA-ORG pair adjusted by the

national Consumer Price Index for All Urban Consumers for airfares (CPI-AFARE) published on

a monthly basis by the U.S. Bureau of Labor Statistics. Airline ticket prices to each Florida

destination were deflated by the national average prices. The CPI-AFARE is a reflection of

airline ticket prices from all domestic origins to all domestic destinations. Therefore, FARE

reflects deflated airline ticket prices to Florida relative to airline ticket prices to all destinations

nationwide. If deflated airline ticket prices to Florida were exceptionally high, consumers may

go somewhere else because prices to Florida are more expensive. Using the CPI-ALL to deflate

ticket prices would give similar results since the correlation between both indexes is 0.93.

Refer to Chapter 3 for an explanation of the methodology used in the selection and

aggregation of airline ticket prices for each CSA-ORG pair. The INC variable represents the

average per capital personal disposable income for each region. Note that the average per capital

personal disposable income from Florida was excluded from the South region. Similarly to the









population variable, the annual average personal disposable income reported by the Bureau of

Economic Analysis (BEA) was used to estimate monthly values.

The next three variables included in the model are related to weather either in the origin or

at the destination. TEMP and PCP refer to indexes of average monthly temperature (in degrees

Fahrenheit) and average monthly rainfall (in inches) at the origin region. FIRE represents an

index of wildfires affecting Florida and HCATLOW, HCATMED, and HCATHGH refer to

dummy variables for tropical depressions and storms and hurricanes affecting Florida.

TEMP and PCP reflect the index of the temperature and precipitation reported in each U.S.

region on a monthly basis normalized by an estimated monthly average value. A value of 1.0

represents average conditions during that month, whereas any value less or greater than 1.0

represents temperature and rainfall below or above the normal. The FIRE variable refers to a

value of weighted wildfires normalized by an estimated average value of such weighted wildfires

for each month. The weighted value of fire is a combination of the size and number of wildfires

where the number and size of wildfires are weighted by the size of the small wildfires. Therefore,

the value will give a higher value to larger wildfires (greater impact) compared to the smaller

wildfires.

HCATLOW includes the number of tropical depressions and storms that affected Florida.

HCATMED includes number of hurricanes of category 1 and 2 and HCATHGH includes

number of hurricanes of category 3, 4, and 5. These categories follow the classification defined

by the Saffir-Simpson hurricane scale.

The ADV variable represents total advertising expenditures by destination CSA. All

destination CSAs include total generic advertising efforts performed at the city, county, and state

level. Note that these totals were not available for each destination CSA and therefore, the total









was used in each destination CSA. The variable also includes, if available, brand advertising

expenditures of those companies that do business in a specific destination CSA. For example the

ADV for the South Florida CSA includes total generic advertising expenditures for Florida plus

brand advertising expenditures of Carnival Cruise Lines. In addition to total generic advertising

expenditures, the ADV for the Orlando CSA is composed by brand advertising expenditures

from the Walt Disney Company, Universal Studios, and SeaWorld. Similarly, advertising

expenditures from Busch Gardens were included in the ADV variable for the Tampa-St.

Petersburg CSA. Since no specific brands were available for the CSAs of Jacksonville and Fort

Myers, total generic advertising expenditures for Florida were used.

The dummy variables TER1-TER7 represent the terrorist attacks of September 11, 2001 in

New York City, Washington, D.C., and Pennsylvania. TER1 takes a value of one for all months

prior to September, 2001 and 0 otherwise. TER2 takes a value of one for the twelve months

following the terrorist attack including September 2001 and zero otherwise. TER3-TER7 takes a

value of one on a twelve-month sequence from September 2002 to August 2007. For example,

TER3 takes a value of one for each month from September 2002 to August 2003, zero otherwise.

TER4 takes a value of one for each month from September 2003 to August 2004, zero otherwise,

and so on. TER1 was dropped from the estimation to avoid the dummy variable trap (i.e., perfect

multicollinearity). Therefore, all inferences of TER2-TER7 are compared to the base, TER1 (pre

9-11 levels).

By setting each TER level in a twelve-month sequence, the effect of the variable has not

been confounded with the adjustment associated to seasonality. Another possibility to construct

the dummy variable was the one where a dummy variable would take a value of zero for all pre

9-11 months and a value of one for all months after 9-11. Such before and after 9-11 approach









would have hindered the ability to make inferences about the recovery process that could occur

through time and therefore was ultimately not considered.

Monthly seasonality dummies and lagged dependent variables are included in the right-

hand side to account for seasonal and consumption patterns in previous periods, respectively.

Dummies for monthly seasonality reflect any event that occurs every year influenced by either

weather or institutional structure. For example, Major League Baseball's spring training occurs

in March and April of every year. Holidays such as spring break, Fourth of July, Memorial Day

Weekend, Thanksgiving Weekend, and so on, give structure to the supply of services available to

consumers.

Availability of these services is limited, since the event only occurs at a certain time of the

year. Therefore, consumers will travel that time of the year when these services are available. For

example, a consumer interested in bird-watching in Florida will most likely visit the state during

the bird-watching season. Also, car-racing fans will be enticed to travel to Daytona Beach in

February more than any other month for the annual Daytona 500. Clearly, such events are a

source of seasonal patterns and the use of monthly seasonal dummies in the model will capture

them.

On the other hand, the lagged dependent variables show how past events influence

consumer's decision to travel in the current period, and hence, capture consumer behavior.

Monthly dummy variables are represented by MTH1-MTH12 to account for seasonal effects.

Similarly to TER1, MTH1 was dropped from the estimation in order to avoid the dummy

variable trap. Therefore, all inferences of MTH2-MTH12 are compared to the base, MTH1

(January). Transformations of the variables were performed and stored using TSPTM software. A

copy of the TSPTM program is illustrated in Appendix B.









Identification of the Variables in the Dynamic Component of the DAP-PAM

Identification of the dynamic component can be made by either running a correlogram or

conducting a spectral analysis on domestic airline passenger data. Since the correlogram presents

some ambiguity surrounding the choice of a suitable dynamic model (Harvey and Todd 1983), a

spectral analysis has been conducted instead.

The spectral analysis helps identify any seasonal fluctuations of different lengths in the

data. Figure 4-1 shows the power spectrum of the domestic air passenger traffic data for the

Florida-Northeast pair. Periodogram values were plotted against the time domain (months) and

are interpreted in terms of variance of the data at the respective month. The spectral analysis

shows a high level of variance every two, four, and twelve months. All other 23 CSA-ORG pairs

exhibited similar behavior in the level of variance at months two, four, and twelve.

The lack of information in identifying which months report the highest variance is a major

limitation of the spectral analysis. One must use some theoretical expectations to explain the

peaks illustrated by the spectral analysis. It can be hypothesized that individuals go on vacation

once a year to explain the twelve months peak in the variance of passenger traffic. For example,

the school holiday season (June through August) represents a key season for those touristic

destinations trying to attract visitors to their states. Since most of the Florida attractions are

family-oriented such as Disney World theme parks, it can be assumed that Florida will

experience an increase in visitors every year during the school holiday season.











14.0 4U

12.0

i 10.0

8.0 24

6.0 119

4.0

2.0




Periodocity (in months)

Figure 4-1 Power spectrum of domestic air passenger traffic data for the Florida-Northeast (33-
1001) pair.

The four-month peak can be attributed to the institutional structure of holidays. A major

calendar holiday comes around roughly every four months a such as Spring Break (March),

Fourth of July (July), and Thanksgiving Day (November). Florida is a major destination during

spring break while the Fourth of July is in the middle of summer which attracts many to the

Florida beaches. Thanksgiving holiday records one of the highest levels of air passenger traffic

around the country also and Florida benefits from the long holiday weekend. It is more difficult

to explain the two-month peak since no clear reason to explain a high variance of the passenger

traffic can be identified. Nevertheless, the lag was included in the model as the spectral analysis

suggested.

Given the results given by the spectral analysis performed on the dependent variable and

the theoretical reasoning discussed above, the dynamic component of the model for domestic air

passenger traffic includes the dependent variable lagged two, four, and twelve months. Table 4-1

presents a summary of the variables included the DAP-PAM.









Table 4-1. Summary of the variables included in the DAP-PAM.
Variable Description
PASSENGERS DAP D,O= Domestic air passenger traffic traveling from region (0) to
destination (D) adjusted by population of origin (0); measured in
passengers per 100,000.
FARE FARED,O= Average air ticket price to destination (D) from origin (0)
adjusted by the CPI-U Airfare component; measured in U.S. dollars.
INCOME PCDIo= Per capital personal disposable income in region (0),
measured in U.S. dollars.
TEMPERATURE TEMPo=Index of average temperature in region (0); unity.
PRECIPITATION PCPo= Index of total precipitation in region (0); unity.
FIRE FIRED= Index of wildfires in Florida; unity.
STORMS HCATLOWD=(Tropical storms & depressions); unity.
HCATMEDD=(HCAT1+HCAT2); unity.
HCATHGHD=(HCAT3+HCAT4+HCAT5); unity.
ADVERTISING ADVD =Advertising expenditures incurred to promote destination
(D); measured in thousands of U.S. dollars.
9-11 ATTACKS TER2=[ (YY:MM>01:08 & YY:MM<02:09)=1] or (otherwise=0)
TER3=[ (YY:MM>02:08 & YY:MM<03:09)=1] or (otherwise=0)
TER4=[ (YY:MM>03:08 & YY:MM<04:09)=1] or (otherwise=0)
TER5=[ (YY:MM>04:08 & YY:MM<05:09)=1] or (otherwise=0)
TER6=[ (YY:MM>05:08 & YY:MM<06:09)=1] or (otherwise=0)
TER7= (YY:MM>06:08=1) or (otherwise=0)
SEASON MTHS2=(February=l) or (otherwise=0)
MTHS3=(March=l) or (otherwise=0)
MTHS4=(April=l) or (otherwise=0)
MTHS5=(May=1) or (otherwise=0)
MTHS6=(June=l) or (otherwise=0)
MTHS7=(July=1) or (otherwise=0)
MTHS8=(August=l) or (otherwise=0)
MTHS9=(September=l) or (otherwise=0)
MTHS10=(October=l) or (otherwise=0)
MTHS1 l=(November=l) or (otherwise=0)
MTHS12=(December=l) or (otherwise=0)
PASSENGERSt-2 DAP t-2,D,O= Domestic air passenger traffic traveling from region (0)
to destination (D) adjusted by population of origin (0) lagged two
months; measured in passengers per 100,000.
PASSENGERSt-4 DAP t-4,D,O= Domestic air passenger traffic traveling from region (0)
to destination (D) adjusted by population of origin (0) lagged four
months; measured in passengers per 100,000.
PASSENGERSt-12 DAP t-12,D,O= Domestic air passenger traffic traveling from region (0)
to destination (D) adjusted by population of origin (0) lagged twelve
months; measured in passengers per 100,000.









Using the log-linear form of Equation 4-5 and including the result in Equation 4-7 and

applying the restriction on the weights of the lagged dependent variables, an expression for the

DAP-PAM can be represented in log-linear as

InDAP(D,O),t = Of(D,O),t + OP1(D,O) InFARE(D,o),t + 8O2(D,O) InPCDI(o),t +

Of3(D,O) In TEMP(o),t + OP4(D,) In PCP(o),t +

8P5(D,O) In FIRE(D),t + Of6(D,o)HCATLOW(D),t +

8Of7(D,o)HCATMED(D),t + Ofl8(D,O)HCATHGH(D),t +

7
Ofl9(D,O)ADV(D),t + 0 YP8+j(D,O)TERj(D,O) +
k=2
12
0 f 14 +j (D,o)MTHj (D,) +
j=2

(1 O)w2(ln DAPt_2 In DAPt-12) +

(1 O)w4(ln DAPt_4 In DAPt-12) +

(1 0)In DAPt-12 + 0E(D,O)t (4-11)

Also note that there are 24 CSA-ORG pairs resulting from four origin regions with

passenger traffic to six destination CSAs. The four pairs for the CSA=33 represent the addition

of the top five destination CSAs and the other smaller destination CSAs not included in the

individual analysis.

Identification of the destination CSA (D) and the origin region (0) is presented in Table 4-

2. Refer to Chapter 3 for a detailed explanation regarding the selection and aggregation

procedures performed to construct the five individual CSAs.









Table 4-2. Identification of destination CSAs and origin U.S. regions.
Destination (D) Destination CSA Origin (0) Origin U.S. Region

1 South Florida 1001 Northeast

5 Orlando 1002 Midwest

8 Tampa-St. Petersburg 1003 South

9 Jacksonville 1004 West

14 Fort Myers

33 Florida


Coefficients (fl) in Equation 4-5 and Equation 4-11 can be interpreted as the long run

relationship between the demand for air passenger traffic and X. Coefficients (fl) associated with

the variables in logarithm form are interpreted as the average propensity to travel given a change

in X. Note that this long run relationship cannot be directly estimated since the equilibrium

quantity demanded (DAP*) cannot be observed because explanatory variables are continually

changing. However, the partial adjustment model specification provides a useful way to derive

estimates of the long run relationship. Equation 4-11 shows that the long run coefficients (fl) are

multiplied by the speed of adjustment coefficient (0).

The speed of adjustment coefficient (0) can now be interpreted as the elasticity of

adjustment since the model has been specified in logarithm form. Since there are two coefficients

(0, f) associated with X, a transformation is needed to estimate Equation 4-11. The new

estimable coefficients (7Ti) are illustrated in Equation 4-12 and interpreted as the short run

relationship between the demand for air passenger traffic and X.









InDAP(D,O),t = LO(D,O),t + 71(D,O) InFARE(D,O),t + T2(D,O) InPCDI(o),t +

73(D,O) In TEMP(o),t + 7r4(D,O) In PCP(o),t + 75(D,O) In FIREt +

7T6(D,o)HCATLOWt + 7T7(D,o)HCATMEDt +

n8(D,O)HCATHGHt + C9(D,o)ADV(D),t +

7 12
Y, 78+j(D,O)TERj(D,O) + 14+j(D,O)MTHj(D,O) +
k=2 j=2

T27(D,O)(In DAPt_2 -In DAPt-12) +

T28(D,O)(ln DAPt-4 In DAPt-12) + 729(D,O) In DAPt-12 +

E(D,O)t

where 729 =(1 0), 0 < 8 < 1, Y.j=2,4,12 = 1, and Ti = 0/i i= 1,...26. (4-12)

There is no partial adjustment if 0 = 0 and the demand for domestic air passenger traffic

is determined exclusively by its past values. The opposite extreme would be if 0 = 1, which

implies that limos, DAPt = DAP* Therefore, demand for domestic air passenger traffic is

solely determined by explanatory variables X and the dynamic component of the model specified

in Equation 4-9 is no longer relevant. It is hypothesized that the elasticity of adjustment (0) lies

between 0 and 1. In other words, demand for air passenger traffic to Florida can be explained by

conditions occurring in the current period and conditions that occurred in past periods.

Estimation Possibilities

Initially, the model in Equation 4-12 was estimated using ordinary least squares (OLS).

Several tests were performed to identify any heteroskedasticity, serial correlation, and

stationarity. Since lagged dependent variables were included in the right-hand side of the

equation, the augmented Dickey-Fuller test was performed to determine if the data contained a









unit root. The null hypothesis was rejected at a 95% confidence level concluding that residuals

from the OLS regression are stationary.

The Durbin h-test was conducted to determine if there was a correlation in the errors

corresponding to successive time periods. Given the results from this test, the null hypothesis of

no serial correlation was rejected. The OLS estimation approach can no longer be considered a

viable estimation procedure for Equation 4-12 because it yields inconsistent estimates. In order

to correct for this problem, a first-order auto-regressive process (AR1) for the error was included

in Equation 4-12.

Et = iEt-1 + Tt, where t~N(0,a2) (4-13)

Therefore, by combining Equation 4-12 and Equation 4-13, an estimating equation that

takes into account the AR(1) process can be defined as

In DAP(D,O),t = TO(D,O),t + 1 (D,O) In FARE(D,O),t + 72(D,O) In PCDI(o),t +

7T3(D,O) In TEMP(o),t + 74(D,O) In PCP(o),t + 5S(D,O) In FIREt +

7T6(D,o)HCATLOWt + 7T7(D,o)HCATMEDt +
7
W8(D,O)HCATHGHt + T9(DO)ADV(D),t + Y 8+j(Do)TERj(Do) +
k=2
12
Y r14+j (D,0)MTHj(D,O) + 27(D,)(lnDAPt-2 nDAPt-12) +
j=2

7T28(D,O) (In DAPt-4 In DAPt-12) +

7T29(D,O) In DAPt-12 + PE(D,O),t-1 + r(D,O),t,


where 29 =(1- 0), 0 < 8 < 1, w = 1, 7-t~N(0, 2)
j=2,4,12

and 7i = fli i = 1,...,26. (4-14)









Another estimation possibility that had to be considered was the fact that the errors could

be correlated not only over time, but also across cross-section units. Initially, OLS has treated

each of the 24 CSA-ORG pairs in the DAP-PAM as separate equations. The procedure fails to

address any auto-correlation and also the possibility that the errors from some of the equations

could be correlated. The AR (1) estimation procedure takes into account the auto-correlation but

not the cross-section correlation. Then, the technique of multivariate regression generally gives

more efficient estimates than OLS and AR (1) regressions applied separately to each equation.

There are three ways to model this possible cross-section correlation. One general

possibility is the fact that all 20 equation residuals are correlated with each other due to some set

of unobservables common to all destinations and origins alike. The second possibility is that

there is a set of unobservables unique to the origin region not captured by the specified model.

The third and final possibility is that there are some unobservables unique to the destination for

which the variables specified in the model do not account.

Therefore, four variations of the model were specified and are summarized as follows:

1. AR1 estimation for each equation separately: assumes that there is no presence of
dependent errors across equations. It only takes into account an AR (1) process in the
error term. There is only one set of 24 equations.

2. SUR-AR1-ALL estimation that includes all U.S. regions traveling to all destination
CSAs; assumes that there is information in all U.S. regions and destination CSAs not
being explained by the model but that could be temporally dependent across all U.S.
origins and destination CSAs, and hence, a source of correlation. The model is also
considering a correlation in the errors corresponding to successive time periods (i.e., AR
(1) process). There is only one set of twenty equations. The destination CSA representing
Florida (i.e., CSA= 33) was not included in each set since it represents the aggregation of
all destination CSAs in Florida and was estimated separately.

3. SUR-AR1-ORG estimation for each U.S. region traveling to all destination CSAs:
assumes that there is information unique to a particular region not explained in the model
that could be temporally dependent across destination CSAs. The model is also
considering a correlation in the errors corresponding to successive time periods (i.e., AR
(1) process). There are four sets of equations representing each U.S. origin and each set
includes five equations. For example, the set for the Northeast region includes all five









destination CSAs related to the Northeast region: 1-1001, 5-1001, 8-1001, 9-1001, and
14-1001. The destination CSA representing Florida (i.e., CSA= 33) was estimated
separately.

4. SUR-AR1-CSA estimation for all U.S. regions coming to a particular destination CSA:
assumes that there is information in a particular destination CSA not explained in the
model that could be temporally dependent across regions. The model is also considering a
correlation in the errors corresponding to successive time periods (i.e., AR (1) process).
There are six sets of equations representing each U.S. origin and each set includes four
equations. For example, the set for the South Florida CSA includes all four origin U.S.
regions related to the South Florida CSA: 1-1001, 1-1002, 1-1003, and 1-1004.

Chapter Summary

Chapter 4 presented the theoretical framework of the partial adjustment model and also the

empirical application to the demand for air passenger traveling to Florida. The model, which

includes a static and a dynamic component, is an attractive approach since it yields a speed of

adjustment coefficient that shows how fast passengers adjust to events that affect their decision

to travel to Florida. An empirical model was introduced to show how the demand for air

passenger traffic could be represented and estimated. Economic theory was used to build the

static component and spectral analysis was performed to determine the dynamic component of

the model. Preliminary diagnostics showed that the model is stationary (augmented Dickey-

Fuller test). In addition, the Durbin h-statistic suggested the errors have an auto-regressive

scheme of order 1. These diagnostics helped concluded that OLS estimates will be inconsistent

and therefore, other estimation procedures were needed.









CHAPTER 5
RESULTS

Separate equations for each CSA-ORG pair were constructed. Each equation includes a

dependent variable, passenger traffic per 100,000 persons, on the left-hand side and a constant

term and 28 variables on the right-hand side. The right-hand side includes a CPI-adjusted air

ticket price specific to the CSA-ORG pair, income, temperature index, and rainfall index specific

to the origin region, a fire index specific to the destination CSA, three hurricane dummy

variables, advertising expenditures, six 9-11 terrorist attack dummy variables, eleven monthly

dummy variables, and the dependent variable lagged four and twelve months.

Comparison of the Estimation Alternatives

The (1 0) coefficient estimates were analyzed to determine which of the three

approaches yields more stable coefficients. Recall that the speed of adjustment parameter (0)

must lie between zero and one, equivalent to state that the (1 0) coefficient must lie between

zero and one, to achieve stability. Results show that the SUR-AR1-ALL estimation yields the

fewest unstable coefficients among the four estimation alternatives. The estimate from the CSA-

ORG pair 14-1004 has a negative sign, but it is not significant. Three pairs exhibit unstable

(1 0) coefficient estimates using the SUR-AR1-CSA estimation, but none were significantly

different from zero. The SUR-AR1-ORG yielded similar results. Table 5-1 shows the (1 0)

coefficient estimates and their corresponding t-value for each CSA-ORG pair under each

estimation procedure.

Coefficient estimates from the SUR-AR1-ALL are shown in the following section. Results

are presented in sub-sections according to each destination CSA.









Results for Demand for Passengers by CSA

This section presents results of the estimated model developed in Chapter 4 for Florida and

each destination CSA using the SUR-AR1-ALL estimation approach. A copy of the TSPTM

program is illustrated in Appendix C. Note that the SUR-AR1-ALL model was estimated

allowing the auto-correlation coefficient (p) to differ across CSA-ORG pairs following the

specification adopted by Maekawa and Hisamatsu (2002) and Greene (2003).

Each sub-section includes a discussion of the results and a table with the coefficient

estimates and corresponding t-values, R-squared, and number of observations. Elasticity of

adjustment coefficient estimates and other explanatory variables (e.g., airfare, income, dummies

for storms and terror) from the static component will be discussed. Note that estimates for the

airfare, income, advertising, temperature, precipitation, and fire variables can be interpreted as

short run elasticities since the estimate (rTi) is the multiplication of 0 and /? .

Long run elasticities can be calculated by dividing the estimate by the elasticity of

adjustment (0). Long run elasticities yield greater /fi, which is in line with demand theory.

Passengers have a greater ability to respond to changes in the demand drivers in the long run

than in the short run. Due to asymmetry in information and relatively inflexible budget

allocations, it takes time before changes affect demand (Syriopoulus 1995).

Estimates of the dummies controlling for monthly seasonality are not discussed in this

chapter. But results suggest that there are strong seasonal patterns in each CSA-ORG pair and

therefore the model required the inclusion of these dummy variables. Since inference of these

coefficient estimates only allows a comparison against January, little insight can be obtained.

Chapter 6 includes a section that discusses seasonal patterns of demand for air passengers

traveling to Florida.









Since there is a strong theoretical expectation of the sign of airfare (negative), income

(positive), fire (negative), dummies for storms (negative), advertising (positive), and dummies

for terror (negative), their calculated t-values were compared against a critical t-value of |1.6581

which represents a 95% confidence on a one-tailed t-distribution. If the t-value (in absolute

value) of the coefficient is greater than |1.6581, the null hypothesis would be rejected. Calculated

t-values of the constant term, temperature, precipitation, and dummies for monthly seasonality

were compared against a critical t-value of |1.961 which represents a 95% confidence on a two-

tailed t-distribution.

The validity of the final model, SUR-AR1-ALL, was checked by analyzing the residuals of

each set of equations. Auto-correlation and partial correlation coefficients of the residuals with

several time lags were examined. Results indicated that residuals were random (no signs of any

other auto-correlation scheme).

One can let X change today and calculate how long the effect takes to be fully realized.

The following sections discuss the interpretation of the rigidity coefficient (1129) slightly

different. The (1 0) coefficient (i.e., 1129) tells whether complete rigidity ( 0 = 0), partial

adjustment (0 < 0 < 1), or complete adjustment (0 = 1) occurs. If complete rigidity is present,

the model could be re-estimated as a pure ARIMA and determine the significance of each

coefficient associated to the lagged dependent variables. Given the results obtained from the

model estimated, this step was not necessary because no CSA-ORG pair showed complete

rigidity. Therefore, the following discussion centers in short and long run changes in demand

given a change in explanatory variables.









Florida-Results

Table 5-2 presents the coefficient estimates and their corresponding t-values for the

demand for air passengers from four U.S regions traveling to Florida using the SUR-AR1-ALL

approach. Elasticity of adjustment estimates are less than one and significant at 95% confidence

in three U.S. regions traveling to the Florida CSA. These results suggest that current demand for

air passengers traveling to The Florida CSA is driven by both current and past events.

Analysis of coefficients in dynamic component

The elasticity of adjustment estimate from the Northeast region suggests that for a one

percentage increase in past demand, current demand for air passengers on average increases by

0.35%. Almost two thirds of the adjustment occurs in the current period, while events that

occurred in past periods affect current demand by 0.35. Similarly, the elasticity of adjustment

estimate of the other three U.S. regions can be analyzed. Results suggest that similar to the

Northeast region, demand from the West region adjusts somewhat immediately to events in the

current period (0.64). The South region adjusts more rapidly to current events (0.79) than to past

events. The remaining 0.21 is attributed to past events. The coefficient for the Midwest region

was significant given a one-tailed test.

Analysis of coefficients in static component

Within the economic variables, airfare estimates have the correct sign in all U.S. regions,

but they are only significant in the Northeast and South regions. For example, the airfare

elasticity of demand is -0.249 for the Northeast region and can be interpreted as follows: a

percentage increase in the price of airline tickets from the Northeast region to The Florida CSA

represents a 0.249% decrease in demand for air passenger traffic.

Income estimates have a positive sign and are significant in all U.S. regions. The short-

term income elasticities of demand are 1.011, 1.420, 1.008, and 1.253 for the Northeast region,









Midwest, South, and West region, respectively. Note that income has a similar effect (-1%

increase) on the demand for air passengers from every region. Coefficient estimates for

advertising expenditures have the correct sign but they are all not significant. In review, only

income estimates are significant across all U.S. regions, while airfare is significant in two U.S.

regions (Northeast and South).

Temperature and precipitation, which are weather variables related to the origin, are not

significant, while some of the weather variables related to The Florida CSA are significant. The

HCATHGH, HCATMED, and FIRE estimates have the correct sign (negative) and are

significant in most cases. However, HCATLOW behaves as expected (i.e., estimates have

negative sign) in the West region only. Estimates from the Northeast, South, and West regions

suggest that tropical storms and depressions have no effect on demand. Hurricanes have a

negative impact on demand from all four U.S. regions, at a significant level. Only the estimate

for the HCATMED from the West region is not significant. Wildfires also have a negative

impact on demand, but only the estimate from the South region is significant.

As expected, dummies for terror (TER2-TER7) have a negative effect on demand from all

U.S. regions. Although the effect is still negative, coefficient estimates suggest that the impact is

reduced as time increases. Meanwhile, the effect of the 9-11 terrorist attacks on demand from the

Northeast region is negative but since September 2003 is no longer significant. The effect of the

terrorism dummies on demand for air passengers traveling to The Florida CSA is presented in

Figure 5-1. Note that the gray-colored cylinders indicate that the estimate is not significant with

95% confidence.

South Florida CSA-Results

Coefficient estimates and their corresponding t-values of the SUR-AR1-ALL model for

four U.S regions traveling to the South Florida CSA are presented in Table 5-3. The stability









condition holds since elasticity of adjustment estimates are less than one. In addition, all

coefficients are significant in the four U.S. regions traveling to the South Florida CSA. These

results suggest that current demand to the South Florida CSA depends on both current and past

events.

Analysis of coefficients in dynamic component

The elasticity of adjustment estimate from the Northeast region suggests that for a one

percentage change in past demand, current demand for air passengers changes by 0.24. In other

words, almost three fourths of the adjustment occurs in the current period, while the remaining

portion of the adjustment depends on events that occurred in past periods.

Results suggest that approximately 81% of the demand from the West region is driven by

current events and the remaining 19% depends on events from past periods. The Midwest and

South regions yielded higher estimates for the elasticity of adjustment than the Northeast and

Midwest regions. Both U.S. regions adjust more rapidly to current events (0.85 and 0.86,

respectively) than to past events.

Analysis of coefficients in static component

All estimates of the short-term elasticities of airfare, income and advertising at the current

period have correct sign and are significant in all U.S. regions. For example, the airfare elasticity

of demand is -0.194 for the Northeast region and can be interpreted as follows. A percentage

increase in the price of airline tickets from the Northeast region to the South Florida CSA

represents a 0.194% decrease in demand for air passenger traffic. Income has a positive effect

and is significant in all four U.S. regions, while advertising expenditure coefficient estimates are

not significant in any region.

The weather variables related to Florida, HCATHGH, HCATMED, and FIRE, have correct

sign (negative) and are significant in most cases. However, HCATLOW behaves as expected in









the West region only. Nevertheless, estimates from all U.S. regions suggest that tropical storms

and depressions have no effect on demand. Hurricanes have a negative impact on demand from

all four U.S. regions, at a significant level. The HCATMED from the West region is the

exception. Wildfires also have a negative impact on demand from the Midwest, South, and West

regions. Demand from the Northeast region is not affected by wildfires in Florida.

Most of the dummies for terror (TER2-TER7) have a negative effect on demand from all

U.S. regions. But coefficient estimates suggest that the impact has been decreasing through time.

Meanwhile, the effect of the 9-11 terrorist attacks on demand from the Northeast and West

regions is negative but since September 2003 is no longer significant. Figure 5-2 exhibits the

effect of the terrorism dummies on demand for air passengers traveling to the South Florida

CSA. Note that the gray-colored cylinders indicate that the estimate is not significant with 95%

confidence.

Orlando CSA-Results

Table 5-4 shows the coefficient estimates and their corresponding t-values, R-squared, and

number of observations of the SUR-AR1-ALL model for the four U.S regions traveling to the

Orlando CSA. Elasticity of adjustment estimates are less than one but only significant in the

Northeast and Midwest regions. These results suggest that current demands for air passengers

from the South and West regions traveling to the Orlando CSA do not respond to events that

occurred in past periods. In other words, their demand is in long run equilibrium because the

response is immediate.

Analysis of coefficients in dynamic component

The elasticity of adjustment estimate from the Northeast region suggests that for a one

percentage increase in past demand, current demand for air passengers on the average increases









by 0.41. Fifty-nine percent of the adjustment occurs in the current period, while the remaining

portion of the adjustment depends on events that occurred in the past.

Results suggest that approximately 79% of the demand from the Midwest region is driven

by current events and the remaining 21% depends on events from past periods. The West and

South regions yielded higher estimates for the elasticity of adjustment than the Northeast and

Midwest regions. Also, the estimate 129 from the West and South regions were statistically not

different from zero. In other words, both regions adjust instantly to current events.

Analysis of coefficients in static component

Short-term airfare elasticities of demand are not significant in any region traveling to the

Orlando CSA, while income elasticity estimates at the current period have the correct sign and

are significant in all U.S. regions. The income elasticity of demand is 2.364 for the West region

which suggests that a percentage increase in income from the West region represents a 2.4%

increase in demand for air passenger traffic to the Orlando CSA. All estimates of the advertising

expenditures have the wrong sign and were significant in three of the four U.S. regions.

Weather variables related to Florida, HCATHGH, HCATMED, and FIRE, have correct

sign (negative) and are significant in most cases. However, HCATLOW behaves as expected in

the West region only. Hurricanes of higher categories (i.e., HCATHGH) have a negative impact

on demand from all U.S. regions, while hurricanes of category 1 and 2 have a negative impact on

demand from the Midwest region only. Estimates for FIRE indicate that wildfires affecting

Florida have an impact on demand from the South region only.

All dummies for terror (TER2-TER7) have a negative and significant effect on demand

from all U.S. regions. But coefficient estimates suggest that the impact has been decreasing

through time. The effect of the terrorism dummies on demand for air passengers traveling to the









Orlando CSA is shown in Figure 5-3. Note that all estimates are significant with 95%

confidence.

Tampa-St. Petersburg CSA-Results

Coefficient estimates and their corresponding t-values, R-squared, and number of

observations of the SUR-AR1-ALL model for four U.S regions traveling to the Tampa-St.

Petersburg CSA are presented in Table 5-5. Elasticity of adjustment estimates are less than one

but only one of them is significant. These results suggest that current demand for air passengers

from all U.S. regions but the Northeast region do not depend on events that occurred in past

periods. In other words, there is no rigidity in the demand for air passengers from the Midwest,

South, and West regions.

Analysis of coefficients in dynamic component

The elasticity of adjustment estimate is not significant in three U.S. regions indicating that

demand to the Tampa-St. Petersburg CSA is solely determined by current events. Consumers

traveling from the Midwest, South, and West regions instantly adjust to any changes in the

drivers of demand for air passengers to the Tampa-St. Petersburg CSA. Conversely, demand for

air passengers from the Northeast region responds to past events. The elasticity of adjustment

estimate from the Northeast region suggests that for a one percentage increase in past demand,

current demand for air passengers on the average increases by 0.16%. In other words, eighty-four

percent of the adjustment occurs in the current period, while the remaining portion of the

adjustment depends on past events.

Analysis of coefficients in static component

Estimates of the short-term income elasticities of demand at the current period have correct

sign and are significant in each of the four U.S. regions. Short-term airfare elasticities of demand

are significant in two of the four U.S. regions. For example, the estimate of the airfare elasticity









of demand for the South region is -0.159 suggesting that a percentage increase in the average

price of an airline ticket from the South region represents a 0.159% decrease in demand for air

passenger traffic to the Tampa-St. Petersburg CSA. Most of the estimates of the advertising

expenditures have incorrect sign and one of them is significant.

Weather variables related to Florida, HCATHGH, HCATMED, and FIRE, have the correct

sign (negative) and are significant in some U.S. regions. However, HCATLOW behaves as

expected in the West region only. Hurricanes of higher category have a negative impact on

demand from all U.S. regions but only estimates from the South and West regions were

significant. The Midwest region yielded significant values for hurricanes of lower categories,

while estimates of the tropical storm variables related to the Northeast region were not

significant. Estimates for FIRE indicate that wildfires affecting Florida do have a negative

impact on demand from any region, but it is only significant in the South region.

Similarly to the Orlando CSA, virtually all dummies for terror (TER2-TER7) have a

negative and significant impact on demand from every region to the Tampa-St. Petersburg CSA.

Figure 5-4 presents the effect of the terrorism dummies on demand for air passengers traveling to

the Tampa-St. Petersburg CSA.

Jacksonville CSA-Results

Table 5-6 shows coefficient estimates and their corresponding t-values, R-square, and

number of observations of the SUR-AR1-ALL model for four U.S regions traveling to the

Jacksonville CSA. Elasticity of adjustment estimates are less than one in all four U.S. regions.

Results in the Midwest and South regions suggest that the current demand for air passengers

traveling to the Jacksonville CSA depend on both current and past events. Elasticity of

adjustment estimates from the Northeast and West regions are not significant, indicating that

demand from these U.S. regions does not depend on events that occurred in past periods.









Analysis of coefficients in dynamic component

The elasticity of adjustment estimate from the Midwest region is statistically different from

zero indicating that demand to the Jacksonville CSA is determined by current and past events.

The elasticity of adjustment estimate suggests that for a one percentage increase in past demand,

current demand for air passengers on the average increases by 0.32%. In other words, nearly

two-thirds of the adjustment occurs in the current period, while the remaining portion of the

adjustment depends on events that occurred the past.

Results suggest that approximately 72% of the demand from the South region is driven by

current events and the remaining 28% depends on events from past periods. Conversely, the

estimate from the Northeast and West regions suggests that demand from these two U.S. regions

is not driven by past events.

Analysis of coefficients in static component

Estimates of the short-term elasticities of airfare, income, and advertising at the current

period have correct sign but are significant in some U.S. regions only. Estimates of the short-

term income elasticities of demand are significant in three of the four U.S. regions: Northeast,

Midwest, and South regions. Short-term price elasticities of demand are significant in the

Northeast and West regions only. For example, the estimate of the airfare elasticity of demand

for the West region is -1.177 suggesting that a percentage increase in the price of airline ticket

from the West region represents a 1.18% decrease in demand for air passenger traffic to the

Jacksonville CSA. Estimates of the advertising expenditures at the current period (ADV) are not

significant in any of the four U.S. regions.

Most of the estimates from the weather variables related to Florida have either incorrect

sign (positive) or are not significant. However, HCATMED from the Midwest region is negative

and significant suggesting that hurricanes of category 1, 2, and 3 have a negative impact on









demand to the Jacksonville CSA. Estimates for FIRE and HCATHGH from the South region are

both negative and significant suggesting that demand for air passengers from this U.S. region is

adversely affected by hurricanes of higher categories and wildfires.

Estimates of dummies for terror (TER2-TER7) have a negative and significant impact on

the demand from three of the four U.S. regions. Although most of the estimates from the West

region are negative, they are not significant. The effect of the terrorism dummies on demand for

air passengers traveling to the Jacksonville CSA is presented in Figure 5-5. Note that all

coefficients from the West region (gray-colored cylinders) indicate that estimates are not

significant with 95% confidence.

Fort Myers CSA-Results

Coefficient estimates and their corresponding t-values, R-square, and number of

observations of the SUR-AR1-ALL model for the four U.S regions traveling to the Fort Myers

CSA are exhibited in Table 5-7. Elasticity of adjustment estimates are less than one in three of

the four U.S. regions. The coefficient estimate from the West region is negative and hence,

unstable. Nevertheless, the estimate is not statistically significant. The estimate from the

Northeast region exhibits a stable estimate at a significant level. These results suggest that

current demand for air passengers from the Northeast region traveling to the Fort Myers CSA

responds to events that occurred in past periods. Conversely, there is no habit persistence in the

demand for air passengers from the Midwest, South, and West regions.

Analysis of coefficients in dynamic component

The elasticity of adjustment estimate is not significant in three U.S. regions and indicates

that demand to the Fort Myers CSA is solely determined by current events. It also indicates that

consumers traveling from the Midwest, South, and West regions instantly adjust to any changes

in the drivers of demand for air passengers. Conversely, demand for air passengers from the









Northeast region responds to past events, especially to those events that occurred twelve months

ago. The elasticity of adjustment estimate from the Northeast region suggests that for a one

percentage increase in previous demand, current demand for air passengers increases by 0.27%.

Seventy-three percent of the adjustment occurs in the current period, while the remaining portion

of the adjustment depends on events that occurred in past periods.

Analysis of coefficients in static component

Estimates of the short-term airfare elasticities from the Northeast and West regions have a

negative impact on demand. Estimates for the advertising expenditures are not significant in

three U.S. regions, while the estimate for the advertising expenditures from the Northeast region

has the incorrect sign and it is significant. Estimates of the short-term elasticities of income are

significant in three of four U.S. regions: Northeast, Midwest, and South regions. The estimate of

the income elasticity of demand from the South region is 1.413 suggesting that a percentage

increase in personal disposable income from the South region represents a 1.41% increase in

demand for air passengers to the Fort Myers CSA.

Only a few estimates from the weather variables related to Florida have correct sign

(negative) and are significant. FIRE and HCATMED estimates from the Midwest region are

negative and significant suggesting that wildfires and hurricanes of category 1, 2, and 3 have a

negative impact on demand to the Fort Myers CSA. The HCATHGH estimates from the

Northeast and South regions are also negative suggesting that hurricanes of category 4 and 5

affect negatively demand for air passengers to the Fort Myers CSA. Wildfires also have a

negative effect on demand for air passengers from the South region.

Estimates of the dummies for terror (TER2-TER7) have a negative sign but only a few are

significant. Estimates from the West region are positive and significant which contradicts

theoretical expectations established earlier. The effect of the terrorism dummies on demand for









air passengers traveling to the Fort Myers CSA is presented in Figure 5-5. Note that the gray-

colored cylinders indicate that estimates are not significant with 95% confidence.

Chapter Summary

This chapter discussed the approach chosen to estimate the model presented in Equation 4-

13 of Chapter 4. The SUR-AR1-ALL estimation approach performed better than the other three

estimation alternatives since it yielded the fewest unstable 129 coefficient estimates. Results were

presented and discussed for each destination CSA and for Florida (CSA=33). The discussion

focused on statistical significance of various estimates such as the elasticity of adjustment

estimates and the economic, weather, and terrorism variables. Estimates of the dummies

controlling for monthly seasonality suggested that there are strong seasonal patterns in each

CSA-ORG pair and that the inclusion of these dummy variables was necessary. Results also

showed that the advertising expenditure estimates were not well-behaved in almost all CSA-

ORG pairs. Since advertising expenditures are highly seasonal, the dummy variables for monthly

seasonality may be capturing the effect of the advertising variable.











0.100


0.000


-0.100


-0.200


-0.300


-0.400 I
Northeast Midwest South West
U.S. Region


Figure 5-1. Florida CSA: effect of terrorism dummies on demand for airline passengers from
four U.S. regions. Note: Gray-colored cylinders indicate estimate not significant at
95% confidence.


0.100


0.000


-0.100


-0.200


-0.300


-0.400 '-
Northeast Midwest South West
U.S. Region
Figure 5-2. South Florida CSA: effect of terrorism dummies on demand for airline passengers
from four U.S. regions. Note: Gray-colored cylinders indicate estimate not significant
at 95% confidence.











0.100


0.000


-0.100


-0.200


-0.300


-0.400 I
Northeast Midwest South West
U.S. Region


Figure 5-3. Orlando CSA: effect of terrorism dummies on demand for airline passengers from
four U.S. regions.


0.100

0.000

-0.100

-0.200

-0.300

-0.400


-0.5001

-0.600
Northeast Midwest South West
U.S. Region
Figure 5-4. Tampa-St. Petersburg CSA: effect of terrorism dummies on demand for airline
passengers from four U.S. regions. Note: Gray-colored cylinders indicate estimate not
significant at 95% confidence.











0.100

0.000

c -0.100

W -0.200 -

-0.300

oe -0.400

-0.500 -

-0.600
Northeast Midwest South West
U.S. Region
Figure 5-5. Jacksonville CSA: effect of terrorism dummies on demand for airline passengers
from four U.S. regions.


4.000


3.000


2.000


1.000


0.000


-1.000


-2.000


H H


I I


{ 4 H .


Northeast Midwest South West
U.S. Region
Figure 5-6. Fort Myers CSA: effect of terrorism dummies on demand for airline passengers from
four U.S. regions.


-


-


-


-











Table 5-1. Coefficient estimates of 1129 and their corresponding t-value using five different
estimation approaches.
Estimation
Approach- OLS AR1 SUR-AR1-ALL SUR-AR1-CSA SUR-AR1-ORG


CSA-ORG Pair
1-1001
1-1002
1-1003
1-1004
5-1001
5-1002
5-1003
5-1004
8-1001
8-1002
8-1003
8-1004
9-1001
9-1002
9-1003
9-1004
14-1001
14-1002
14-1003
14-1004
Number of


0.134 1.037 0.114
-0.086 -0.446^ -0.096
-0.104 -0.739^ -0.066
0.050 0.387 0.093
0.194 1.299 0.188
0.210 1.421 0.133
-0.102 -0.678A 0.070
0.262 1.938* 0.188
-0.266 -1.969A -0.140
0.111 0.679 -0.038
0.015 0.142 0.046
0.124 0.819 0.161
0.029 0.221 0.040
-0.034 -0.239^ 0.274
0.068 0.533 0.151
0.104 0.785 0.115
0.197 1.651 0.287
0.032 0.265 -0.041
0.138 1.036 0.197
-0.246 -1.790A -0.135


HF


1.069
-0.925^A
-0.660A
0.925
1.782*
1.265
0.707
1.917*
-1.094^A
-0.342^A
0.492
1.534
0.349
2.475*
1.440
1.102
2.552*
-0.370A
1.933*


0.237
0.151
0.137
0.186
0.409
0.215
0.052
0.133
0.163
0.071
0.130
0.025
0.042
0.319
0.278
0.245
0.268
0.157
0.172


-1.302^ -0.211


3.667* 0.297
2.434* 0.146
2.490* 0.203
2.229* 0.248
5.558* 0.377
2.880* 0.168
0.750 0.078
1.510 0.397
2.278* 0.163
0.948 0.053
1.860* 0.192
0.206 -0.017
0.309 -0.183
3.559* 0.331
3.525* 0.246
1.919* 0.094
2.454* 0.142
1.344 0.002
1.811* 0.218
-1.609A -0.061


3.564* 0.087 1.070
1.815* 0.077 0.928
2.851* 0.065 0.902
2.707* 0.002 0.017
4.270* 0.372 4.106*
1.805* 0.187 1.988*
0.873 -0.027 -0.027^
3.895* -0.127 -1.164^
1.731* 0.090 1.001
0.536 0.017 0.197
2.250* 0.133 1.691*
-0.122^ 0.039 0.272
-1.292^ 0.005 0.035
3.117* 0.497 4.870*
2.634* 0.288 3.118*
0.660 0.272 1.978*
1.015 0.212 1.756*
0.011 0.058 0.489
1.804* 0.168 1.694*
-0.450A -0.085 -0.614^


Unstable 6 6 1
Estimates
* indicates significance at 95% confidence using a one-tailed test.
A indicates coefficient estimate is unstable (negative sign).










Table 5-2. Florida CSA: coefficient estimates and their corresponding t-values for the demand
for air passengers traveling from four U.S. regions using the SUR-AR1-ALL
approach.


Northeast (1001) Midwest (1002) South (1003) West (1004)


* indicates significance at 95% confidence using a two-tailed test.
A indicates significance at 95% confidence using a one-tailed test.


Origin -

Variable 1
C
FARE
INC
TEMP
PCP
FIRE
HCATLOW
HCATMED
HCATHGH
ADV
TER2
TER3
TER4
TER5
TER6
TER7
MTH2
MTH3
MTH4
MTH5
MTH6
MTH7
MTH8
MTH9
MTH10
MTH11
MTH12
H127


P
H29
p
0
W2
W4
Wi2
R2
N=146


Estimate T-value
-5.383 -2.601*
-0.249 -3.318^
1.011 4.465A
0.025 0.846
0.002 0.855
-0.004 -0.522
0.005 0.707
-0.024 -1.924^
-0.054 -2.972A
0.008 0.484
-0.278 -8.016^
-0.152 -3.523A
-0.070 -1.472
-0.081 -1.511
-0.074 -1.305
-0.089 -1.413
0.029 1.554
0.128 4.984*
0.056 2.002*
-0.090 -3.203*
-0.139 -4.440*
-0.063 -1.661*
-0.044 -1.157
-0.327 -8.917*
-0.097 -3.836*
-0.014 -0.420
0.019 0.677
0.08 1.542
0.06 1.157
0.35 4.267*
0.648 12.87*
0.651
0.224
0.163
0.613
0.963


Estimate
-8.130
-0.055
1.420
0.013
0.000
-0.010
0.013
-0.046
-0.044
0.001
-0.265
-0.325
-0.306
-0.344
-0.359
-0.324
0.089
0.276
0.015
-0.204
-0.185
-0.181
-0.257
-0.474
-0.155
-0.062
0.052
0.095
0.073
0.154
0.713
0.846
0.614
0.476
-0.090
0.963


T-value
-3.687*
-0.726
5.773A
0.742
0.034
-1.575
1.788A
-3.732^
-2.558A
0.037
-7.308A
-6.725A
-5.571A
-5.527A
-5.364A
-4.511^
4.196*
9.276*
0.457
-6.056*
-5.141*
-3.912*
-6.470*
-12.27*
-5.299*
-1.876
2.144*
1.955*
1.506
1.808A
16.19*


Estimate
-4.112
-0.246
1.008
0.058
-0.004
-0.013
0.007
-0.027
-0.051
0.014
-0.280
-0.173
-0.128
-0.170
-0.174
-0.171
0.028
0.166
0.056
0.021
0.014
0.010
-0.048
-0.216
-0.027
0.002
0.053
0.045
0.022
0.208
0.668
0.792
0.215
0.104
0.682
0.948


T-value
-2.513*
-3.852^
5.483A
1.805
-1.676
-2.243A
1.068
-2.533^
-3.271^
1.033
-9.311^
-4.531^
-3.131^
-3.643A
-3.570A
-3.181A
1.969*
8.262*
2.677*
0.982
0.624
0.332
-1.740
-7.813*
-1.203
0.087
2.612*
1.006
0.499
2.779*
14.50*


Estimate T-value
-8.904 -3.429*
-0.106 -1.090
1.253 4.324A
0.080 0.927
-0.002 -1.816
-0.007 -1.026
-0.005 -0.661
-0.011 -0.901
-0.053 -3.108A
0.007 0.428
-0.215 -6.133A
-0.167 -3.664A
-0.063 -1.252
-0.142 -2.557A
-0.074 -1.258
-0.078 -1.218
-0.084 -4.937*
0.054 2.629*
-0.068 -2.665*
-0.121 -4.216*
-0.076 -2.874*
-0.085 -2.463*
-0.090 -2.791*
-0.290 -8.233*
-0.158 -5.630*
-0.132 -4.505*
-0.025 -1.001
0.019 0.328
0.189 3.477*
0.363 3.733*
0.684 11.74*
0.637
0.053
0.521
0.426
0.970










Table 5-3. South Florida CSA: coefficient estimates and their corresponding t-values for the
demand for air passengers traveling from four U.S. regions using the SUR-AR1-ALL
approach.
Origin Northeast (1001) Midwest (1002) South (1003) West (1004)


* indicates significance at 95% confidence using a two-tailed test.
A indicates significance at 95% confidence using a one-tailed test.


Variable 1
C
FARE
INC
TEMP
PCP
FIRE
HCATLOW
HCATMED
HCATHGH
ADV
TER2
TER3
TER4
TER5
TER6
TER7
MTH2
MTH3
MTH4
MTH5
MTH6
MTH7
MTH8
MTH9
MTH10
MTH11
MTH12
1H27

H29
p
0
W2
W4
Wi2
R2
N=146


Estimate T-value
-3.574 -2.224*
-0.194 -3.032A
0.873 5.380A
0.021 0.633
0.002 0.665
-0.007 -0.847
0.010 1.137
-0.031 -2.154^
-0.058 -2.709^
-0.002 -0.650
-0.199 -6.385A
-0.134 -3.619^
-0.038 -0.947
-0.028 -0.618
-0.046 -0.959
-0.066 -1.224
0.002 0.109
0.096 3.882*
-0.004 -0.147
-0.193 -6.527*
-0.256 -8.009*
-0.128 -3.724*
-0.104 -3.143*
-0.461 -13.85*
-0.203 -7.968*
-0.053 -1.737*
0.030 1.231
0.110 2.654*
0.068 1.711^
0.237 3.667*
0.543 12.93*
0.763
0.463
0.286
0.251
0.942


Estimate
-4.960
-0.108
1.023
0.007
-0.005
-0.018
0.012
-0.064
-0.060
-0.001
-0.213
-0.290
-0.277
-0.343
-0.379
-0.367
0.038
0.193
-0.102
-0.351
-0.389
-0.372
-0.396
-0.638
-0.326
-0.125
0.059
0.099
0.124
0.151
0.639
0.849
0.657
0.820
-0.477
0.963


T-value
-2.811*
-2.403A
5.573^
0.401
-1.703
-2.430^
1.520
-4.881^
-3.162A
-0.218
-6.422A
-6.810A
-5.861A
-6.349A
-6.436A
-5.695A
2.053*
7.497*
-3.329*
-10.90*
-11.30*
-9.333*
-11.39*
-18.36*
-11.97*
-4.287*
2.836*
2.688*
3.420*
2.434*
16.69*


Estimate
-4.475
-0.148
1.023
0.065
-0.004
-0.017
0.011
-0.037
-0.052
0.003
-0.236
-0.208
-0.155
-0.192
-0.218
-0.202
0.001
0.151
0.033
-0.029
-0.052
-0.037
-0.095
-0.322
-0.105
-0.002
0.075
0.123
0.024
0.137
0.614
0.863
0.897
0.175
-0.073
0.927


T-value
-3.129*
-3.281^
6.808A
1.986*
-1.530
-2.583A
1.616
-3.226^
-3.046^
1.052
-8.567A
-6.082A
-4.173A
-4.608A
-4.904A
-4.082A
0.090
8.141*
1.553
-1.365
-2.307*
-1.461
-4.023*
-12.86*
-4.909*
-0.086
4.539*
3.647*
0.733
2.490*
17.05*


Estimate T-value
-7.344 -2.412*
-0.103 -1.116
1.137 3.651^
0.144 1.532
0.000 0.052
-0.014 -1.794A
-0.001 -0.147
-0.017 -1.237
-0.052 -2.598A
0.005 0.951
-0.139 -3.324A
-0.174 -3.052A
-0.034 -0.503
-0.095 -1.257
0.009 0.114
0.042 0.466
-0.126 -7.047*
0.040 1.712
-0.056 -1.942
-0.154 -4.894*
-0.093 -3.020*
-0.050 -1.433
-0.050 -1.512
-0.389 -11.14*
-0.241 -7.998*
-0.109 -3.705*
0.013 0.522
0.117 2.305*
0.121 2.400*
0.186 2.229*
0.758 15.44*
0.814
0.626
0.650
-0.276
0.949










Table 5-4. Orlando CSA: coefficient estimates and their corresponding t-values for the demand
for air passengers traveling from four U.S. regions using the SUR-AR1-ALL
approach.
Origin Northeast (1001) Midwest (1002) South (1003) West (1004)


* indicates significance at 95% confidence using a two-tailed test.
A indicates significance at 95% confidence using a one-tailed test.


Variable 1
C
FARE
INC
TEMP
PCP
FIRE
HCATLOW
HCATMED
HCATHGH
ADV
TER2
TER3
TER4
TER5
TER6
TER7
MTH2
MTH3
MTH4
MTH5
MTH6
MTH7
MTH8
MTH9
MTH10
MTH11
MTH12
1H27
n28
H29
p
0
W2
W4
Wi2
R2
N=146


Estimate T-value
-8.654 -4.050*
-0.095 -1.422
1.225 5.211^
0.037 1.106
0.002 0.771
0.002 0.207
0.006 0.742
-0.019 -1.408
-0.042 -2.097A
-0.013 -1.192
-0.269 -8.010^
-0.179 -4.119^
-0.099 -2.148A
-0.129 -2.593A
-0.124 -2.287A
-0.186 -3.007A
0.060 3.131*
0.181 7.074*
0.141 5.506*
0.027 1.077
-0.027 -0.976
0.057 1.682
0.057 1.712
-0.245 -7.057*
-0.010 -0.388
0.057 1.906
0.006 0.272
0.136 3.111*
0.048 1.149
0.409 5.558*
0.598 13.04*
0.591
0.333
0.118
0.550
0.957


Estimate
-10.509
0.007
1.543
0.003
0.002
-0.004
0.027
-0.031
-0.048
-0.029
-0.232
-0.314
-0.270
-0.305
-0.298
-0.294
0.079
0.225
0.074
-0.013
0.036
0.027
-0.101
-0.299
-0.010
-0.025
0.012
0.051
0.020
0.215
0.650
0.785
0.237
0.091
0.672
0.937


T-value
-4.890*
0.094
6.536*
0.149
0.507
-0.577
3.219A
-2.245^
-2.450^
-2.85 0^
-6.448A
-6.796A
-5.131^A
-5.217A
-4.723A
-4.324A
3.995*
9.260*
2.783*
-0.463
1.175
0.733
-2.999*
-8.417*
-0.339
-0.883
0.585
1.193
0.457
2.880*
15.96*


Estimate
-8.659
-0.064
1.464
0.056
-0.003
-0.013
0.010
-0.016
-0.040
-0.015
-0.278
-0.258
-0.206
-0.225
-0.211
-0.219
0.049
0.206
0.095
0.080
0.087
0.086
-0.037
-0.238
0.033
0.023
0.038
0.059
-0.003
0.052
0.701
0.948
1.123
-0.049
-0.074
0.926


T-value
-4.290*
-1.119
6.800A
1.473
-0.892
-1.897A
1.335
-1.304
-2.162A
-1.808A
-8.078A
-5.714A
-4.098A
-3.998A
-3.475^
-3.249A
3.068*
9.835*
4.265*
3.305*
3.403*
2.933*
-1.363
-8.202*
1.307
0.954
2.180*
1.509
-0.069
0.750
19.10*


Estimate T-value
-19.331 -5.850*
0.059 0.606
2.364 6.748A
-0.211 -2.060*
-0.004 -2.964*
0.002 0.220
0.000 -0.027
-0.004 -0.300
-0.049 -2.431^
-0.038 -2.941^
-0.292 -7.053A
-0.330 -5.965A
-0.252 -4.027A
-0.327 -4.728A
-0.309 -4.045A
-0.356 -4.209A
-0.069 -3.877*
0.105 4.850*
-0.036 -1.419
-0.069 -2.408*
-0.031 -1.089
-0.066 -1.999*
-0.132 -4.147*
-0.321 -8.984*
-0.113 -3.925*
-0.147 -5.468*
-0.042 -1.990*
0.004 0.086
0.139 2.909*
0.133 1.510
0.700 14.78*
0.867
0.034
1.046
-0.079
0.953










Table 5-5. Tampa-St. Petersburg CSA: coefficient estimates and their corresponding t-values for
the demand for air passengers traveling from four U.S. regions using the SUR-AR1-
ALL approach.
Origin Northeast (1001) Midwest (1002) South (1003) West (1004)


* indicates significance at 95% confidence using a two-tailed test.
A indicates significance at 95% confidence using a one-tailed test.


Variable 1 Estimate T-value
C -14.351 -7.455*
FARE -0.054 -0.983
INC 1.856 8.917A
TEMP 0.024 0.739
PCP 0.001 0.344
FIRE -0.008 -0.969
HCATLOW 0.011 1.301
HCATMED -0.015 -1.086
HCATHGH -0.031 -1.510
ADV -0.008 -1.337
TER2 -0.202 -6.576A
TER3 -0.179 -4.777A
TER4 -0.065 -1.629
TER5 -0.071 -1.594
TER6 -0.068 -1.434
TER7 -0.119 -2.178A
MTH2 0.090 4.935*
MTH3 0.257 10.29*
MTH4 0.162 6.541*
MTH5 0.000 -0.010
MTH6 -0.026 -0.960
MTH7 0.092 2.882*
MTH8 0.070 2.330*
MTH9 -0.256 -8.756*
MTH10 -0.005 -0.187
MTH11 0.065 2.363*
MTH12 0.040 1.922
H27 0.146 3.469*
H28 -0.014 -0.355
H29 0.163 2.278*
p 0.544 12.81*
0 0.837
W2 0.897
W4 -0.088
Wi2 0.191
R2 0.970
N=146


Estimate
-12.970
-0.143
1.808
0.028
-0.002
-0.008
0.026
-0.028
-0.033
-0.008
-0.235
-0.355
-0.327
-0.356
-0.433
-0.409
0.149
0.384
0.121
-0.082
-0.043
-0.022
-0.137
-0.378
-0.077
0.003
0.115
0.095
0.009
0.071
0.594
0.929
1.333
0.124
-0.457
0.942


T-value
-6.204*
-2.113^
7.918A
1.150
-0.407
-0.875
2.806A
-1.886^
-1.485
-1.324
-6.380^
-7.598A
-6.231^
-6.182A
-6.973A
-5.895^
7.110*
13.25*
4.017*
-2.638*
-1.315
-0.545
-4.122*
-11.26*
-2.707*
0.097
5.426*
2.247*
0.220
0.948
14.14*


Estimate
-9.928
-0.159
1.497
0.041
-0.008
-0.013
0.010
-0.018
-0.032
0.001
-0.262
-0.221
-0.192
-0.210
-0.275
-0.285
0.047
0.228
0.094
0.039
0.032
0.032
-0.037
-0.196
0.007
0.039
0.078
0.088
0.012
0.130
0.636
0.870
0.678
0.091
0.231
0.943


T-value
-6.149*
-2.989A
8.420A
0.969
-2.438*
-1.979A
1.359
-1.568
-1.843A
0.207
-8.671^
-5.809A
-4.708A
-4.595A
-5.667^
-5.215^A
3.185*
10.57*
4.378*
1.719
1.359
1.184
-1.537
-7.939*
0.323
1.772
4.881*
2.118*
0.302
1.860^
14.44*


Estimate T-value
-32.606 -6.732*
0.107 0.815
3.608 7.063^
0.229 1.544
-0.004 -2.236*
-0.014 -1.365
-0.003 -0.276
-0.014 -0.783
-0.045 -1.762A
-0.030 -2.412^
-0.157 -3.293A
-0.164 -2.755A
-0.101 -1.581
-0.136 -1.925^
-0.173 -2.238A
-0.214 -2.372A
-0.023 -0.980
0.200 6.455*
0.005 0.154
-0.075 -1.868
-0.045 -1.200
-0.025 -0.570
-0.095 -2.569*
-0.337 -8.055*
-0.152 -4.238*
-0.142 -3.844*
-0.007 -0.290
-0.006 -0.078
0.116 1.817^
0.025 0.206
0.637 10.83*
0.975
-0.223
4.581
-3.359
0.976










Table 5-6. Jacksonville CSA: coefficient estimates and their corresponding t-values for the
demand for air passengers traveling from four U.S. regions using the SUR-AR1-ALL
approach.
Origin Northeast (1001) Midwest (1002) South (1003) West (1004)


Variable 1
C
FARE
INC
TEMP
PCP
FIRE
HCATLOW
HCATMED
HCATHGH
ADV
TER2
TER3
TER4
TER5
TER6
TER7
MTH2
MTH3
MTH4
MTH5
MTH6
MTH7
MTH8
MTH9
MTH10
MTH11
MTH12
1H27
n28
H29
p
0
W2
W4
Wi2
R2
N=146


Estimate T-value
-15.938 -2.697*
-0.311 -2.427A
1.871 3.116A
0.132 2.196*
0.000 -0.059
-0.008 -0.827
0.023 2.201^
0.002 0.095
-0.010 -0.387
0.034 1.643
-0.323 -5.342A
-0.380 -4.353A
-0.270 -2.558A
-0.179 -1.488
-0.090 -0.657
-0.015 -0.095
0.119 4.954*
0.319 8.375*
0.292 7.498*
0.227 6.037*
0.208 5.206*
0.234 4.723*
0.250 4.969*
-0.017 -0.381
0.185 4.515*
0.194 5.350*
0.215 6.109*
-0.012 -0.193
0.016 0.256
0.042 0.309
0.826 16.34*
0.958
-0.294
0.375
0.919
0.947


* indicates significance at 95% confidence using a two-tailed test.
A indicates significance at 95% confidence using a one-tailed test.


Estimate
-9.923
-0.044
1.211
0.034
-0.009
-0.006
0.033
-0.038
-0.023
0.026
-0.277
-0.337
-0.408
-0.428
-0.346
-0.324
0.105
0.378
0.257
0.179
0.207
0.238
0.167
-0.049
0.180
0.190
0.194
0.192
0.085
0.319
0.555
0.681
0.602
0.266
0.132
0.917


T-value
-3.783*
-0.489
4.283A
1.058
-1.662
-0.662
3.202A
-2.230^
-0.947
1.486
-6.557A
-6.080A
-6.316A
-5.457^
-4.482A
-4.086A
4.922*
11.37*
8.405*
5.786*
6.172*
5.882*
4.362*
-1.360
5.109*
5.502*
6.384*
3.376*
1.578*
3.559*
9.740*


Estimate
-13.188
-0.019
1.657
0.059
-0.001
-0.017
0.016
-0.014
-0.030
0.014
-0.300
-0.253
-0.197
-0.255
-0.293
-0.321
0.028
0.228
0.163
0.149
0.147
0.165
0.109
-0.064
0.101
0.101
0.113
0.103
0.077
0.278
0.596
0.722
0.370
0.276
0.354
0.958


T-value
-6.798*
-0.308
7.502^
1.302
-0.415
-2.526A
2.177A
-1.179
-1.667A
1.284
-9.914^
-6.349A
-4.649A
-5.274A
-5.620^
-5.610^
1.902
10.27*
7.443*
6.495*
5.964*
5.713*
3.965*
-2.400*
4.009*
4.283*
5.532*
2.216*
1.796A
3.525*
12.97*


Estimate T-value
-13.900 -0.983
-1.177 -1.792^
1.421 1.013
0.099 0.123
0.008 0.818
0.071 1.506
0.015 0.294
0.072 0.901
-0.037 -0.318
0.101 0.996
-0.117 -0.592
-0.190 -0.805
-0.352 -1.361
-0.185 -0.641
-0.173 -0.561
0.225 0.663
-0.086 -1.023
0.159 1.485
-0.151 -1.217
-0.079 -0.570
0.048 0.311
0.109 0.646
0.104 0.585
-0.439 -2.392*
-0.495 -3.260*
-0.368 -2.806*
-0.074 -0.550
0.125 1.359
0.204 2.594*
0.245 1.919^
0.585 7.589*
0.755
0.510
0.830
-0.340
0.840










Table 5-7. Fort Myers CSA: coefficient estimates and their corresponding t-values for the
demand for air passengers traveling from four U.S. regions using the SUR-AR1-ALL
approach.


Northeast (1001) Midwest (1002) South (1003) West (1004)


* indicates significance at 95% confidence using a two-tailed test.
A indicates significance at 95% confidence using a one-tailed test.
9 indicates unstable coefficient.


Origin -

Variable 1
C
FARE
INC
TEMP
PCP
FIRE
HCATLOW
HCATMED
HCATHGH
ADV
TER2
TER3
TER4
TER5
TER6
TER7
MTH2
MTH3
MTH4
MTH5
MTH6
MTH7
MTH8
MTH9
MTH10
MTH11
MTH12
1H27


P
H29
p
0
W2
W4
Wi2
R2
N=146


Estimate T-value
-13.315 -3.948*
-0.402 -3.120^
1.692 4.827A
-0.025 -0.372
-0.002 -0.358
-0.007 -0.625
0.012 0.979
0.016 0.792
-0.076 -2.620^
-0.044 -2.054A
-0.317 -5.835A
-0.159 -2.419^
0.034 0.461
0.135 1.544
0.103 1.057
0.095 0.880
0.050 1.527
0.163 3.392*
0.022 0.396
-0.297 -4.553*
-0.436 -6.164*
-0.493 -6.116*
-0.508 -6.634*
-0.768 -10.48*
-0.354 -7.610*
-0.248 -4.726*
-0.171 -4.334*
-0.047 -0.729
0.083 1.425
0.268 2.454*
0.653 11.36*
0.732
-0.174
0.311
0.863
0.978


Estimate
-12.005
0.139
1.604
-0.035
-0.006
-0.020
0.024
-0.045
0.012
0.012
-0.210
-0.224
-0.182
-0.108
-0.144
-0.102
0.145
0.359
-0.019
-0.599
-0.682
-0.628
-0.720
-0.919
-0.421
-0.144
-0.001
0.128
0.024
0.157
0.495
0.843
0.817
0.156
0.027
0.986


T-value
-5.016*
1.279
5.668^
-1.008
-0.911
-2.009A
2.336^A
-2.536^
0.479
0.665
-5.492A
-4.796A
-3.568A
-2.006A
-2.607A
-1.653
3.507*
5.868*
-0.268
-6.772*
-7.437*
-5.954*
-7.610*
-10.63*
-7.072*
-2.270*
-0.038
1.832A
0.384
1.344
7.146*


Estimate
-10.596
0.039
1.413
-0.003
0.000
-0.014
-0.004
-0.014
-0.041
0.018
-0.168
-0.072
-0.037
-0.038
-0.054
-0.068
0.106
0.258
0.057
-0.129
-0.206
-0.189
-0.279
-0.430
-0.119
-0.059
0.009
0.044
0.017
0.172
0.698
0.828
0.256
0.099
0.644
0.972


T-value
-4.083*
0.489
5.145A
-0.053
0.070
-1.711^
-0.496
-0.959
-1.857A
1.364
-4.113^
-1.316
-0.590
-0.545
-0.738
-0.828
4.218*
7.661*
1.442
-2.766*
-4.079*
-3.222*
-5.347*
-8.656*
-3.287*
-1.525
0.382
0.748
0.308
1.811^
13.88*


Estimate T-value
38.241 1.432
-2.508 -2.265A
-3.911 -1.478
-0.588 -0.423
0.002 0.091
0.105 1.229
0.062 0.669
-0.049 -0.334
-0.047 -0.225
0.208 1.177
0.437 1.108
1.640 3.183A
2.662 4.481^
3.207 4.786A
3.415 4.730A
3.714 4.652A
0.016 0.100
0.369 1.756
-0.161 -0.623
-0.641 -2.290*
-0.941 -2.987*
-1.118 -3.069*
-0.958 -2.658*
-1.731 -4.849*
-0.963 -3.624*
-0.248 -1.032
0.356 1.530
-0.020 -0.218
-0.061 -0.766
-0.211 -1.6091
0.668 8.557*
1.211
0.096
0.288
0.616
0.893









CHAPTER 6
SIMULATION ANALYSIS

Coefficient estimates for each CSA-ORG pair presented in Chapter 5 were used to conduct

a simulation analysis on the demand for airline passengers traveling to Florida. The present

chapter includes seven sections. The first section discusses the simulation analysis and identifies

the explanatory variables used in the analysis. Sections two to six present results of the

simulation analysis for income, airline ticket prices, terror, hurricanes, and monthly seasonality,

respectively. The seventh section discusses simulations performed on the fire, rainfall, and

temperature variables. Results from the simulation analysis determine how sensitive the demand

for air passengers is to changes in a specific variable while all others are set to their actual

values.

Also, the analysis simulated demand for air passengers in the absence of the 9-11 terrorist

attacks. Then, this simulated demand is compared to the one that accounts for the terrorist attack.

The difference between the two shows the impact of the terrorist attack on demand for air

passengers to Florida. Finally, the simulations were conducted to identify seasonal patterns

exhibited by the demand for air travel to Florida. These inferences are a useful tool to draw

recommendations for policymakers in the travel industry across the state.

Introduction

The simulation analysis was conducted to measure the magnitude and the speed of the

response of the dependent variable to specific changes on one explanatory variable, with all other

variables set at their actual values. All measures of the response are relative to the monthly

average number of passengers per 100,000 over the period between 1996 and 2006.

The expected average number of passengers for each of the 24 CSA-ORG pairs was

calculated using coefficient estimates presented in Chapter 5. It is the average number of









passengers calculated with all continuous explanatory variables set to their actual values during

the 11-year period, the terror dummy variables set in the presence of the 9-11 attack, and the

monthly seasonality dummy variables set in the presence of seasonal variation.

Total number of passengers traveling from each U.S. region can be calculated by

multiplying the average number of passengers by the population factor for each region:

Northeast-540 (e.g., 54,000,000 people), Midwest-528, South-859, and West-649. For example,

the average number of passengers per 100,000 from the Florida (33)-Northeast (1001) pair under

average conditions is 1,955. Total number of passengers for the entire 11-year period is

1,955x540x12 months 11 years or 139 million passengers. On average, approximately 12.6

million passengers traveled from the Northeast region each year since 1996. This estimate is

consistent with values presented in Figure 3-2 of Chapter 3.

Note that the population factor is the average for the whole 11 year period. Appendix E

presents monthly population factors during the period between 1996 and 2006 for each U.S.

region. Since the simulations for the terror variables are presented by year and the seasonality

variables by month, these factors can be used to calculate total passengers traveling from a

particular region for a given year or month.

Eight sets of simulations for eight explanatory variables were conducted to calculate the

overall average, average by month, and average by year. The set of simulations for each variable

was set to change at different levels and were defined as follows:

Income: 5% decline to 5% increase in per capital personal disposable income, in
increments of 1%

Airline ticket prices: 20% decrease to 20% increase in airfares, in increments of 5%

Terror: absence of 9-11 terrorist attack

Hurricanes: absence of hurricanes affecting Florida









Wildfires: absence of wildfires affecting Florida

Rainfall: 20% decrease to 20% increase in rainfall from a region, in increments of 5%

Temperature: 20% decrease to 20% increase in temperatures from a region, in increments
of 5%

Seasonality: average number of passengers by month

The next five sections discuss results from each set of simulations performed to each CSA-

ORG pair. Results of the simulations for each variable are presented in separate sections and

each section discusses the effects of the changes by destination CSA. Note that the dependent

variable represents number of passengers per 100,000 habitants in the origin U.S. region per

month unless otherwise noted. Note that some of the explanatory variables were not statistically

significant in some CSA-ORG pairs according to the model estimation presented in Chapter 5. If

the explanatory variable was not significant for a particular CSA-ORG pair, simulation analysis

has no relevance because the variable effect is not statistically significant. Inferences are only

applicable to those CSA-ORG pairs where the variable was statistically significant.

The magnitude of the response is discussed in each section as well as the speed of

adjustment. Note that even if a variable is statistically significant, inferences may only be valid

for the long run adjustments. If the speed of adjustment coefficient (0) is not statistically

significant different from one, there is no difference between short and long run adjustment

implying that the dependent variable adjusts immediately to changes in structural variables.

Thus, inferences were made about the differences between the short run and long run

adjustments only for those CSA-ORG pairs where the speed of adjustment coefficient (0) was

significantly different from one.

A final set of simulations was performed to identify the presence of seasonal patterns in the

demand for air transportation to Florida and the simulation results are presented following the









section on hurricanes. It is worthwhile to indicate that no simulations were performed using the

advertising variable. Results presented in Chapter 5 indicate that coefficient estimates for this

variable exhibit wrong sign and most of them were not significant.

Simulations for Income

The first set of simulations was conducted on the income variable which represents per

capital personal disposable income for a particular U.S. region (ORG): Northeast, Midwest,

South, and West regions. The income variable was set to fluctuate between a 5% decrease to a

5% increase in increments of 1%. Results presented in Table 5-2 to 5-7 in Chapter 5 showed that

the income variable was statistically significant in 22 of the 24 CSA-ORG pairs. Therefore,

simulation values are valid for all but two pairs.

Also note that income elasticities of demand for 22 CSA-ORG pairs were greater than one

suggesting that air transportation is not only a normal good but also a luxury good. That is,

demand for air transportation increases more rapidly than income. The following discussion

centers on the magnitude (measured in elasticities) and response speed (comparing short run

adjustment to full effect) across each of the six destination CSAs including Florida in relation to

each of the four originating U.S. regions.

Florida CSA-Income Simulations

Table 5-2 in Chapter 5 indicates that income was significant in each of the four U.S.

regions traveling to the Florida CSA. Therefore, the simulation results presented in this section

have statistical validity. Nevertheless, the speed of adjustment coefficient from the Midwest

region is not statistically different from one. Short run and long run simulated values from the

Midwest region are statistically equal and therefore, the response to changes in income occurs

instantaneously.









Only inferences about the differences between short and long run adjustments from the

other three U.S. regions are valid. The South region yielded 2,106 passengers per month during

the 11-year period covered under average conditions and the presence of the 9-11 terrorist attack,

followed by the Northeast (1,955 passengers), Midwest (1,335 passengers), and West (370

passengers) regions. The West region has the largest long run income elasticity of demand at

1.90, followed by the Midwest (1.67), Northeast (1.51), and South (1.25) regions. Figure 6-1

presents the relationship between different levels of per capital personal disposable income and

domestic demand for air passengers from four U.S. regions traveling to the Florida CSA.

Comparing income elasticities of demand, demand from the West region is more sensitive

to changes in income than any other U.S. region but its response is more rigid compared to the

other U.S. regions. In other words, the magnitude of the response is greater but it takes more

time to realize the full effect related to a change in income. For example, a 3% increase in

income represents a 5.8% increase in demand but only 65% of this increase is realized in the

short run.

The opposite occurs for the demand from the South region. The income elasticity of

demand from this U.S. region is the lowest, but its response in the short run is faster. A 3%

increase in income represents a 3.8% increase in demand and 80% of this increase is realized in

the short run. Short and long run changes in domestic demand for air passengers at different

levels of per capital personal disposable income from four U.S. regions traveling to the Florida

CSA are illustrated in Figure 6-2.











Florida (33)
Northeast (1001)
----------------

----------------

----------------

- - ---4


Short run*
-Long run

--------------
--------------


1,500

1,450

1,400

1,350

1,300

1,250

1,200

1,150

1,100


2,150

2,100

2,050

2,000

1,950

1,900

1,850

1,800

1,750

1,700

1,650


Florida (33)
Midwest (1002)


Sholi run
Long run


---------------- -------------

---------- --- ----------------


-\ O O ~ --- -
c r r 0


Florida (33)
South (1003)
---------------

---------------



---------------


Short run*
-Long run
---------------


Florida (33)
West (1004)
500 --------------


450---------------


400 ---------------


Short run*
Long run


2,050 -- -------------------- 350 -- ------ ----------------

S2,000 ----------- ---------------

1,950 ---------- ---- ----------300----
S1,900 -- --- --------
1,900------------------------------250 --------------- ----------------

1,850
1,80 -------------------------------200
1,800 --------------- --------------- 200 --------------- ----------------



PC-PDI (USD 000) PC-PDI (USD 000)

Figure 6-1. Florida CSA-Income simulations: relationship between different levels of personal
disposable income and domestic demand for air passengers from four U.S. regions.
Asterisk (*) denotes short run adjustment is statistically different from the long run
adjustment given a two-tailed test at 95% confidence.


~~0 0 0-3 r- 3 -c
--- -- -- -- --


2,300

2,250

2,200

2,150

S100


-------------


c--,--.--.--.-1--I--~--~--~--


I














200

150

100

50

0

-50

-100

-150

-200








200

150

100

50

0


Florida (33) E Short run*
Northeast (1001) Long run -


a-- --- --- --- ---^-i- --------- ----
























Florida (33) ]- Short run*
south-(1003)g ------------run
--------------------- --------



-- -----------------------------



-------------------------------------

-------------------------------------











--------------------------------------

------------------------- -------

------------------------- -


200

150

100

50

0

-50

-100

-150

-200








200

150

100

50

0


SI I 0

S 50 -------- -50


5 0 -5 0 --- - - - - - - - - -
-100 --1 --1------------------------------- -100 --------------------------------------

-150 ------------------------------------- -150 --------------------------------------

-200 ------------------------------------ -200 ------------------------------------




PC-PDI (USD 000) PC-PDI (USD 000)
Figure 6-2. Florida CSA-Income simulations: short and long run changes in domestic demand for
air passengers from four U.S. regions at different levels of per capital personal
disposable income. Asterisk (*) denotes short run adjustment is statistically different
from the long run adjustment given a two-tailed test at 95% confidence.


Florida (33) Short run
Midwest (1002) ---------- Long run

------------------------------- -------

--------------------------------------

------------------------------ -- -





















Florida (33) Short run*
est (100-4)- ----------------------------
--- ----------------------------------

--------------------------------------

--------------------------------------








-West (1004) --Lonrun


--------------------------------------



I - - - - - - - - - _









South Florida CSA-Income Simulations

Demand from the Northeast region amounted to 911 passengers per month during the 11-

year period covered under average conditions and the presence of the 9-11 terrorist attacks,

followed by the South (750), Midwest (392), and West (156) regions. Comparing income

elasticities of demand, the West region is more sensitive to changes in income than any other

U.S. region in both the short and long run, while the Northeast region is the least sensitive. A

percentage change in income leads to a 1.40% change in demand from the West region compared

to a 1.14% change in demand from the Northeast region in the long run. The long run income

elasticity of demand from the Midwest and South regions was 1.19 and 1.20, respectively. Figure

6-3 illustrates the relationship between different levels of per capital personal disposable income

and number of passengers for each of the four U.S. regions traveling to the South Florida CSA.

The speed of adjustment of all four U.S. regions was statistically different from one,

implying that the full effect of a change in income is not completely realized in the current

period. Part of the demand adjusts immediately and the remainder is realized in subsequent

periods. Within the differences between the short run and the long run adjustment to changes in

income, response from the Northeast region indicates that 77% is realized in the short run. For

example, a 3% increase in income yields a long run increase of 31 passengers per month (3.4%

increase) of which 24 are accounted as the short run or immediate increase in demand and the

remaining seven are related to the lagged effect from the increase in income. Faster responses

were found in the demands from the other three U.S. regions. The Midwest region shows a short

run response of 85%, followed by the South (86%) and West (81%) regions. Figure 6-4 shows

short and long run changes in domestic demand for air passengers at different levels of per capital

personal disposable income from four U.S. regions traveling to the South Florida CSA.














980

960

o 940

" 920

8 900

E 880

9 860

S840

820

800


South Florida (1)
Northeast (1001)

----------------*

----------------*

----------------*


Short run*
-Long run


--------------



-- ------------


00- ~r 0r 3
00-------m- --


South Florida (1)

820 South (1003)
820

800 ----------------

780 ----------------

760 ---------------

740 ---------

720 -- --------

700 ---------------


Short run*
--Long run


----------------

--------- ---


-- -----------

------ ----------

----------------

----------------


South Florida (1) SholT run*
Midwest (1002) Long run
420

410 ---------------- ------------ --

400 ---------------- --------

40390 --------------------------------
390 -- --- ---


380------- ------------------------

370 --------------- ---------------

360----------------- -----------------

350---------------- -------------

340---------------- -------------






South Florida (1) Short run*
West (1004) Long run
170 ----------------- -------------

165

160 ---------------------- ---------

155

150 ----- -------------------------

145 ----------------------------------

1n ----------------- -----------------
1A(


680 --------------- ---------------- 135 ----------------------------------

660 --------------- ---------------- 130 ----------------- -----------------



PC-PDI (USD 000) PC-PDI (USD 000)

Figure 6-3. South Florida CSA-Income simulations: relationship between different levels of
personal disposable income and number of passengers for each of the four U.S.
regions. Asterisk (*) denotes short run adjustment is statistically different from the
long run adjustment given a two-tailed test at 95% confidence.






202









South Florida (1) E Short run*
Northeast (1001) Long run


- i --- -- --- -- -

--- Q- Q-- Q-- O O O- --- Cl---



South Florida (1) El Short run*
South (1003) Long run





........... -1 i


South Florida (1) Short run*
Midwest (1002) Long run


















South Florida (1) Short run*
West (1004) Long run


U I U 1 1 1 1 1 1 1 1 1 1 1



S-40 ---------------------------------- -40------------------------------------

-60 -------------------------------- -60 ------------------------------------
4N xO xO xO N N 4N 4N OC 0 OC

PC-PDI (USD 000) PC-PDI (USD 000)
Figure 6-4. South Florida CSA-Income simulations: short and long run changes in the number of
passengers at different levels of per capital personal disposable income for each of the
four U.S. regions. Asterisk (*) denotes short run adjustment is statistically different
from the long run adjustment given a two-tailed test at 95% confidence.









Orlando CSA-Income Simulations

Demand from the Northeast region totaled 592 passengers per month during the 11-year

period covered under average conditions and the presence of the 9-11 terrorist attacks, followed

by the South (516), Midwest (451), and West (134) regions. In the long run, demand from the

West region is more sensitive to changes in income than any other U.S. region, while the South

region is the least sensitive. A percentage change in income leads to a 2.72% change in demand

from the West region compared to a 1.54% change in demand from the South region in the long

run. The long run income elasticity of demand from the Northeast and Midwest regions was 2.00

and 1.94, respectively. The relationship between different levels of per capital personal

disposable income and the number of passengers for each of the U.S. regions traveling to the

Orlando CSA is shown in Figure 6-5.

Within the differences between the short run and the long run adjustment to changes in

income, only the elasticities from the Northeast and Midwest regions can be compared. The

speed of adjustment coefficient from the South and West regions is not statistically different

from one. Thus, the response to changes in income occurs instantaneously.

Fifty-nine percent of the full response from the Northeast region is realized in the short

run. For example, a 3% increase in income yields a long run increase of 36 passengers per month

(6.1% increase) of which 22 are accounted as the short run or immediate increase in demand and

the remaining 14 are related to the lagged effect from the increase in income.

A quicker response was observed in the demand from the Midwest region. This U.S.

region yielded a short run response of 79%. Figure 6-6 illustrates short and long run changes in

domestic demand for air passengers at different levels of per capital personal disposable income

from four U.S. regions traveling to the Orlando CSA.









Orlando (5) Short run*
Northeast (1001) ==Long run


650 ---------------


600 --------------


550 -----


500---------------


----00 'N -- --
C r~ lr 1 =


Orlando (5)
South (1003)
---------------,


Short run
'Long run
---------------


Orlando (5)
West (1004)
160-----------

150---------------


- -i


Short run
Long run




----- ----------
------------- --


420 ------------------------------ 100 ------


PC-PDI (USD 000) PC-PDI (USD 000)
Figure 6-5. Orlando CSA-Income simulations: relationship between different levels of personal
disposable income and number of passengers for each of the four U.S. regions.
Asterisk (*) denotes short run adjustment is statistically different from the long run
adjustment given a two-tailed test at 95% confidence.


Orlando (5)
Midwest (1002)





- - -




- - - -


Short run*
Long run





------ ----



- - - - -


--------------


__ _ _ __ _I __ _ _ _
lcCrN r ri r01 =
n n rl r- r 0


- - - -


-------- ----


---------------


- - -












Orlando (5) n Short run*
Northeast (1001) m Long.run


i--------------------------------I


.------------------------------ -


----------------------- -


------------------------






a a a -- -----------------------------

S-----------------------------------
-----------------------------------
':r::: ::: .


Orlando (5) 0 Short run
south (1003) -_ __- Long- run









-------------P.-_


80


60


40


20


0


-20


-40


-60


-80


------------------------------------ -----------------------------------
i '- o a -- 1 o > 1- 1 -^ i- 0 r o '



PC-PDI (USD 000) PC-PDI (USD 000)

Figure 6-6. Orlando CSA-Income simulations: short and long run changes in the number of
passengers at different levels of per capital personal disposable income for each of the
four U.S. regions. Asterisk (*) denotes short run adjustment is statistically different
from the long run adjustment given a two-tailed test at 95% confidence.


Orlando (5) Short run*
Midwest (1002) Long run


-------------------------------------


----------------------------------- -




----------------------------


- -- ----------------------------


-- ----------------------------------


-------------------------------------


-------------------------------------







Orlando (5) Short run
West (1004) Long run










I-I- I-I-I-I-I-I-I-I-I









Tampa-St. Petersburg CSA-Income Simulations

The South region totaled the largest number of passengers per month (347) during the 11-

year period covered under actual conditions and the presence of the terrorist attacks of

September 11th, followed by the Northeast (284), Midwest (283), and West (57) regions.

Demand from the West region is more sensitive to changes in income with a long run elasticity

of 3.61, followed by the Northeast (2.21), Midwest (1.81), and South (1.50) regions. Figure 6-7

shows the relationship between different levels of per capital personal disposable income and the

number of passengers for each of the U.S. regions traveling to the Tampa-St. Petersburg CSA.

From the results in Table 5-5 in Chapter 5, only the speed of adjustment coefficient from

the Northeast region is statistically different. Therefore, only inferences about the short and long

run adjustments from this U.S. region are valid. For example, if there is a 3% increase in income,

the immediate increase in demand from the Northeast region is 16 passengers. The full effect is

an increase of 19 passengers, implying that 84% of the full response occurs in the short run.

Short run and long run levels of demand from the Midwest, South, and West regions are

statistically equal. Therefore, the response to changes in income is instantaneous; there is little

rigidity in the responses to changes in income. A 3% increase in income leads to an increase in

demand of 15 passengers in the Midwest region, 17 in the South region, and 7 in the West

region. Short and long run changes in domestic demand for air passengers at different levels of

per capital personal disposable income for each of the four U.S. regions traveling to the Tampa-

St. Petersburg CSA are presented in Figure 6-8.













340

320

300

280

260

240

220

200


Tampa-St. Pete (8) Short run*
Northeast (1001) -Long run

--------------- ----------------

--------------- ------- -------















S---------- ---------- -----
-- ------------ ----------------
--------------- - -

--------------- ----------------




C 00 0^ 0 0 0N



Tampa-St. Pete (8) Short run
South (1003) = Long run


200 ---------------J-------------- ------------------------------



PC-PDI (USD 000) PC-PDI (USD 000)

Figure 6-7. Tampa-St. Petersburg CSA-Income simulations: relationship between different levels
of personal disposable income and number of passengers for each of the four U.S.
regions. Asterisk (*) denotes short run adjustment is statistically different from the
long run adjustment given a two-tailed test at 95% confidence.


Tampa-St. Pete (8) Short run
Midwest (1002) Long run
340------------- --------------

320-------------------------------

300--------------- -----------

280------------------------

260

240 ------------- -----------

220 --------------- ----------------

200 ------------------ ----------------





Tampa-St. Pete (8) Short run
00 est (1004) Long run
100



80 --------------- ----------------



60 --------------- -- -----------


400 --------------



350



300 ---------------


---------------








Tampa-St. Pete (8) E Short run*
Northeast (1001) _._- _Long run






I WI










Tampa-St. Pete (8) B Short run
South (1003) Lo______ng run-
--p------------


Tampa-St. Pete (8) Short run
Midwest (1002) ----------- Long run

















Tampa-St. Pete (8) Short run
West (10--04)--___ Lo___ng run






- - - - -


-U -------------------------------- -Uf -


PC-PDI (USD 000) PC-PDI (USD 000)
Figure 6-8. Tampa-St. Petersburg CSA-Income simulations: short and long run changes in the
number of passengers at different levels of per capital personal disposable income for
each of the four U.S. regions. Asterisk (*) denotes short run adjustment is statistically
different from the long run adjustment given a two-tailed test at 95% confidence.









Jacksonville CSA-Income Simulations

The South region yielded the largest number of passengers per month (165) during the 11-

year period covered under average conditions and the presence of the terrorist attack of

September 11th', followed by the Midwest (49), Northeast (47), and West (3) regions. From

Table 5-6 in Chapter 5, the coefficient estimate of the income variable for the West region is not

statistically different from zero. Hence, no valid inferences can be made since income does not

affect demand from the West region. Also, the speed of adjustment coefficient from the

Northeast region is not statistically different from one. Therefore, differences between short and

long run adjustments from this U.S. region are statistically equal to zero. Demand responses to

changes in income from this U.S. region are still significant but occur immediately.

The long run elasticity of demand (2.27) from the South region was the largest compared

to the other two U.S. regions, followed by the Northeast (1.95) and Midwest (1.76) regions.

These results suggest that the South region is more sensitive to changes in income, while the

Midwest region is less sensitive. Figure 6-9 shows the relationship between different levels of

per capital personal disposable income and number of passengers for each of the U.S. regions

traveling to the Jacksonville CSA.

Demand from the South region responds more quickly to changes in income than the one

from the Midwest region. For example, the full effect of a 3% increase in income represents a

total increase in demand from the South region of 11 passengers from which 72% (8 passengers)

is realized in the short run. Meanwhile, the Midwest region yielded a response of 68% in the

short run. Short and long run changes in domestic demand for air passengers at different levels of

per capital personal disposable income from four U.S. regions traveling to the Jacksonville CSA

are presented in Figure 6-10.











Jacksonville (9)
Northeast (1001)
80---------------

70--------------

S60 ---------------

50 -------------

" 40- ------

= 30-------------

20-------------

10----------------

0 ---------------


Short run
- Long run


Jacksonville (9) Short run*
Midwest (1002) Long run
54--------------------------


52-------------------------


50---


If.-l-.I


44 ---------------------------


40 --------------- ----------------
"- ',0 1CN "1" 1C 0 tN qr r- o
rII rI rI II rI rI II rI


Jacksonville (9)
South (1003)


180 ---------------



160



140 ---------------



120 ---------------


Short run*
-Long run





----------


Jacksonville (9)
West (1004)
6 ----------------


Short run
S=-Long run


PC-PDI (USD 000)


PC-PDI (USD 000)


Figure 6-9. Jacksonville CSA-Income simulations: relationship between different levels of
personal disposable income and number of passengers for each of the four U.S.
regions. Asterisk (*) denotes short run adjustment is statistically different from the
long run adjustment given a two-tailed test at 95% confidence.


C r_ r_O


---------------










Jacksonville (9) E Short run
Northeast (1001) Long run












--- --> ---0 > 0 > 0 00-- ------ Cl-


- - - - - - - - - -






Jacksonville (9) n Short run*
South (1003) ____ Lonz run-




:---------


Jacksonville (9) Short run*
.Midwyest J}0 _ _ Long run











Midwest (1002) _____Long run












WesI (1004) -I--


-tU ------------------------------------ -v-U
^ i ----- -o i o -- --1 -\ -- -i -^ -. -r -o -- -
lc0 :;. 11Cll 0 e
en 4t 4t V* VN tN tN xO xO tN xO xO O NNN OC OC OC

PC-PDI (USD 000) PC-PDI (USD 000)
Figure 6-10. Jacksonville CSA-Income simulations: short and long run changes in the number of
passengers at different levels of per capital personal disposable income for each of the
four U.S. regions. Asterisk (*) denotes short run adjustment is statistically different
from the long run adjustment given a two-tailed test at 95% confidence.









Fort Myers CSA-Income Simulations

The Midwest region yielded the largest number of passengers per month (147) during the

11-year period covered under average conditions and the presence of the terrorist attacks of

September 11th', followed by the Northeast (105), South (99), and West (2) regions. Given the

results presented in Table 5-7 in Chapter 5, the coefficient estimate of the income variable for the

West region is not statistically different from zero. Hence, no valid inferences can be made since

income does not affect demand from the West region. Demand from the Northeast region is more

sensitive to changes in income than the other two U.S. regions. The income elasticity of demand

from the Northeast region is 2.26, followed by the Midwest (1.60) and South (1.41) regions.

Figure 6-11 presents the relationship between different levels of per capital personal disposable

income and number of passengers for each of the U.S. regions traveling to the Fort Myers CSA.

Only the Northeast region yielded a speed of adjustment coefficient statistically different

from one. Therefore, the difference between short and long run adjustments from this U.S. region

are statistically different and the response in demand from the Midwest and South regions to

changes in income is still significant but occurs immediately. Demand response from the

Midwest and South regions to the changes in income occurs in the same period, while the

response from the Northeast region to changes occurs in the short run and the full effect is

completed in the long run. For example, if there is a 3% increase in income, 73% of the response

in increased demand from the Northeast region occurs immediately. In other words, there is an

increase in demand of five passengers in the short run, while the full effect of an increase of

seven passengers is realized in subsequent periods. Short and long run changes in domestic

demand for air passengers at different levels of per capital personal disposable income from four

U.S. regions traveling to the Fort Myers CSA are illustrated in Figure 6-12.












Fort Myers (14) Short run*
Northeast (1001) -=Long run


Ou --------------- ----------------






Fort Myers (14) Short run
South (1003) --= Long run
110


105 ------------------------


100 --------------- ------------


95 ------ ----------------------


an


---------------------- ---------

-------------------- ------



----------------


80 ------------ ----------------




PC-PDI (USD 000) PC-PDI (USD 000)

Figure 6-11. Fort Myers CSA-Income simulations: relationship between different levels of
personal disposable income and number of passengers for each of the four U.S.
regions. Asterisk (*) denotes short run adjustment is statistically different from the
long run adjustment given a two-tailed test at 95% confidence.






214


Fort Myers (14) Short run
Midwest (1002) Long run






------ ---------------------
---------------m-----------------

---------------m-------------- -
- - - 1 - - -






















West (1004) -- Longrun
-----------------
--------------- -----------------
- - ~-- - - -

- - ~-- - - -










Fort Myers (14) En Short run*
Northeast (1001) Lonrun_.











------------ --- -- -- N^ 14 _- 1^





C0>>00 0C0lN


Fort Myers (14) B Short run
South (1003) Long.run _
-- - - - - -l -


Fort Myers (14) Short run
Midwest (1002) _______ _Long run





















Fort Myers (14) [ Short run
est (_1004)_ -- -_-L-ng. ru
- -- ---- ---- ---- ---- --- ----l'-- -


-----------------U ------------------ -U

Cl Cl Cl Cl Cl Cl Cl Cl Cl Cl C l C l C l C l C l C l C l C l C l C0 C0 0

PC-PDI (USD 000) PC-PDI (USD 000)
Figure 6-12. Fort-Myers CSA-Income simulations: short and long run changes in the number of
passengers at different levels of per capital personal disposable income for each of the
four U.S. regions. Asterisk (*) denotes short run adjustment is statistically different
from the long run adjustment given a two-tailed test at 95% confidence.




215









Simulations for Airline Ticket Prices

The second set of simulations was conducted on the airfare variable which represents

average airline ticket prices for a specific CSA-OG pair. Recall from Chapter 4 that airline ticket

prices were deflated by the CPI-airfare component to represent real airline ticket prices. The

airfare variable was set to fluctuate between a 20% decrease to a 20% increase in increments of

5%. Results presented in Table 5-2 to 5-7 in Chapter 5 showed that the airfare variable was

statistically significant in only 10 of the 24 CSA-ORG pairs. This section presents a discussion

of the Florida, South Florida, Tampa-St. Petersburg, Jacksonville, and Fort Myers CSAs. Only

those CSA-ORG pairs where the airfare variable is significant are illustrated and discussed. The

CSA-ORG pairs are the following:

* Florida (33)-Northeast (1001)
* Florida (33)-South (1003)
* South Florida (1)-Northeast (1001)
* South Florida (1)-Midwest (1002)
* South Florida (1)-South (1003)
* Tampa-St. Petersburg (8)-Midwest (1002)
* Tampa-St. Petersburg (8)-South (1003)
* Jacksonville (9)-Northeast (1001)
* Jacksonville (9)-Midwest (1002)
* Fort Myers (14)-Northeast (1001)


The Orlando CSA was not included in the discussion since results from Table 5-4 in

Chapter 5 indicated that the airfare variable was not significant in any of the four U.S. regions.

These results suggest that changes in airfare do not affect demand for air passengers to the

Orlando CSA.

Differences between short run and long run price elasticities of demand of 10 CSA-ORG

pairs are also presented and discussed. The following sections center on the magnitude









(measured in elasticities) and speed (comparing short run adjustment to full effect) of response of

domestic demand for air passengers across each of the 10 CSA-ORG pairs.

Florida CSA-Airline Ticket Price Simulations

Results in Table 5-2 in Chapter 5 indicate that the airfare variable is significant for

domestic demand from the Northeast and South regions only. Therefore, simulation results for

these two U.S. regions have statistical validity. Also, the speed of adjustment coefficient from

the Northeast and South regions is statistically different from one. There is a significant

difference between short run and long run adjustments to changes in airfares.

Under actual conditions and the presence of the 9-11 terrorist attacks, monthly demand

from the South region totaled 2,016 passengers per 100,000, while demand from the Northeast

region yielded 1,955 passengers per 100,000. In the short run, both the Northeast and South

regions are equally sensitive to changes in airline ticket prices, but the full impact of a change in

prices is greater in the demand from the Northeast region than from the South region. For

example, a 20% increase in airline ticket prices leads to an instantaneous decrease in demand of

4.4% in both U.S. regions. The percentage decrease in demand represents a decrease of 87

passengers in the Northeast region and 92 passengers in the South region.

But in the long run the decrease in demand from the Northeast region is 6.6% (equivalent

to a decrease of 128 passengers) while the decrease in demand from the South region is only

5.4% (equivalent to a decrease of 114 passengers). The difference is attributed to the speed of

adjustment coefficient. While both U.S. regions present a short run elasticity of-0.25, this value

represents 80% of the full response in the demand from the South region but only 67% of the

response from the Northeast region. These results imply that the Northeast region exhibits more

rigidity in the response to changes in prices than the South region. Figure 6-13 illustrates a) the

relationship between different levels of airline ticket prices and domestic demand for air











passengers and b) short and long run changes in domestic demand for air passengers at different


levels of airline ticket prices for the Northeast and South regions traveling to the Florida CSA.


A) Florida (33) Short run* Florida(33) Short run*
2,150Northeast(1001) ----Long run 2,300 So-u-th-Cl -Q31- TLong run

2,100 ----------------------------- 2,250

2,050 ----- ------ ------- 2,200 -------------------------

C 2,000 ---------- 2,150 ----
e- O00 --- --- --- --- --2, 150 -------- --- ---------------
""


1,YVJU

1,900

1,850

1,800

1,750

1,700

1,650


2,100

2,050

2,000

1,950

1,900

1,850


Ft 0O0 N0 00 0 (N 0 0 fl 0 0
---- -- --- n- e e
--------------- --------- -

---------------- ---------------

---------------- ---------------

---------------- ---------------

-------------------------------
r- II ll t C W'
7t 11C CD 7-t
rII rI M M


Florida (33) 0 Short run*
-Northeast(_1001)_ _____ .- Long run




--- ---------------------------------
-- -- -----------------------------

-- -------------------------


200

150

100

50

0


--------------- -- -----

--------------- ------------

--------------- ---------------

--------------- ---------------

--------------- ---------------
C \o N eM oC 't C tn
IC \o OC 0 (N ^t \o N
rII rII rII


Florida (33) o Short run*
South (10031 ......... Long run
--- ---------------------------------

--- --------------------------------


"-- - -- - - - - -


....I
a -50 -5011 n ll [i 11 1

-200 ----------------------------- -00
.- I -I I .1
100 --------------------- -.100 --------------|----------

S 150 -------------------------------------- -150 --------------------------------------

-200 -------------------------------------- -200 --------------------------------------
(^ (N OC e 0> 'ot C) w C C) x N^ M oC 7t C tf
7t xO N 0' C (N all Cn Nt W xo oC 0 (N OC a,, N7
rI rII rI r e e e e e (N (N e en en en en

Airline Ticket Prices in USD Airline Ticket Prices in USD
Figure 6-13. Florida CSA-Airline ticket price simulations: A) Relationship between different
levels of airline ticket prices and number of passengers; and B) Short and long run
changes in the number of passengers at different levels of airline ticket prices for two
U.S. regions. Asterisk (*) denotes short run adjustment is statistically different from
the long run adjustment given a two-tailed test at 95% confidence.


200

150

100

50

n









South Florida CSA-Airline Ticket Price Simulations

Results in Table 5-3 in Chapter 5 indicate that the airfare variable is significant for three of

the four U.S. regions: Northeast, Midwest, and South. Simulation results for these three U.S.

regions are presented in this section. Also, the speed of adjustment coefficient from these three

U.S. regions is statistically different from one. There is a significant difference between the short

run and long run adjustments to changes in airfares.

Under average conditions and the presence of the 9-11 terrorist attacks, monthly demand

from the Northeast region totaled 911 passengers per 100,000, followed by the South (750

passengers) and Midwest (392 passengers) regions. The Northeast region exhibit the largest long

run price elasticity of demand (-0.26), followed by the South (-0.17) and Midwest (-0.13)

regions. A 20% increase in airfares leads to a 4.5% decrease in demand from the Northeast

region, followed by the South (3.1% decline) and Midwest (2.3% decline) regions.

All three U.S. regions adjust relatively quickly to changes in airfares. For example, an

increase of 20% in airfares leads to a 3.5% (32 passengers) decrease in demand in the short run.

The short run decrease represents 77% of the full response in demand from the Northeast region.

Likewise, the Midwest and South regions experience a 2% decrease (8 passengers) and 2.7%

decrease (20 passengers), respectively. This short run adjustment accounts for 85 and 86% of the

full response from the Midwest and South regions, respectively. Figure 6-14 shows a) the

relationship between different levels of airline ticket prices and number of passengers and b)

short and long run changes in the number of passengers at different levels of airline ticket prices

for the Northeast, Midwest, and South regions traveling to the South Florida CSA.












South Florida (1) Short run*
980 Northeast (1001) "=*Long run
980 - -

960

940

920


B)
60
o
o
i 40
2
-

b 20

I


r I- -l -i r- -- o-


South Florida (1) B Short run*
NortheastC- 01) -- Long run- -----



S-------------- ---
------ -- --- --- ---------










----- ---- --- ----------------


South Florida(l)
405 Midwest (1002)__,

400 -- -------.

395 ------------


390

385

380

375

370





60

40

20


Short run*
Long run


S--------------
---------------


South Florida(l) Short run*
790 South_l(oo3i Long run
780 ----- ------------------

780 ------------- --------------

770 ----------------------

760 ------- ------------------

750

740 ------------

730 -----------------------

720

710 -------------- --------------

700 -------------- --------------




South Florida (1) rn Short run*

60 South (1003) ------- _______Long run -

40 --------------------------------

20 -------------------


S-0 ------------------------20 ----------------------------------0 ----------

-40 ----------------------- ---- -40 ---------------------------------- -40 ---------------------------------

-60 --------------------- -- ----------- -- ---------------------------------
== -60 -60 -60



Airline Ticket Prices in USD Airline Ticket Prices in USD Airline Ticket Prices in USD
Figure 6-14. South Florida CSA-Airline ticket price simulations: A) Relationship between different levels of airline ticket prices and
number of passengers; and B) Short and long run changes in the number of passengers at different levels of airline ticket
prices for three U.S. regions. Asterisk (*) denotes short run adjustment is statistically different from the long run
adjustment given a two-tailed test at 95% confidence.


900

880

860

840

820


--------------- -- -----------

-------------- --------- -

-------------- ---------------









South Florida (1) Short run*
Midwest _100_2)_ Long run

-------------------------------f

---------------------------------









Tampa-St. Petersburg CSA-Airline Ticket Price Simulations

The airfare variable was significant in the demand from the Midwest and South regions

traveling to this destination CSA, as presented in Table 5-5 in Chapter 5. Also, the speed of

adjustment coefficient from the three U.S. regions is not statistically different from one. There is

no significant difference between short run and long run adjustments in demand to changes in

airfares. Therefore, the adjustment is realized immediately.

Under average conditions and the presence of the 9-11 terrorist attacks, monthly demand

from the Midwest region totaled 263 passengers per 100,000 and the South region yielded 347

passengers per 100,000. The South region exhibited a slightly higher long run price elasticity of

demand of -0.146 compare to -0.161 from the Midwest region. A 20% increase in airfares leads

to a 2.8% decrease in demand from the Midwest region, while a similar decrease in airfares leads

to a 3.3% decrease in demand from the South region. In terms of number of passengers, the 2.8%

decrease in demand from the Midwest region accounts for 7 passengers, while a 3.3% decrease

in demand from the South region totals 10 passengers.

But if a 20% decrease in airfares occurs, the increase in demand from the Midwest region

is 3.5% and the South region would experience a 4.1% increase in demand. These results suggest

that the response is higher when there is a decrease in airfares than when there is an increase in

airfares.

Figure 6-15 illustrates a) the relationship between different levels of airline ticket prices

and number of passengers and b) short and long run changes in the number of passengers at

different levels of airline ticket prices for the Midwest and South regions traveling to the Tampa-

St. Petersburg CSA.











A) Tampa-St. Pete (8) Short run
Midwest (1002) Long run
275 ----------------


270 -- ----------- ----------------


- - - - -


20 -

15 -

10 -

5-

0


--------------- -----------------
M 0> Wf' N en oM OC '
Wfi 1o OC CD 7 e t Co o



Tampa- St. Pete (8) Short run
Midwest (1002) Long run

-----------------------------------

-----------------------------------

- -- ----------------------------


Tampa-St. Pete (8) -Short run
5 South (1003) Long run
365 -------------

360 ------------ -------------

355 ---- -------- -------------
35 ------- -- ------- -----
350 ----------------------------

345 ---------------- ------------

340

335 ---------------- -------------

330 ---------------- ------------

325 ---------------- -------------

320 ---------------- ------------
oo n emon N O Ml N
Cl l e en en n e 0


20

15

10

5

0
-


Tampa- St. Pete (8) n Short run
South (1003) Long run

I-- ------------------------------

-- ---------------------------


----------------------


S---------------------------------- ---------------

-10 ------------------------------------ -10

-15 ----------------------------------- -15

-20 ---------------------------------- -20---------------------------------
en 0>o N M en 'to C) l- Cn eII C) N ^


Airline Ticket Prices in USD Airline Ticket Prices in USD
Figure 6-15. Tampa-St. Petersburg CSA-Airline ticket price simulations: A) Relationship
between different levels of airline ticket prices and number of passengers; and B)
Short and long run changes in the number of passengers at different levels of airline
ticket prices for two U.S. regions.







222









Jacksonville CSA-Airline Ticket Price Simulations

Table 5-6 in Chapter 5 indicates that the airfare variable was significant in the demand

from Northeast and West regions. Also, the speed of adjustment coefficient from both U.S.

regions is not statistically different from one. There is no significant difference between the short

run and long run adjustments in demand from these U.S. regions to changes in airfares.

Therefore, inferences of the difference between the short run and long run adjustments in

demand from these U.S. regions are not valid.

Under average conditions and the presence of the 9-11 terrorist attacks, monthly demand

from the Northeast region totaled 47 passengers per 100,000 and the West region yielded 3

passengers per 100,000. The West region has a long run price elasticity of demand of -1.194

compare to -0.32 from the Northeast region. This result suggests that the West region is more

sensitive to changes in airfares than the Northeast region. For example, a 20% increase in

airfares leads to a 19% decrease in demand from the West region, while a similar increase in

airfares leads to a 5.7% decrease in demand from the Northeast region. Even though the decrease

in demand from the West region is large in relative terms, it only represents a decrease of one

passenger per 100,000. On the other hand, the 5.7% decrease represents a decrease of three

passengers per 100,000 in the Northeast region.

Also, demand from the West region is the only CSA-ORG pair showing to be price elastic.

That is, the change in demand is proportionately greater than the change in prices. This finding is

consistent with other studies that have found that as distance increases demand for air travel

becomes more price elastic. Figure 6-16 illustrates a) the relationship between different levels of

airline ticket prices and number of passengers and b) short and long run changes in the number

of passengers at different levels of airline ticket prices for the Northeast and West regions

traveling to the Jacksonville CSA.










Jacksonville (9) Short run
Northeast (1001) =*Long run
--------------- -- r ---- ---------


Jacksonville (8)
West (1004)


I --- I I 1 1 I


C M 0 w C) 00



Jacksonville (9) E Short run
Northeast (1001) Long run


lO 00 O t \O 00



Jacksonville (9) Short run
West (1004) Long run
10 -------------------------- -------


.I


-10 1---------------------------------
1 '00 l 00
1i o 00 M0 7 m t \o
t( I N Ml M M m


-10 -------- ----------------

\t O 00 C I' 7 O 00
m m m m 7t 7t 7t 7t 7


Airline Ticket Prices in USD Airline Ticket Prices in USD
Figure 6-16. Jacksonville CSA-Airline ticket price simulations: A) Relationship between
different levels of airline ticket prices and number of passengers; and B) Short and
long run changes in the number of passengers at different levels of airline ticket
prices for two U.S. regions. Asterisk (*) denotes short run adjustment is statistically
different from the long run adjustment given a two-tailed test at 95% confidence.


Slion run
Long run


50 -- ----------


48 ----------


46 ---------------


44 ---------------


----- -------


t MoI










Fort Myers CSA-Airline Ticket Price Simulations

Table 5-7 in Chapter 5 indicates that the airfare variable was significant in demand from

Northeast only. Also, the speed of adjustment coefficient is statistically different from one.

Therefore, inferences of the difference between short run and long run adjustments in demand

from this U.S. region are valid. Under average conditions and the presence of the 9-11 terrorist

attacks, monthly demand from the Northeast region totaled 105 passengers per 100,000.

Simulation results suggest that a 20% increase in airfares leads to a 9.4% decrease in the demand

from this U.S. region. Seventy-five percent of the decrease in demand is realized in the short run.

The remaining 25% increase was attained in subsequent periods. Figure 6-17 illustrates a) the

relationship between different levels of airline ticket prices and number of passengers and b) the

short and long run changes in the number of passengers at different levels of airline ticket prices

for the Northeast region traveling to the Fort Myers CSA.

A) Fort Myers (14) Short run* B) Fort Myers (14) EI Short run*
140 Northeast (1001) = Lon.g run 10 Northeast (1Qfl) -- Long run


S120 ------------------------ --------------------
130 ------------------------------

110 5 ------------11




0 -----------
90 ----------------------------

80 ---------------- ---------------- -5 -------------------------------------

70 -------------------------------

60 ---------------- --------------- -10 --------------------------------------


Airline Ticket Prices in USD Airline Ticket Prices in USD
Figure 6-17. Fort Myers CSA-Airline ticket price simulations: A) Relationship between different
levels of airline ticket prices and number of passengers; and B) Short and long run
changes in the number of passengers at different levels of airline ticket prices for the
Northeast region. Asterisk (*) denotes short run adjustment is statistically different
from the long run adjustment given a two-tailed test at 95% confidence.









Simulations for Terror

Dummy variables for terror were also used to conduct a set of simulations. The variables

TER2-TER7 were set to zero to represent the absence of the 9-11 terrorist attacks. Simulation

results yield the level of passengers from each U.S. region that would have traveled to Florida

and its destinations each year after 2000. Then, these simulated results are compared to the

average under actual conditions and in the presence of the 9-11 terrorist attacks. Recall that

results from Chapter 5 indicate that the terror variables have a negative and significant impact in

almost all CSA-ORG pairs in the six years following the 9-11 terrorist attacks. Therefore, the

simulated line for no terror (absence of 9-11 terrorist attacks) is greater that the simulated line for

terror (presence of 9-11 terrorist attacks). The difference between both simulated lines represents

the loss in passenger traffic to each CSA-ORG pair. Results also show whether the CSA-ORG

pairs have recovered from the 9-11 terrorist attacks (i.e., the simulated line with terror eventually

meets or crosses the predicted line for no terror).

Simulation results are presented in monthly average passengers by 100,000 by year. That

is, for example, demand of the Florida (33)-Northeast (1001) pair totaled 1,539 passengers per

100,000 per month in 1996. Note that the simulated line for terror matches the simulated line for

no terror. Both lines are equal until 2001 when the 9-11 terrorist attacks occurred. In all cases,

both simulated lines show a growth trend that can be explained not only by the dummy variables

for terror, but also by the increase in income and the decrease in real airfares during the 11-year

period. Also, there has not been a full recovery from the 9-11 terrorist attacks in most CSA-ORG

pairs. Latent possibilities of a similar event occurring again and the reflection of the increase in

inconvenience in the air travel experience may be hindering a full recovery in the demand for air

travel to Florida.









The following sections present a discussion of the simulated results in terms of levels of

demand that would have been attained in the absence of the 9-11 terrorist attacks for each

destination CSA. Note that two percentages are reported when comparing the difference between

the simulated line with terror and the simulated line without terror in relative terms to denote that

the relative change will differ depending on the base used. The convention used is to report the

relative change from the simulated line with terror followed by the relative change from the

simulated line without terror in parenthesis.

Florida CSA-Terror Simulations

In general, the difference between the simulated demand for passengers to the Florida CSA

with terror and without terror was 1,480 passengers or a 26% (21%) decrease in demand. Figure

6-18 illustrates the comparison between simulated results of domestic demand for air passengers

from four U.S. regions traveling to the Florida CSA in the presence and absence of the 9-11

terrorist attacks. Simulation results suggest that the Northeast and West regions experienced the

largest decrease in demand in 2002. The decrease totaled 595 and 95 passengers per 100,000

which is equivalent to a 35% (26%) and 30% (23%) decrease in demand from the Northeast and

West regions, respectively. Results from Chapter 5 indicate that no valid inferences can be made

from the demand from Northeast region after September 2003 (TER4 not significant). Similarly,

demand from the West region presented no significant values for TER4 and TER6. Demand

from the Northeast region during the two-year period post 9-11 has decreased by 415 passengers

per 100,000 or a 22% (18%) decrease in demand per year. Likewise, the decrease in demand

from the West region since 2001 is 59 passengers or a 20% (17%) decline per year.

The South region had a decrease of 608 passengers in 2002, while the decline in 2006

totaled 576 passengers. The initial decrease in 2002 accounted for a 32% (24%) decrease in

demand. During the five year period post 9-11 the average loss has been 480 passengers,











equivalent to a 22% (18%) decline in demand. Meanwhile, the Midwest region has experienced

an increasing decline in passengers since 2001 and the trend has not reversed. The decrease in


demand from the Midwest region totaled 608 in 2002 and it has increased to a loss of 678

passengers in 2006. The overall loss in demand from this U.S. region adds up to 526 passengers,

equivalent to a 40% (28%) decline in demand five years after the 9-11 terrorist attacks.


Florida (33) Terror Florida (33) Terror
3,000 Northeast(.001C )_ No Terror 2500MidwestC (102)_ No Terror

o 2,500 ---------------
o 2,000 ---------------------------- ---

z 2000 ------------


1,500

1,500

1,000

500


1,500


1,000


500


I I I I I I I I I
\ThooON~ ~


3,500

S3,000

" 2,500

2,000

1,500

S1,000

500


Florida (33)
Sl-SouthC.00 3)_

----------------

-----------------


----- --------

----------------


-Terror
.No Terror

----------------
----------------

00- ----------


----------------

----------------


Florida (33) Terror
West (100_4_)_ No Terror



--- --------------------
- - - - -


' 1 11+


C' ( ( (C all C) (NI (MI Y ea r
Year Year
Figure 6-18. Florida CSA-Terror simulations: comparison between the simulated results of
demand for air passengers from four U.S. regions in the presence and absence of the
9-11 terrorist attacks.


' I 1 1


----------------


- 1-0--------









South Florida CSA-Terror Simulations

The difference between the simulated demand for passengers to the South Florida CSA

with terror and without terror was 393 passengers or a 20% (16%) decrease in demand. Figure 6-

19 presents the comparison between simulated results of domestic demand for air passengers

from four U.S. regions traveling to the South Florida CSA in the presence and absence of the 9-

11 terrorist attacks.

Results from Table 5-3 in Chapter 5 indicate that the Northeast and West regions have

recovered from the 9-11 terrorist attacks since September 2003. Nevertheless, simulation results

suggest that both U.S. regions experienced the largest decrease in demand in 2002. Demand

decreased by 203 and 29 passengers per 100,000 which is equivalent to a 25% (20%) and 21%

(17%) decrease in the Northeast and West regions, respectively. Demand from the Northeast

region during the three-year period (2001 to 2003) post 9-11 decreased by 134 passengers per

100,000 or a 15% (13%) decrease in demand per year. Likewise, the decrease in demand from

the West region since 2001 is 21 passengers or a 14% (13%) decline per year.

Both the Midwest and South regions have not recovered from the 9-11 terrorist attacks.

The Midwest region had a decrease of 125 passengers in 2002. The initial decrease in 2002

accounted for a 34% (25%) decrease in demand. During the five-year period post 9-11 the

average loss has been of 144 passengers, equivalent to a 38% (27%) decline in demand.

Meanwhile, the decrease in demand from the South region totaled 208 passengers in 2002

equivalent to a 30% (23%) decrease in demand. Overall loss during the five-year period post 9-

11 is 176 passengers, equivalent to a 23% (19%) decline in demand over the five years following

the 9-11 terrorist attacks.













1,400


1,200

1,000

800

600


400


200










1,200


1,000


800


600


400


200


South Florida (1) Terror
Northeast (1001) == No Terror


--------------------------------


----------------


---- -------------

--------------------------------

--------------------------------


--------------------------------



a l ll C C C C C C "l C




South Florida (1) Terror
South (1003) =No Terror


Year Year

Figure 6-19. South Florida CSA-Terror simulations: comparison between the simulated results of
demand for air passengers from four U.S. regions in the presence and absence of the
9-11 terrorist attacks.







230


South Florida (1) Terror
Midwest (1002) No Terror
700 ---------

600 --------------------------------

500 ------------------ ------------


400------ ---------- --------------

300 ----------------------------


200 -------------------------------
200


100--------------------------------








South Florida (1) Terror
250 est (1004) No Terror
250



200 -----------------------------



150 ------------- ---- ----------


I


I


- - -



- - - -


000









Orlando CSA-Terror Simulations

Overall, the difference between the simulated demand for passengers to the Orlando CSA

with terror and without terror was 506 passengers or a 29% (22%) decrease in demand. Figure 6-

20 presents simulated demand for air passengers from four U.S. regions traveling to the Orlando

CSA (5) in the presence and absence of the 9-11 terrorist attacks.

Results from Table 5-4 in Chapter 5 indicate that none of the four U.S. regions have

recovered from the 9-11 terrorist attacks. Simulation results suggest that the Midwest and West

regions experienced the largest decline representing a 36% (27%) and a 37% (27%) decrease in

demand, respectively. The Northeast and South regions experienced a decrease of 25% (20%)

and 24% (20%) in demand, respectively.

In addition, these two U.S. regions suffered the largest decrease in demand in 2002. The

decrease totaled 197 and 153 passengers per 100,000 which is equivalent to a 38% (27%) and

33% (25%) decrease in demand from the Northeast and South regions, respectively. Demand

from the Northeast region during the five-year period post 9-11 decreased by 158 passengers.

Likewise, the decrease in demand from the South region since 2001 totaled 130 passengers.

The Midwest region had a decrease of 141 passengers in 2002. The initial decrease in 2002

accounted for a 35% (25%) decrease in demand. During the five-year period post 9-11 the

average loss has been of 166 passengers. Meanwhile, the decrease in demand from the West

region totaled 50 passengers in 2002, equivalent to a 42% (29%) decrease in demand. The

overall loss during the five-year period post 9-11 is 52 passengers.













1,000

900

800

700

600

500

400

300

200

100


Orlando (5) Terror
Northeast (1001) -==No Terror

-----------------------------

------------------------ ------

----------------- -------- --

----------- --- ----- --------

--- --------------------------

--------------------------------

--------------------------------

--------------------------------





c^ ^ c^ oo oo


Orlando (5)
South (1003)


-Terror
-No Terror
- - -


Orlando (5) Terror
West (1004) No Terror
250



200 ------------------------ ------



150 -------------- ----------- ---



100 ----------------------------



50 ----------------------------


) O" 0 0 o C C, C) C) C5 )o


Year Year

Figure 6-20. Orlando CSA-Terror simulations: comparison between the simulated results of
demand for air passengers from four U.S. regions in the presence and absence of the
9-11 terrorist attacks.







232


Orlando (5) Terror
Midwest (1002) No Terror







------------- ---------------.

-- -----------------------------












Tampa-St. Petersburg CSA-Terror Simulations

The difference between the simulated demand for passengers to the Tampa-St. Petersburg

CSA with terror and without terror was 261 passengers or a 27% (21%) decrease in demand.

Figure 6-21 presents simulated demand for air passengers from four U.S. regions traveling to the

Tampa-St. Petersburg CSA in the presence and absence of the 9-11 terrorist attacks.

Results from Table 5-5 in Chapter 5 indicate that the Northeast region has recovered from

the 9-11 terrorist attacks since September 2003. Simulation results suggest that this U.S. region

experienced the largest decrease in demand in 2002. The decrease totaled 65 passengers per

100,000 which is equivalent to a 25% (20%) decrease in demand from the Northeast region.

Demand from the Northeast region during the three-year period (2001 to 2003) post 9-11

decreased by 46 passengers per 100,000 or a 17% (14%) decrease in demand per year.

The other three U.S. regions have not recovered from the 9-11 terrorist attacks. The

Midwest region had a decrease of 88 passengers in 2002, while demand from the South region

decreased by 104 passengers and demand from the West region decreased by 10.

During the five-year period post 9-11 the average loss in the demand from the Midwest

region has been of 106 passengers, equivalent to a 40% (28%) decline in demand. The decrease

in demand from the South region is 98 passengers, equivalent to a 27% (21%) decline in

demand. Finally, demand from the West region declined by 10 passengers or a 15% (13%) drop

during the five-year period post 9-11.










Tampa-St. Pete (8) -Terror
450 Northeast (1001) -- -No Terror
450

400 ---------------------------

o 350 ---------------------

S300 -------------------- -----

250 -------- -------------------

' 200----------

2 150 ---------- ---------------

A 100 -----------------------------
50 --------------------------------
'- ^ o ^ '. m ^ r -
c^ ^ ^c^o o o o
c^ ^ ^c^o o o o
'- 'i -i '- c c r r r r r


Tampa-St. Pete (8) -Terror
South (1003) No Terror


0

o 400


" 300


q 200
sS


Tampa-St. Pete (8) Terror
Midwest (1002) No Terror




-------------. -------------------
-------------------- -----------



----- ---------- ---------




--------------------------------











West (1004) No Terror


Year Year

Figure 6-21. Tampa-St. Petersburg CSA-Terror simulations: comparison between the simulated
results of the demand for passengers from four U.S. regions in the presence and
absence of the 9-11 terrorist attacks.






234








----------- ---
---------------


--------------- 0


00000 -----------44









Jacksonville CSA-Terror Simulations

The difference between the simulated demand for passengers to the Jacksonville CSA with

terror and without terror totaled 107 passengers or a 39% (28%) decrease in demand. Figure 6-22

shows simulated demand for air passengers from four U.S. regions traveling to the Jacksonville

CSA (9) in the presence and absence of the 9-11 terrorist attacks.

Results from Table 5-6 in Chapter 5 indicate that the Northeast region has recovered from

the 9-11 terrorist attacks since September 2004. Those results also suggest that the West region

had no significant response to the 9-11 terrorist attacks.

Simulation results suggest that the Northeast region experienced the largest decrease in

demand in 2003. The decrease totaled 18 passengers per 100,000 which represented a 41%

(29%) decrease in demand. Demand from the Northeast region during the four-year period (2001

to 2004) post 9-11 decreased by 14 passengers per 100,000 or a 30% (23%) decline in demand

per year.

The other two U.S. regions have not recovered from the 9-11 terrorist attacks. The

Midwest region had a decrease of 23 passengers in 2002 and has been growing larger ever since.

Similar results are experienced by demand from the South region which decreased by 66

passengers in 2002.

During the five-year period post 9-11 the average loss in demand from the Midwest region

has been of 27 passengers, equivalent to a 57% (36%) decline in demand. The decrease in

demand from the South region is 66 passengers, equivalent to a 37% (27%) decline in demand.









Jacksonville (9) Terror
Northeast (1001) ==No Terror


Jacksonville (9) Terror
Midwest (1002) No Terror




------------------------------










I'D o000CWI
011 011 011 011 C)
011 011 011 oooo


Jacksonville (9)
South (1003)


- Terror
- No Terror


Jacksonville (9)
WPat (1 004


Terror
--Nnr TPrrcir


i ..


00I 0 C W I

Year


6* --


IIIIIIIIIII
~~~0000000
~~~0000000
o i i i i i i Yei r

Year


Figure 6-22. Jacksonville CSA-Terror simulations: comparison between the simulated results of
the demand for air passengers from four U.S. regions in the presence and absence of
the 9-11 terrorist attacks.





236


-------------

----- --- 4 ____
-------- ------
Cx


------------

------ --------

----------------


I I I I I I
11 r- C 1 C) II M t 1
al al INOI C C C C C C C
al al llal C C C C C C C









Fort Myers CSA-Terror Simulations

The difference between the simulated demand for passengers to the Fort Myers CSA with

terror and without terror amounted to 107 passengers or a 39% (28%) decrease in demand.

Figure 6-23 illustrates simulated demand for air passengers from four U.S. regions traveling to

the Fort Myers CSA in the presence and absence of the 9-11 terrorist attacks.

Results from Table 5-7 in Chapter 5 indicate that the Northeast region has recovered from

the 9-11 terrorist attacks since September 2003 and the South region since 2002. Recall that no

valid inferences can be made from the West region because its speed of adjustment coefficient is

unstable.

Simulation results suggest that the Northeast region experienced the largest decrease in

demand in 2002. The decrease totaled 28 passengers per 100,000 habitants, equivalent to a 35%

(26%) decrease in demand. Demand from the Northeast region during the three-year period

(2001 to 2003) post 9-11 decreased by 19 passengers per 100,000 or a 20% (17%) decline in

demand per year.

Similarly, the South region experienced a decrease in 2002 when its demand declined by

15 passengers equivalent to a 17% (14%) decrease. Demand from the South region during the

two-year period (2001 to 2002) post 9-11 decreased by 10 passengers per 100,000 or an 11%

(10%) decline in demand per year.

The Midwest region has not recovered from the 9-11 terrorist attacks. Its demand has

decreased by 38 passengers in 2002. Although the trend reversed in 2004, a deficit in demand

still exists. During the five-year period post 9-11 the average loss in demand from the Midwest

region has been of 30 passengers, equivalent to a 19% (16%) decline in demand.













Fort Myers (14) Terror
Northeast (1001) ==No Terror


Fort Myers (14)
Midwest(1002)


i1C i^ Oi i^ C ri i i M 't iV i1C






Fort Myers (14) Terror
South (1003) No Terror


I", I I I I




Year


I I I I I I I I I I I

IDr 0 0 0"0 C C C M ZT





Fort Myers (14) Terror
West (1004) -No Terror























I"III----- -
c --- ---- --- -- -- --- ---- ---
c --- ---- --- -- -- --- ---- ---
----------------------------





011 0 0 0


Year


Figure 6-23. Fort Myers CSA-Terror simulations: comparison between the simulated results of
the demand for passengers from four U.S. regions in the presence and absence of the
9-11 terrorist attacks.









238


Terror
No Terror


------------

------------

------ -------

------ ---------

-- ------------


--------------

------- ------

----------------


----------------






--- ----------

----------------


---------------


---- ---------


----------------









Simulations for Hurricanes

The fourth set of simulations is related to the presence or absence of hurricanes affecting

Florida. In this case, the variables HCATIMED and HCATHGH were set to zero to represent the

absence of hurricanes during an 11-year period (1996-2006). Simulation results yield level of air

passengers from each U.S. region that would have traveled to Florida in the absence of a

hurricane making landfall in Florida. Then, these simulated results are compared to demand

under actual conditions as described in previous sections.

Results from Chapter 5 indicate that the two variables representing hurricanes affecting

Florida have a negative and significant impact in almost all 24 CSA-ORG pairs. Results are

presented in changes in the number of passengers that would have traveled to a particular

destination in the absence of hurricanes. Due to the seasonality of hurricanes (hurricane season

runs from June to November), the results are presented by months. Presenting the results by

years will negate the impact of hurricanes since it will average out the changes over the entire

year. Therefore, results by month clearly show the impact of hurricanes on the demand for air

travel to Florida.

The following sections present a discussion of the simulated results in terms of changes in

number of passengers that would have traveled to Florida in the absence of the hurricanes. The

response speed (comparing short run adjustment to full effect) across each of the six destination

CSAs including Florida in relation to each of the four U.S. regions is also presented.


Florida CSA-Hurricane Simulations

As expected, an immediate increase in demand for air passengers traveling to Florida in the

absence of hurricanes occurs from July to November. Also note that although there are no

hurricanes that could instantaneously impact demand during the period between December and









May, a long run effect is present. Nevertheless, results presented in Table 5-2 in Chapter 5

indicate that long run effects are only statistically significant for the Northeast, South, and West

regions. Increase in demand from the Midwest region is realized immediately and thus, no

lagged impact is observed.

August, September, and October presented the largest changes in demand for air

passengers that would have traveled to Florida in the absence of hurricanes. Specifically,

September would have experienced an immediate increase in demand for passengers from the

Northeast region of 19 passengers if no hurricanes had affected Florida. Similar results were

found in the Midwest, South, and West regions. These three U.S. regions show the highest

increase in demand during September. Demand from the South region yields an increase of 23

passengers during this month, followed by the Midwest region with 17 passengers and the West

region with three passengers.

Figure 6-24 illustrates short run and long run changes in the number of passengers from

four U.S. regions traveling to the Florida CSA in the absence of hurricanes in Florida. These

results are well below the increases in demand shown in the absence of a terrorist attack. For

example, during the hurricane season from June to November, the average increase due to

absence of hurricanes accounts to 0.68% of total demand from the Northeast region during the

hurricane season. Similarly, absence of hurricanes only produces a 0.73% increase in demand

from the Midwest region, a 0.66% increase in the South region, and a 0.54% increase in the West

region.











Florida (33) n Short run*
Northeast (1001) Long run



--------------------------- ----------


------------------------ -- ----------


----------------------- -- -------


-------------------- -- ----- --


-------------------









Florida (33) 0 Short run*
South (1003) Long run



--------------------------- ----------


-.------------------------ ----------


----------------------- ----------


-.----------------- -- - -




-- -- -- - - - -


Florida (33) Short run
Midwest (1002) Long run



25 --------------------------------------


20 --------------------------------------


15 -------------------------- ----------


10 -------------------------- ----------


5 ----------------------- -- ----
30


25


20





10





0* 111111111 111.





Florida (33) Short run*
West (1004) Long run



25 --------------------------------------


20 --------------------------------------


15 --------------------------------------


10 --------------------------------------


5 --------------------------------------
25


20





10





0 I I I I I I I I I I I I


Month Month

Figure 6-24. Florida CSA-Hurricane simulations: short and long run changes in the number of
passengers from four U.S. regions in the absence of hurricanes in Florida. Asterisk (*)
denotes short run adjustment is statistically different from the long run adjustment
given a two-tailed test at 95% confidence.






241









South Florida CSA-Hurricane Simulations

Similar to the Florida CSA, an immediate increase in demand for air passengers traveling

to the South Florida CSA in absence of hurricanes occurs from July to November. In addition,

there is a long run impact in demand during months that naturally do not experience hurricanes.

Inferences of the long run effect are valid since results presented in Table 5-3 in Chapter 5

indicate that the speed of adjustment coefficient is statistically significant across all U.S. regions

traveling to this destination CSA.

August, September, and October presented the largest changes in domestic demand for air

passengers that would have traveled to the South Florida CSA in the absence of hurricanes.

September would have yielded an immediate increase in demand for passengers from the

Northeast region of 10 passengers if no hurricanes had affected Florida. Similar results were

found in the Midwest, South, and West regions. These three U.S. regions show the highest

increase in demand during September. Demand from the South region yields an increase of 10

passengers during this month, followed by the Midwest region with seven passengers and the

West region with one passenger.

Figure 6-25 illustrates short and long run changes in domestic demand for air passengers

from four U.S. regions traveling to the South Florida CSA in absence of hurricanes in Florida.

Overall, absence of hurricanes barely increases demand for air transport. For example, during the

hurricane season from June to November, the average increase due to an absence of hurricanes

accounts for 0.76% of total demand from the Northeast region during the hurricane season.

Similarly, absence of hurricanes only produces a 1% increase in demand from the Midwest

region, a 0.75% increase in the South region, and a 0.61% increase in the West region.












South Florida (1) Short run*
Northeast (1001) Long run



12 --------------------------------------


10 -------------------------- ----------


8 ----------------------- ----------


6 --------------------- -- -------


4 -------------------- --


2 -------------------- -----


0 5 0
14 South(1003) Longrun


















12 --------------------------------------


10 -------------------------- ----------


8 -------------------------- ----------


6 ----------------------- ----------
12
0C
0
4 10---------- ----
aC
' 8




" 4 -
6







S2-------------------- -















0 I, I ". I I I I I .
South Florida (1) n Short run*
South (1003) Long run
14


O 12
0
0
^ 10


' 8

u 6


4


- 2


0 -


South Florida (1) Short run*
Midwest (1002) Long run



12 --------------------------------------


10 --------------------------------------


8 --------------------------------------


6 -------------------------- ----------


4 -------------------------- ----------


2 ----------------------- ---- ----


C4A 0






South Florida (1) Short run*
SWest (1004) Long run



12 --------------------------------------


10 --------------------------------------


8 --------------------------------------


6 --------------------------------------


4 --------------------------------------


2 --------------------------------------
14


12


10














0 I1I11I11I11I1 I I


Month Month

Figure 6-25. South Florida CSA-Hurricane simulations: short and long run changes in the
number of passengers from four U.S. regions in the absence of hurricanes in Florida.
Asterisk (*) denotes short run adjustment is statistically different from the long run
adjustment given a two-tailed test at 95% confidence.






243









Orlando CSA-Hurricane Simulations

Demand for air passengers to the Orlando CSA experience an immediate increase in the

number of passengers traveling to Florida in the absence of hurricanes from July to November.

Also, there is a long run impact in demand from two U.S. regions during months that naturally

do not experience hurricanes. The long run impact is only significant in demand from the

Northeast and Midwest regions as results presented in Table 5-3 in Chapter 5 indicate. The speed

of adjustment coefficient is statistically significant only in the Northeast and Midwest regions

traveling to this destination CSA. August, September, and October presented the largest changes

in the number of passengers that would have traveled to the Orlando CSA in the absence of

hurricanes. September would have yielded an immediate increase in demand for passengers from

the Northeast region of 10 passengers if no hurricanes had affected Florida. Similar results were

found in the Midwest, South, and West regions. These three U.S. regions show the highest

increase in demand during September. Demand from the South region yields an increase of 10

passengers during this month, followed by the Midwest region with seven passengers and the

West region with one passenger.

Overall, absence of hurricanes barely increases domestic demand for air transport. For

example, during the hurricane season from June to November, the average immediate increase

due to an absence of hurricanes accounts to 0.54% (0.79% in the long run) of total demand from

the Northeast region during the hurricane season. Similarly, absence of hurricanes only produces

a 0.63% increase (0.78% in the long run) in demand from the Midwest region. Statistically the

immediate increase of 0.47% in the South region and 0.43% in the West region is equal to the

long run impact. Short and long run changes in the number of passengers from four U.S. regions

traveling to the Orlando CSA in the absence of hurricanes in Florida are presented in Figure 6-

26.












Orlando (5) E Short run*
Northeast (1001) Long run



6 --------------------------- ----------



5 ------------------------ -- -- --------


4 --------------------- -- -- -----


3 --------------------- -- -- ----


2 -------------------- -- ----
7





r 5

4




"2





o rla, I(,5 r, hr ru n,


7


6




4
3





2
0 1

u



Bf


Orlando (5) n Short run
South (1003) m Long run


--------------------------------------



--------------------------------------



-------------------------- ----------


-------------------- --|-------


-------------------- I- ----
:::::::::::::::::::I:lI::::I:


Orlando (5) Short run*
Midwest (1002) Long run



6 --------------------------------------


5 --------------------------- ----------


4 -------------------------- ----------


3 ---------------------- ----------


2 -------------------- -------


I ----------- --------- ----


0 I I I I I I I I I I I I






Orlando (5) Short run
West (1004) Long run



6 I I I I I I I I I------------------------------------I I --


5 -.-------------------------------------


4 --------------------------------------


3 --------------------------------------


2 --------------------------------------


1 --------------------------------------


0) I I I I I I I I I I I I


C < C1 C O


Month Month

Figure 6-26. Orlando CSA-Hurricane simulations: short run and long run changes in the number
of passengers from four U.S. regions in the absence of hurricanes in Florida. Asterisk
(*) denotes short run adjustment is statistically different from the long run adjustment
given a two-tailed test at 95% confidence.






245









Tampa-St. Petersburg CSA-Hurricane Simulations

Results presented in Table 5-4 in Chapter 5 indicate that the hurricane variables are not

statistically significant and hence do not affect demand from the Northeast region. Also, the

speed of adjustment coefficient is not statistically significant in any of the remaining three U.S.

regions. Therefore, inferences of the demand from the Northeast region are not included in the

discussion and the difference between the short run and long run is statistically equal to zero.

Hence all three U.S. regions adjust immediately to an absence of hurricanes.

The Tampa-St. Petersburg CSA experienced an instantaneous increase in the number of

passengers in the absence of hurricanes during the months covering the hurricane season (from

June to November). August, September, and October presented the largest changes in the number

of passengers that would have traveled to the Tampa-St. Petersburg CSA in the absence of

hurricanes. September would have yielded an immediate increase in demand for passengers from

the Midwest region (+2 passengers) if no hurricanes had affected Florida. Similar results were

found in the South and West regions. These U.S. regions show the highest increase in demand

during September. Demand from the South region yields an increase of three passengers during

this month. Numerically, the increase in demand from the West region is close to zero.

Absence of hurricanes barely increases domestic demand for air transport to the Tampa-St.

Petersburg CSA. During the hurricane season, the average immediate increase due to an absence

of hurricanes accounts for 0.51% of the total demand from the Midwest region. Similarly, the

absence of hurricanes only produces a 0.44% increase in demand from the South region.

Statistically the immediate increase in demand from these two U.S. regions equals the long run

impact. Short and long run changes in domestic demand for air passengers from four U.S.

regions traveling to the Tampa-St. Petersburg CSA in the absence of hurricanes in Florida are

presented in Figure 6-27.












Tampa-St. Pete (8)
Northeast (1001


] Short run*
SI r"n nin


---- ---- -- ---- -- L ?----- --





S1 5
87


6





4 -------------------------------------


3 -------------------------------------


2 --------------------------- ----------
aI





S------------- ------- -------


0
^ - -

a - -
Bf 1
= I
-E -.. .. .. .


a UC- o& r
i ^

^fcS^S^ ^^0^^


Tampa-St. Pete (8) r Short run
South (1003) Long run










--------------------------------------i
--------------------------------------


--------------------------------------



-------------------------- ----------


-------------------- ----


"J


Tampa-St. Pete (8) Short run
Midwest (1002) E Long run
7


6 --------------------------------------


5 -.-------------------------------------


4 --------------------------------------


3 --------------------------------------


2 -------------------------- ----------


I ----------------------- ----------









Tampa-St. Pete (8) Short run

7 West (1004) Long run


6 --------------------------------------


5 -.-------------------------------------


4 --------------------------------------


3 --------------------------------------


2 --------------------------------------


1 --------------------------------------
6


5


4


3


2


1









Tampa-St. Pete (8) Short run
West (1004) Long run
7


6


5


4


3


2




S I I I I I I I I I I I I


i i i >i i i & &C C1 C t i to i i i i i i i i


Month Month

Figure 6-27. Tampa-St. Petersburg CSA-Hurricane simulations: short and long run changes in
the number of passengers from four U.S. regions in the absence of hurricanes in
Florida. Asterisk (*) denotes short run adjustment is statistically different from the
long run adjustment given a two-tailed test at 95% confidence.






247


7


6





4
3





2
0







Bf









Jacksonville CSA-Hurricane Simulations

The hurricane variables are statistically significant only in the demand from the Midwest

and South regions, as shown in Table 5-6 in Chapter 5. Also, the speed of adjustment coefficient

from the Midwest region is statistically significant. Only inferences related to demand from the

Midwest and South regions are included in the discussion. The South region adjusts immediately

to an absence of hurricanes. In the absence of hurricanes, the immediate increase in domestic

demand for air passengers occurs during the months covering the hurricane season (from June to

November). But numerically, the increase is of one passenger or less. In relative terms the

increase is only of 0.51% in the Midwest region and 0.40% in the South region. Short and long

run changes in domestic demand for air passengers from four U.S. regions traveling to the

Jacksonville CSA in the absence of hurricanes in Florida are presented in Figure 6-28.

Jacksonville (9) Short run* Jacksonville (9) n Short run
Midwest (1002) Long run South (1003) Long run
7 7 ... .

6 -------------------------------------- 6 --------------------------------------
6 ----------------- ----------------- ---

5-------------------5

4 ------------------------------------ 4------------------------------------

S 3 ------------------------------------- 3 -------------------------------------

2 1------------------------------------ 2



0 I I I I I I I I I I I I 0 -I I I I I l



Month Month
Figure 6-28. Jacksonville CSA-Hurricane simulations: short and long run changes in the number
of passengers from four U.S. regions in the absence of hurricanes in Florida. Asterisk
(*) denotes short run adjustment is statistically different from the long run adjustment
given a two-tailed test at 95% confidence.










Fort Myers CSA-Hurricane Simulations

Hurricane variables are statistically significant in the demand from three U.S. regions:

Northeast, Midwest and South, as shown in Table 5-7 in Chapter 5. Recall that the results from

Fort Myers (14)-West (1004) pair indicate that estimates are not valid due to the lack of stability

of its speed of adjustment coefficient. In the absence of hurricanes, the immediate increase in the

number of passengers occurs during the months covering the hurricane season (from June to

November). But numerically, the increase is of one passenger in demand from the South region

and less than one in the other two U.S. regions. In relative terms the increase is only of 0.58% in

the Northeast region, 0.33 in the Midwest region, and 0.48% in the South region. Short and long

run changes in domestic demand for air passengers from four U.S. regions traveling to the Fort

Myers CSA in the absence of hurricanes in Florida are presented in Figure 6-29.

Fort Myers (14) Fort Myers (14) Fort Myers (14)
Northeast (1001) Midwest 1002) South (1003)
7 ----------------- -------- 7 --------------------- ------ 7 --------------------Sh ---
7 Short run Short run 7 Short run
SLong run Long run 0 Long run
"5 6 ----------------------- 6 ------------------------- 6 --------------
o




53 3 3-------------------- ------------------------- 3
4 ------------------------- 4 --------------------------- 4 ---------------------------



2 --- ------------------ 2 ------------------------- 2

------------ ----- ---------------- ------- 1

0 I 0 1 0 I


Month Month Month


Figure 6-29. Fort Myers CSA-Hurricane simulations: short and long run changes in the number
of passengers from four U.S. regions in the absence of hurricanes in Florida. Asterisk
(*) denotes short run adjustment is statistically different from the long run adjustment
given a two-tailed test at 95% confidence.









Simulations for Seasonality

The fifth set of simulations is related to the presence of seasonal patterns in demand for air

transportation to Florida. In this case, the variables MTH2-MTH12 were simulated at the actual

values to identify how demand varies at each month. The following sections present a discussion

of the simulated results in terms of percentage changes from the average number of passengers in

a month. Deviations of more than 10% are noted as large changes from the average.

Florida CSA-Seasonality

The four U.S. regions present high variability during the months of March and September.

The largest increase from the average occurs in March and the largest decrease in September.

Nevertheless, the Midwest region presented more months with a 10% change or greater from the

mean than any other region. February (+17%), March (+41%), April (+12%), and December

(+11%) experienced an increase equal to or greater than 10% from the average (1,335

passengers), while August (-15%), September (-35%), and October (-10%) showed large

decreases from the average.

In contrast, the South region presents the lowest variability. Only March (+19%) and

September (-23%) showed a percentage change of 10% or higher from the average (2,106

passengers). These results suggest that demand from the Midwest region is highly seasonal

compared to the South region. Figure 6-30 presents a) simulated monthly seasonal patterns of

demand and b) percentage change from the monthly average demand for passengers from four

U.S. regions traveling to the Florida CSA.













- Northeast Midwest South West


O 2500
-


O 2000
-


S1500
U,

1 1000


i 500


0


..- ------- Florida (33)
..-------------------- Northeast(1001)
..-------------------------------------
------- ----------------------------


----------- ----- ---- ---3

------------------ ------- ---------

..------------------------ --






..------------------------- ----------


..------------------------------------









-------------------------- ---------
. . .Florida (33)
South (1003)






------------------------------------




..------------------------------------
0 -- ---- =----- -- to ---P --- U
- - - - - - C


Month

Figure 6-30. Florida CSA-Seasonality simulations:


Month

50
40
30
20
10
0
-10
-20
-30
-40
-50





50
40
30
20
10
0
-10
-20
-30
-40
-50


---------------------- Florida (33)
------- ------------- Mid est (1002)
------- Florida (33)
------- ----------------West (100



















Month---- -----------------------
---------------------- -------
------------------------- ----------
------------------------- ----------
-------------------------------------
-------------------------------------





------------------------- Florida(33)
------------------------- Wgest(1004)

-------------------------------------
-------------------------------------



------------------------- ----------
------------------------- ----------
-------------------------------------
-------------------------------------
-------------------------------------




Month


A) Monthly seasonal pattern of domestic


demand for air passengers; and B) Percentage change from the monthly average of
domestic demand for air passengers from four U.S. regions.


3000 -


Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec


------------------ -------------------------------------------------------------


- - - - - - - - - - - -


- - - - - - - - - - - - - - - - - - - -










South Florida CSA-Seasonality

Similar to the Florida CSA, this CSA experienced the largest variability during March

(increase) and September (decrease) from each of the four U.S. regions. The Midwest region

presented more months with a 10% change or greater from the average than any other region.

The first four calendar months showed an increase of 10% or greater than the average, while the

months between May and November experienced a decrease of 10% or greater than the average

(392 passengers). In contrast, the South region presents the lowest variability. Only March

(+18%) and September (-26%) showed a percentage change of 10% or higher from the average

(750 passengers). These results suggest that demand from the Midwest region is highly seasonal

compared to the South region. Figure 6-31 illustrates simulated seasonal patterns of demand for

air passengers and Figure 6-32 shows the percentage change from the monthly average demand

from four U.S. regions traveling to the South Florida CSA.


Northeast Midwest South West
1200---------------------------

S 1050 --------------- ------ ---------------------------------------------------------
goo90 ------------------- --------- ------------ --------------- ------------



| 600 -- ----
450 ------------------------------------------------------ ------------------- -------


300 ---------------------------------------------------- -------- --------------------
o-







Month

Figure 6-31. South Florida CSA-Seasonality simulations: monthly seasonal pattern of domestic
demand for air passengers from four U.S. regions.













50

- 40

S30

S20
S10

o
n 10 -



1 -10 -

-20

-30

-40

-50


South Florida (1)
Northeast (1001)

------------------------------------


50

40

30

20

10

0

-10

-20

-30

-40

-50


South Florida (1)
Midwest (1002)

-------- ----------------------------

-------- ----------------------------


I- --- -- --- -- -- -- ---- -- -- -- -- -- -- -- -- -- --


South Florida (1)

50 South (1003)

40

~ 30

I 20
40 --------------------------------------

30 --------------------------------------

20 --------------------------------------

10 -----
( 0 I-


South Florida (1)

50 est(1004)

40------------

30------------

20------------

10
0 ---------


-10 -------------------------------- ---10--

o -20 ------------ -- -20 ---------- ----

< -30 --------------------------------- -30 --------------

-40 ---------------------------------- -40------------

-50 ---------------------------------- -50 --------------



Month Month

Figure 6-32. South Florida CSA-Seasonality simulations: percentage change from the monthly
average demand for passengers from four U.S. regions.







253


-

-

-

-

-

-


------------- ~3 ~
- - - - - -
- - - - - - -


- - - - - - -
- -

- - - - -










Orlando CSA-Seasonality

March (increase) and September (decrease) recorded the largest variability from each of

the four U.S. regions traveling to the Orlando CSA. Interestingly, variability of the other 10

months is less than 10% in three of the four U.S. regions. Only the Northeast region shows a

15% increase from the average in a third month (April). These results suggest that demand from

each of the four U.S. regions is somewhat steady through almost all year. These results imply

that seasonal patterns in demand for passengers from any U.S. region traveling to the Orlando

CSA are much smaller relative to other destination CSAs. Figure 6-33 presents simulated

seasonal patterns of domestic demand for air passengers and Figure 6-34 shows percentage

changes from the monthly average domestic demand from four U.S. regions traveling to the

Orlando CSA.


Northeast Midwest South West
750 --------------------------------------------------------------------------------


S-----600 ---


450 --------------


lo-


150 ------------ ------ ----------------------------------------------------------
300


150




Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Month


Figure 6-33. Orlando CSA-Seasonality simulations: monthly seasonal pattern of domestic
demand for air passengers from four U.S. regions.













50

S40

S30

S20
So-

o
S10


1 -10 -

S-20

-30

-40

-50


Orlando (5)
Northeast (1001)

[------------------------------------


50

40

30

20

10

0

-10

-20

-30

-40

-50


Orlando (5)
Midwest (1002)

--------------------------------------

--------------------------------------


I ------ ----------------------------


Orlando (5)

50 South (1003)

40

~ 30
I 20
40 --------------------------------------

30 --------------------------------------

20 --------------------------------------

10 -------i------
0-


Orlando (5)

50 est(1004)

40------------

30------------

20------------

10 -------- ----------------------------
10
C)


S-10 -------------------------- ----------------------------- ----------

-20 ------------ -20
S-20 -------------------------- --------- -20 -------------------------- ----------

S-30 ---------------------------------- -30 --------------

-40 --------------------------------- -40------------

-50 ---------------------------------- -50 --------------



Month Month

Figure 6-34. Orlando CSA-Seasonality simulations: percentage change from the monthly
average domestic demand for air passengers from four U.S. regions.







255


-

-

-

-

-

-


- - - - - - - -
- - - - - -
- - - - - - -


- - - - - - -
- - - - -

- - - - - - -










Tampa-St. Petersburg CSA-Seasonality

Similar to the previous CSAs discussed, the Tampa-St. Petersburg CSA shows that March

(increase) and September (decrease) recorded the largest variability from the average in each of

the four U.S. regions. Interestingly, the variability of the other 10 months is less than 10% in the

South and West regions. These results imply that there is no strong seasonality in demand for

passengers from these two U.S. regions. In contrast, the Midwest region exhibits six months with

a variability of at least 10% above or below the average. February (+12%), March (+40%), and

April (+11%) are above the average (263 passengers), while August (-12%), September (-31%),

and October (-10%) are below the average. Figure 6-33 presents simulated seasonal patterns of

demand for air passengers and Figure 6-34 shows the percentage change from the monthly

average demand for the four U.S. regions traveling to the Tampa-St. Petersburg CSA.


Northeast Midwest South West

450 ---------------------------------------------------------------------------------

400 ----------------- -- -----------------------------------------------------------
450-------------------

400
0 350




-2 0 --------- -- ----- --------- ----- ------ -- ------ -------------
200

150
100
10 -- ----- --- -----------




50 ------------------------------- -- ---- -------
S200 ------------------------------------------------------ --------------------------

150 ----------------------------------------------------------------------------------

100 ----------------------------------------------------------------------------------

50 -------------------------------------------------- ---- -----------------
0 1

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Month


Figure 6-35. Tampa-St. Petersburg CSA-Seasonality simulations: monthly seasonal pattern of
domestic demand for air passengers from four U.S. regions.












50

- 40

S30

S20-
1 0
S10


1 -10 -

-20

-30

-40
-50


Tampa-St. Pete (8)
Northeast (1001)

[------------------------------------


50
40
30
20
10
0
-10
-20
-30
-40
-50


Tampa-St. Pete (8)
Midwest (1002)

--------------------------------------
----------- -------------------------

I --------- -------------------------


Tampa-St. Pete (8)
50 South (1003)

40 --------------------------------------
30 --------------------------------------
40
~ 30

S20 -----------------------------------

10 ------
0----------------------------
S- I


Tampa-St. Pete (8)

50 est(1004)
40------------
30------------
20--- ---------
10 ----------- -------------------------
10
C)


-10 -
1.-10 --------------------*---- -10------ -
I
-10 ------------------ -- -- -0 ------------ -------
S-20 -------------------------- --------- -20 ----------------------------- -------
S-30 ---------------------------------- -30 --------------
-40 ---------------------------------- -40------------
-50 ---------------------------------- -50-------------



Month Month

Figure 6-36. Tampa-St. Petersburg CSA-Seasonality simulations: percentage change from the
monthly average domestic demand for air passengers from four U.S. regions.






257


-
-
-
-
-
-


- - - - - - - -
- - - - - -
- - - - - - -


- - - - - - -
- - - - - -


A __ WbiMLO 4










Jacksonville CSA-Seasonality

Three of the four U.S. regions show the same pattern of variability observed in the

previous sections. Demand from the Midwest, South, and West regions exhibited a large

variability from the average in March (increase) and September (decrease). The Northeast

region, on the other hand, presents the largest decline from the average (47 passengers) during

January (-20%) and the largest increase in March (+15%). Also, the Midwest region exhibits the

largest variability compared to the other U.S. regions. This U.S. region shows seven months with

a variability of at least 10% above or below the average. February (+12%), March (+31%), May

(+15), July (+14%), and August (+13%) were above the average of 33 passengers per month,

while September (-30%), October (-30%), and November (-25%) are below the average. Note

that although the West region shows high levels of variability in relative terms, demand from this

U.S. region is very small (only three passengers). Figure 6-37 exhibits simulated seasonal

patterns of demand for air passengers and Figure 6-38 illustrates the percentage change from the

monthly average demand for four U.S. regions traveling to the Jacksonville CSA.

Northeast Midwest South West
200------------------------------------------------------------------------
200


150

0_ 0
^ 10 ------- -------------------------------------------------------------

0 50 -------------------------------------




Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month

Figure 6-37. Jacksonville CSA-Seasonality simulations: monthly seasonal pattern of domestic
demand for air passengers from four U.S. regions.













50

- 40

S30

S20
M)O i
n 10 -

0

S-10

-20

-30

-40

-50


Jacksonville (9)
Northeast (1001)

[------------------------------------


50

40

30

20

10

0

-10

-20

-30

-40

-50


Jacksonville (9)
Midwest (1002)
--- --- --- --------------------------

--------------------------------------

--------------------------------------


I-------- -----------------------------


Jacksonville (9)
50 South (1003)

40

~ 30
I 20
40 --------------------------------------

30 --------------------------------------

20 --------------------------------------

10 ------ ---- ------------------
0 -


Jacksonville (9)

50 est(1004)

40------------

30 -- ----------

20 -- ----------

10 -------- ---- ---- -------------
10
C)


m -
S-10 -- --------------10---------- ---

S -20 -------------------------------------- -20 -------------------------- -

S-30 ------------------------- -30 -------------------

-40 --------------------------------- -40------------

-50 ---------------------------------- -50 --------------



Month Month

Figure 6-38. Jacksonville CSA-Seasonality simulations: percentage change from the monthly
average domestic demand for air passengers from four U.S. regions.







259


-

-

-

-

-

-


- - - - - - - -
- - - - - -
- - - - - - -


- - - - - - -
- - - - - -
I A JEL~:O 4










Fort Myers CSA-Seasonality

Three of the four U.S. regions show high levels of variability. Demand from the Northeast

Midwest, and South regions exhibited a large variability from the average in the winter months

(increase) and summer months (decrease). Recall that results from Fort Myers -West pair

indicate that estimates are not valid due to the lack of stability of its speed of adjustment

coefficient and no inferences from this U.S. region would be valid.

The Midwest region exhibits the largest variability compared to the other three U.S.

regions. This U.S. region shows that in March demand increases to 86% above the average and

in September it decreases by half (54% below the average). Figure 6-37 exhibits simulated

seasonal patterns of demand for air passengers and Figure 6-38 illustrates the percentage change

from the monthly average demand from four U.S. regions traveling to the Jacksonville CSA.


Northeast Midwest South West



a 250 --------------------- ------------------------------------------------------------

8 200 ---------- -------------- ---------------------------------------------------------
300
250




150 --------------- --- ----- ----------------------------------------- -----------
0o
1 200------- --------------------------------------- ------------









0 1 1 1 1
O i I

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Month


Figure 6-39. Fort Myers CSA-Seasonality simulations: monthly seasonal pattern of domestic
demand for air passengers from four U.S. regions.














100

- 80

S60

40

20

S0

-20

-40

-60

-80

-100







100

80

S60

P 40

20


Fort Myers (14)
Northeast (1001)

--------------------------------------

--------------------------------------

-------- ----------------------------



..-- --------------------------

----------------- ----------


--------------------------- ----------

--------------------------------------

-------------------------------------






Fort Myers (14)
South (1003)

--------------------------------------

--------------------------------------

-------- ----------------------------

---- ---------------------------


Fort Myers (14)
Midwest (1002)

-------- ----------------------------

-------- ----------------------------

---- ----------------------------

-- -----------------------


100

80

60

40

20

0

-20

-40

-60

-80

-100








100

80

60

40

20

0


-20 --- -20 ---------

o -40 -40 -------------------- ---

-60 ----------------------------------- -60 --------------------------------------

-80 -----------------------------------80 -------------

-100 --------------------------------- -100------------




Month Month

Figure 6-40. Fort Myers CSA-Seasonality simulations: percentage change from the monthly
average domestic demand for air passengers from four U.S. regions.







261


-------------- - -

----------------------- ----------

--------------------------------------

--------------------------------------

--------------------------------------
a >- i = -5 to &< 3 > o




Fort Myers (14)
West (1004)


-I------------------------------------
-------- ----------------------------

-- --- -------------------------

-**-- -* ----------------------









Simulations for Fire, Rainfall, and Temperature

Three set of simulations were conducted on the fire, rainfall, and temperature variables for

those CSA-ORG pairs on which these variables were statistically significant. All three variables

were set to fluctuate between a 20% decrease to a 20% increase in increments of 5%. Simulation

results indicate that the demand response to changes in these variables is extremely small relative

to other structural variables. Hence, no discussion for each pair is presented.

Chapter Summary

Five sets of simulations were presented in detail for each destination CSA. Income, airline

ticket prices, dummy variables for terror, hurricanes, and monthly seasonality were set to

different scenarios to analyze the demand response to a particular shock. Different variations in

income and airline ticket prices were simulated, but discussions centered in the variation most

likely to occur. For example, discussion in the income section centered on a 3% increase in

income, while the airline prices section discusses an extreme case where a 20% increase in

airline ticket prices occurs. Even at this unlikely event, demand for air passengers to Florida is

inelastic to changes in airline ticket prices. Therefore, simulation results indicate there is a

greater response to changes in income than to changes in airline ticket prices.

Figure 6-41 presents demand response from each U.S. region traveling to Florida and

Figure 6-42 illustrates response in demand for passengers traveling to five destination CSAs

given a 3% increase in income and airfare. Even with increasing fuel prices, a 3% increase in

airfares is more likely to occur than the 20% discussed in previous sections given stiff

competition between low cost and legacy airlines.














Income






33-1001*






S33-1002






S33-1003*






33-1004*


Airline Ticket Prices OLong run

Short run
-----------3----


33-1001*


----i------~------r------i------I


O Short mn

Long run
*-----I------I----- -I-----r-----l
SI I




I I I I--
---------------- ----------1
SI I
I I
I I
I I
I I
I I
I I
I I I I
I I I I I


I I I I I



SI I
SI I
SI I
SI I

SI I
SI I
I I I II I


I I I I I
I I I I I
I I I I I
I I I I I
I I I I I
I I I I I
I I I I I
I I I I I
-----~--~--------,------I



I I I I I
r mI I I

I I I I I
I I I I I
I I I I I


I I I I I I

0 20 40 60 80 100 0 20 40 60 80 100



Increase in Demand Decrease in Demand

(Passengers per 100,000) (Passengers per 100,000)


Note: CSA-ORG pair empty cell indicates variable is not a significant demand driver. Asterisk (*) denotes
short run adjustment is statistically different from the long run adjustment given a two-tailed test at 95%
confidence.

Figure 6-41. Comparison of the demand response from each U.S. region traveling to Florida to a

three percent increase in income and airline ticket prices.


Also, the simulations performed in the absence of the 9-11 terrorist attacks indicate that



Florida was greatly affected by the terrorist attacks. Demand for passengers could have been up



to 14% higher in the absence of the 9-11 terrorist attacks. Moreover, analysis by destination CSA



indicates that the impact has been greater in some U.S. regions, especially in the Midwest and



South regions. Simulations also indicate that the Orlando CSA is the only destination that has not



shown signs of recovery from any of the U.S. region. Table 6-1 presents a comparison of the



simulated demand in the absence of the 9-11 attacks and simulated demand in the presence of the



9-11 terrorist attacks.


33-1002






33-1003*






33-1004















Income


0 Short run


CSA-ORG Pair Long run
-------r-------r-------r- ----1

1-1001*

I II
1-1002*


--------L-------L-------.-------l



1-1004*"




--------r--------r-------r-------r
SI I







S----iI- -i--- -- n-- ----I-- ------

























--- -a----w-r--- ---- ------- .-a-a---
I I
5-1003


-1004






8-1002
**------- ------- ------- -------4



I I II








9-1002*
-. ------L------- .------- .-------
I I I











9-1003* .. _


9-1004*
(--P-s----ne -----p---e




14-1001*









14-1004



0 10 20 30 40


Increase in Demand

(Passengers per 100,000)
(Passengers per lO0,O00)


Airline Ticket Prices


SLong run


CSA-ORG Pair U Short run
---- I--r-------r-------r-------

1-1001* I I
II -;-------r-------r-------!


1-1002*


1-1003*


1-1004


5-1001


5-1002


5-1003


5-1004


8-1001


8-1002


8-1003


8-1004


9-1001


9-1002


9-1003


9-1004*


14-1001*


14-1002


14-1003


14-1004


L


0 10 20 30

Decrease in Demand

(Passengers per 100,000)


Note: CSA-ORG pair empty cell indicates variable is not a significant demand driver. Asterisk (*) denotes
short run adjustment is statistically different from the long run adjustment given a two-tailed test at 95%
confidence.

Figure 6-42. Comparison of the demand response in each CSA-ORG pair to a three percent

increase in income and airline ticket prices.





264


I


.. .- .- .-.
II I
I I I I


I I I
3
I I I
--------L-------L-------L--------
I I I I


E-------J----------------------
I I I I
I I I
I I I I
I I I I







-------- --I -------- ----
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I














]
I I I I






-------------------------1--
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I
















3I
-- -- --L --- ---- J
I I I I
I I I I
I I I I
I I I I








-------- -----I---------- .---
I I I I














---------~------ -------- .-------r
I I I I










]



.-------.------- .------- .-------r
I I I I
II I







.I1
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I








II
II I
II I
S I I I
I I I I
--------------L---L ---I-- -J
I I I I
I I I I
I I I I
I I I I

------------I----I--------Jl
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I
I I I I
SI I


I I I I
I I I I
I I I I
I I I I









Another interesting simulation was the one performed to set a scenario where hurricanes

are not present, and hence, do not affect Florida. The response is very small (less than 1.0% in

most cases) suggesting that hurricanes do not have a great impact in the demand for air

passengers. Table 6-3 presents a comparison between simulated demand in the absence and

presence of hurricanes.

Finally, simulations on seasonality show that the South Florida CSA mimics seasonal

patterns for the entire state. All destination CSAs exhibited the largest increase in demand in

March and the largest decrease during September. These findings are consistent with the

seasonality values reported by Visit Floridac. Also, the Orlando CSA is the only destination

where seasonality is less variable, while the Fort Myers CSA exhibited strong patterns of low

demand during summer and high levels during winter months. Within U.S. regions, the Midwest

region experienced stronger patterns of seasonality in all destination CSAs, while the South

region showed the weakest patterns.













Table 6-1. Comparison between the simulated demand in the presence of the 9-11 terrorist
attacks and the simulated demand in the absence of 9-11 terrorist attacks.



1 1 -

0 .31?( U UV
/ 0 0^C G Cy C ^ C ^ CC (
*s &D o &r- o & t c^pr t t rt tT
^ 0
f0


0 t C ~ Ht 0Y 0
0i 0i .. 0 00

0 -= t= 0 0S
-g -- --- _P 0
< 1 C. 1-S 10* .3 ^3 1- 1 3
*U- *M 'g *>2 M2 *S *< 22 0 T,^ WC,^
oe a ma .a.. ,


911 965
392 476
750 851
156 163
2,209 2,455

592 684
451 547
516 590
134 165
1,692 1,985

284 307
263 325
347 405
57 64
951 1,101


6.25 0.35 64.94 68.79 3.86
3.02 0.53 27.31 33.18 5.86
8.77 1.04 85.06 96.49 11.43
1.29 0.05 13.58 14.18 0.60
19.33 1.98 190.89 212.64 21.75

4.43 0.60 42.18 48.73 6.55
3.47 0.61 31.41 38.14 6.73
6.08 0.77 58.46 66.88 8.42
1.30 0.24 11.67 14.33 2.66
15.28 2.21 143.72 168.08 24.36

1.99 0.15 20.23 21.87 1.65
2.06 0.40 18.33 22.68 4.35
4.17 0.59 39.38 45.92 6.54
0.50 0.05 4.98 5.53 0.55
8.73 1.19 82.92 96.00 13.09


0.35 0.04 3.37 3.81
0.41 0.10 3.40 4.47
2.10 0.40 18.76 23.15
0.02 0.00 0.24 0.27
2.88 0.54 25.77 31.70


1-1001
1-1002
1-1003
1-1004
Total

5-1001
5-1002
5-1003
5-1004
Total

8-1001
8-1002
8-1003
8-1004
Total

9-1001
9-1002
9-1003
9-1004
Total

14-1001
14-1002
14-1003
14-1004
Total


7.50 7.50 0.00 0
10.22 11.39 1.17 11
11.22 11.78 0.56 5


2.63 2.79 0.16 28.94 30.67


1.73 6


33-1001 1,955 2,160
33-1002 1,335 1,640
33-1003 2,106 2,381
33-1004 370 409
Total 5,765 6,590


12.67
8.46
21.71
2.92
45.75


14.00 1.33 139.35 153.96 14.61
10.39 1.93 93.01 114.30 21.28
24.54 2.84 238.78 269.99 31.21
3.24 0.31 32.15 35.61 3.46
52.17 6.42 503.29 573.86 70.57


Total passengers for the 11-year period from 1996 to 2006
r Percentage change = [(Difference/ Simulated Total Demand with 9-11 Attacks)*100]


0.68 0.00
1.04 0.11
1.07 0.05













Table 6-2. Comparison between the simulated demand in the presence of hurricanes and the
simulated demand in the absence hurricanes during the hurricane season (June-

November).

I) I1 I)a
eS p ao P^ u u

E.2 E -E E ca aa
cli o o 5

S-p Je b a*^*a ?
Q ?-g
-- 0^ 0^- 0 s c u" ^ '
-E?



IS
Eo on o
I -s I E QE -5 -I
" "2 r "b: a" a" s s
-o -c -- cc ~c
o o 11
.0 .0 0 Q

C U5 eo e
-. o
cJ o a ,0_E _&n&
U m~


829 830 540
331 331 528
712 712 859
148 148 659
2,020 2,021 0

573 574 540
427 428 528
505 505 859
129 129 659
1,633 1,635 0

270 270 540
237 237 528
335 336 859
54 54 659
897 897 0

47 47 540
49 49 528
167 167 859
3 3 659
266 266 0

84 84 540
100 101 528
84 84 859
2 2 659
270 270 0

1,817 1,820 540
1,172 1,173 528
2,031 2,033 859
349 350 659
5,370 5,376 0


3,116,469
1,215,424
4,257,189
679,337
9,268,420

2,156,547
1,571,935
3,024,341
593,059
7,345,882

1,018,805
873,290
2,010,978
247,860
4,150,933

178,662
179,900
1,002,046
11,880
1,372,488

315,203
370,542
504,784

1,197,620

6,834,664
4,310,170
12,157,881
1,605,444
24,908,159


3,133,581 3,137,018 17,112 20,549 0.55 0.66
1,223,979 1,224,401 8,555 8,977 0.70 0.74
4,279,337 4,281,623 22,148 24,434 0.52 0.57
682,248 682,458 2,911 3,121 0.43 0.46
9,319,146 9,325,500 50,726 57,081 0.55 0.62

2,164,748 2,168,734 8,200 12,187 0.38 0.57
1,578,831 1,580,480 6,896 8,544 0.44 0.54
3,034,143 3,034,575 9,802 10,235 0.32 0.34
594,844 594,914 1,786 1,855 0.30 0.31
7,372,566 7,378,704 26,684 32,822 0.36 0.45

1,021,887 1,022,389 3,081 3,584 0.30 0.35
876,431 876,584 3,141 3,294 0.36 0.38
2,017,135 2,017,844 6,158 6,866 0.31 0.34
248,869 248,869 1,010 1,010 0.41 0.41
4,164,323 4,165,687 13,390 14,754 0.32 0.36

178,748 178,751 86 90 0.05 0.05
180,540 180,724 640 824 0.36 0.46
1,004,828 1,005,492 2,782 3,446 0.28 0.34
11,869 11,869 -11 -11 -0.10 -0.10
1,375,984 1,376,836 3,496 4,348 0.25 0.32

316,501 316,890 1,298 1,687 0.41 0.54
371,405 371,585 864 1,043 0.23 0.28
506,479 506,784 1,695 1,999 0.34 0.40

1,201,502 1,202,368 3,881 4,747 0.32 0.40

6,868,106 6,881,234 33,442 46,569 0.49 0.68
4,332,185 4,334,240 22,015 24,070 0.51 0.56
12,213,773 12,226,169 55,892 68,288 0.46 0.56
1,611,516 1,613,379 6,072 7,935 0.38 0.49
25,025,580 25,055,022 117,421 146,862 0.47 0.59


Note: Averages of months included in the hurricane season (June to November)
*Short run is statistically different from long run adjustment
r Percentage change = [(Difference in Short Run/ Simulated Avg. Demand during Hurricane Season w/ Hurricanes)* 100]
rr Percentage change = [(Difference in Long Run/ Simulated Avg. Demand during Hurricane Season w/ Hurricanes)* 100]
CSA-ORG has unstable estimates. No inferences are valid.


1-1001*
1-1002*
1-1003*
1-1004*
Total

5-1001*
5-1002*
5-1003
5-1004
Total

8-1001*
8-1002*
8-1003
8-1004
Total

9-1001*
9-1002
9-1003
9-1004
Total

14-1001
14-1002*
14-1003*
14-1004C
Total

33-1001*
33-1002
33-1003*
33-1004*
Total


824
329
708
147
2,009

571
425
503
129
1,627

270
236
334
54
894

47
49
167
3
265

83
100
84
2
269

1,808
1,166
2,022
348
5,344









CHAPTER 7
SUMMARY, CONCLUSIONS, IMPLICATIONS, AND FUTURE RESEARCH

The airline industry serves as a vital link between consumers and the tourism industry.

Since nearly one half of the tourists traveling to Florida use air transportation, having

quantitative methods that help identify the factors that affect demand for air passenger traffic are

essential to planning and infrastructure development for the tourist sector and the state in

general. Also, knowing how air passenger traffic responds to unexpected shocks, such as tropical

storms, helps quantify the economic impact on the state economy. It also provides guidelines for

emergency planning, risk assessment, and prevention management that could be implemented in

case of the occurrence of such devastating events.

Although forecasting procedures have dominated the tourism demand analyses,

identification of the determinants of demand have been seldom analyzed. Also, data availability

and the lack of a standardized system of accounting between the tourism and transportation

industries have been cited as a limitation. Most of the studies use annual data which hinder the

ability to model seasonality, an intrinsic characteristic of tourism demand. Nevertheless,

researchers agree that econometric models are vital for policy and strategic planning and risk

management. Moreover, those few analyses aimed to identify the factors influencing demand

have either focused on micro tourism demand analysis for a particular city or attraction or on

macro tourism demand at a country level. This project developed a model that attempts to fill the

gap by addressing those issues and hence contribute to the tourism research literature.

The primary objective of this research was to develop an understanding of the factors

influencing the domestic airline travel demand to Florida and to the top five destination CSAs in

Florida: South Florida, Orlando, Tampa-St. Petersburg, Jacksonville, and Fort Myers. Demand

factors such as airline ticket prices, income, terrorism, seasonal patterns, hurricanes, wildfires,









and advertising expenditures were included in the analysis. An econometric model was

developed to help explain the driving forces for domestic airline travel demand to Florida and to

five destination CSAs in Florida. Significant drivers of demand were identified for each U.S.

region traveling to Florida and its top five destination CSAs.

A partial adjustment model was estimated under the initial hypothesis that air passengers

are unable to respond immediately to changes in demand factors and hence the full response is

realized not until subsequent periods. The SUR-AR1-ALL estimation approach performed better

than other three estimation alternatives considered since it yielded the fewest unstable speed of

adjustment coefficient estimates. Results were presented and discussed for each destination CSA

and for Florida (CSA=33). The discussion focused on statistical significance of various estimates

such as the elasticity of adjustment estimates and the economic, weather, and terrorism variables.

Also, the (1-0) coefficient was analyzed and compared for each U.S. region across destination

CSAs and for each destination CSA across U.S. regions.

Finally, each demand equation was shocked to determine its short and long run

responsiveness to changes in demand drivers. Five sets of simulations were presented in detail

for each destination CSA. Income, airline ticket prices, dummy variables for terror, hurricanes,

and monthly seasonality were set to different scenarios to analyze the demand response to a

particular shock.

Conclusions

Consumer demand theory suggests that prices and income, and to a lower extent, political

and social stability at the destination, constitute major demand drivers in developing nations. It

also suggests that there are certain instances where consumers are unable to immediately adjust

to changes in their demand determinants. Adjustment costs, incomplete information, and habits

are some reasons why consumers exhibit rigidity in their response. Both domestic and









international passengers traveling to Florida may display habit due to ownership of real estate or

time shares in the state, institutional structure such calendar holidays that attract visitors, school

vacations, sense of security and safety, and a supply of unique products (e.g., beaches,

attractions). In addition, consumers may return to Florida because they are already familiarized

with the environment and services provided which reduces search costs and also creates a sense

of comfort and security. Finally, the system in place provides little or no flexibility to consumers

when hurricanes, wildfires, and terrorist attacks occur.

Although the initial hypothesis suggests that demand to each destination CSA is different,

the Florida CSA was estimated to present a general summary of the demand for air travel. While

this destination CSA helps explain certain characteristics of Florida as a whole, it was shown that

there is an added benefit to analyze the destination CSAs individually. Results showed that some

destination CSAs vary significantly from one another. Certainly, the South Florida CSA

somewhat mimics the behavior observed in the Florida CSA. This result was rather expected

because the South Florida CSA is the largest of the five destination CSAs studied in this project

and could overshadow the others. Some commonalities across destination CSAs was the habit

observed in the Northeast region which exhibited some degree of habit to all destination CSAs

but one (Jacksonville CSA). Other similarities where found regarding the positive effect of

income in demand and the dramatic impact of the 9-11 terrorist attacks across all destination

CSAs. Main differences across U.S. origins and destination CSAs were observed in airline ticket

prices, and to a lesser degree, tropical storms and seasonality.

Despite the different results observed across U.S. origins and destination CSAs in terms of

habit persistence or rigidity of demand, the speed of adjustment coefficient suggests that most of

the demand is driven by current events (i.e., structural or static component of the model). Speed









of adjustment coefficient estimates range from 0.60 (Orlando-Northeast) to 0.86 (South Florida-

South) on those CSA-ORG pairs where the estimate was significant.

Ownership of property, timeshare holdings, or other interests could explain why response

of the air passengers from the Northeast region tends to be more rigid. If air passengers own

property in the state or buy a timeshare in a Florida resort, they will more likely visit the state in

a regular basis regardless of the conditions of the economy or weather. Also, most of the time

share resorts are located in the Orlando CSA and it could explain why this destination CSA

exhibited more habit than any other destination CSA in Florida.

Seasonality also provided a wide range of results that indicated high seasonality present in

the Fort Myers CSA and a more steady flow of passengers traveling to the Orlando CSA year-

round. The Orlando CSA has successfully lowered the variability in levels of passengers by

expanding the services and attractions to other key demographics such as young and single adults

and business travelers. Within the U.S. regions, the Midwest region experienced stronger

patterns of seasonality to all destination CSAs, while the South region showed the weakest

patterns. Nevertheless, the common trend shown across all destination CSAs indicate that the

largest increase in demand occurs in March, while the largest decrease occurs during September.

These findings are consistent with seasonality values reported by Visit Floridac. Large volume of

spring breakers traveling to Florida in March and the beginning of classes in September could

explain the variability in demand during those months.

Estimation results indicate that income is the major determinant in the demand for air

passengers traveling to Florida and its specific destination CSAs. These results imply that the

state of the economy directly affects domestic demand and hence the tourism industry in Florida.

The 2008 economic environment poses a threat to the largest industry in Florida and the









livelihood of millions of Floridians who directly or indirectly depend on the tourism industry.

Although the country is currently experiencing a slowdown in the economy, early 2007 estimates

suggest that Florida has not experienced a dramatic decline in domestic passengers. In addition,

Florida is benefiting from a weak dollar, which has attracted international demand. International

visitors tend to stay longer and spend more than domestic visitors, according to Visit Florida"

(2006).

Overall, coefficient estimates suggest that domestic demand for air travel to Florida is price

inelastic. Changes in demand are proportionately smaller than changes in airline ticket prices. At

the destination CSA level, estimations indicate that airfare is a significant determinant of demand

in only eight of 20 CSA-ORG pairs. Also, response tends to be inelastic to changes in airline

prices in seven of eight pairs. These findings would contradict economic theory which suggests

that demand is elastic when possibilities of substitution are high. Since Florida is a premiere

leisure destination, airlines embark in price wars to attract passengers and fill seats. Airlines

realize that consumers have a large number of carriers to choose from. Such a large number of

possibilities would imply that domestic demand for air travel to Florida is price elastic. But there

is another level of substitution that could explain the results found in this study. Florida has

unique characteristics that are difficult to find elsewhere. One can expect that consumers cannot

easily substitute a leisure trip to a different destination. As Brons et.al (2002) explained, multiple

levels of substitution can be distinguished in aviation, one of them being the characteristics of the

destination. The findings of this study are consistent with Brons et.al (2002) who used utility

maximization to rationalize the reasons why demand for air travel can be inelastic in markets

where there is a high level of intra-modal substitution. Even if prices increase, consumers still

find a higher level of utility by traveling to Florida rather than to another destination.









The Orlando CSA showed that none of the origin U.S. regions is sensitive to price changes

in airline ticket prices. These results could be attributed to the fact that data used in the model

include both leisure and business travelers. Economic theory suggests that business and leisure

travelers respond differently to changes in economic factors such as airline ticket prices. Demand

for leisure travel tends to be more sensitive to changes in airline ticket prices than demand for

business travel. The Orlando CSA is also a popular destination for the business travel

demographic and hence, the opposite response of each demographic could be offsetting the

significance of the price variable in this destination CSA.

Furthermore, results indicate that domestic demand for air passengers traveling to Florida

has not fully recovered from the 9-11 terrorist attacks. In the absence of the attacks, more than

70.57 million passengers would have traveled to Florida since 2001, equivalent to a 14%

increase from actual demand. Although domestic demand from some U.S. regions traveling to

certain destination CSAs has shown signs of recovery, simulations suggest that the Orlando CSA

is the only one that has not shown signs of full recovery. In addition, the Midwest region has

exhibited the largest differences compared to other origin U.S. regions.

In regard to weather-related variables, results suggest that "different-from-the-norm"

temperatures and rainfall at the origin do not affect domestic demand for air travel significantly.

Notwithstanding, one must be careful with this assessment since temperature and rainfall records

from a few cities were used to generalize to an entire U.S. region. In addition, impact of

hurricanes and wildfires affecting Florida on domestic demand for air travel is very small.

Simulations suggest that, in the absence of hurricanes, domestic demand would have increased

by 0.60% over an 11-year period. The percentage increase in domestic demand or air travel in

the absence of wildfires was negligible.









Finally, results also showed that advertising expenditure estimates were not well-behaved

in almost all CSA-ORG pairs. Since advertising expenditures are highly seasonal, the dummy

variables for monthly seasonality may be capturing the effect of the advertising variable.

Implications

Tourism will continue to be a major source of employment in Florida but the environment

surrounding the airline industry could dramatically affect Florida's maj or industry in the long

run. The major concern of the tourism industry in Florida is that while consumers will still go on

vacation, they will decide to stay closer to home to save on transportation costs, which in some

cases represent up to 40% of the vacation budget (Visit Florida" 2006). Since air travel tends to

be a discretionary item, price incentives may be necessary to maintain stable demand during

economic downturns. As disposable income shrinks due to increases in fuel and food costs,

consumers allocate less income to expenses such as eating out or traveling. Currently, major

tourist providers in Florida are lowering rates at hotels, resorts, and theme parks to stimulate

demand. Similar trends have been observed in Las Vegas, California, and Hawaii where lodging,

gaming, and major cruise companies are reducing prices.

Also, due to stiff competition observed in the airline industry between low-cost and legacy

airlines, airline ticket prices are not expected to dramatically increase. Media reports indicate that

legacy airlines are cutting costs by cancelling routes, reducing the speed of jets, making aircraft

lighter, cutting free on-board services, and the most recent and least popular, charging for

luggage to offset high fuel prices. And while legacy airlines are cancelling service between

secondary markets and leisure destinations such as Orlando, low cost airlines have opened new

routes. Legacy airlines are been affected on two fronts: their inability to hedge fuel prices and the

lack of brand loyalty among consumers. Therefore, low cost airlines are benefiting. Southwest

Airlines has effectively hedged its fuel costs and has largely benefited from current market









conditions. Also, consumers may be thinking twice before committing to travel. In a rush to fill

seats, airlines are forced to match competitor's prices or withdraw from less profitable markets.

Seasonal patterns are an intrinsic element of products offered by Florida. Natural migration

from the Northern states during winter, school holidays, and supply of unique products such as

warm weather and beaches trigger these seasonal patterns that affect domestic demand for air

transportation. Strong seasonal patterns present in Fort Myers have multiple implications in the

destination's economy including an increased supply of services and number of employment

opportunities during winter. By the same token, significant decline in air passengers during

summer represents a slowdown in the area's economy. On the other hand, weaker seasonal

patterns such as the ones exhibited in the Orlando CSA suggest that its economy is not

considerably affected by seasonal patterns given that there is a steady need for services

throughout the year.

While wildfires and hurricanes could have a devastating effect on the infrastructure of the

tourism industry, they did not have a significant impact on the demand for air travel. This

statement regarding wildfires is consistent with the conclusions provided in the study conducted

by Thapa, Holland, and Absher (2003). Nevertheless, no major damages due to these natural

events were reported by tourism providers in the state during the period covered by this project.

This could obviously change if a major hurricane or wildfire directly affects points of access to

tourism sites (i.e., airports).

Unlike hurricanes and wildfires, terrorist attacks are not acts of nature and could be

prevented up to a certain extent. Such attacks pose a major threat to both the airline and tourism

industries. Although most of the initial decline in demand was driven by the fear of flying, some

of the decline has been attributed also to the more stringent security measures placed as a









response to the 9-11 terrorist attacks. Increased security measures and tighter visa requirements

imposed after the 9-11 terrorist attacks by the U.S. Department of Homeland Security also

contributed to the decline of international travel to the United States. Technological

advancements in passenger screening without undermining security should be a priority of

policymakers. Such advancements could provide a streamline process that would reduce queue

lines for both domestic and international visitors at airport security checkpoints. Hassle-free

travel experiences contribute to the traveler's overall satisfaction and consumers will be more

likely to repeat the experience.

And while Florida offers more commercial airports than any other state, it faces major

challenges related to air transportation and tourism. Increases in air passenger traffic combined

with security concerns bring nuisances to air travelers, such as delays due to traffic overflow,

computer glitches, and security checkpoints. More runways, gates, and security checkpoints at

Florida airports could ease such increase in passenger traffic and reduce travel time. Such

expansion should take into account any environmental concerns that may arise. Proper

development is paramount to minimize any negative impact on surrounding communities.

Moreover, cancellation of point-to-point routes means passengers will need to connect

flights through hubs, and hence, increase their travel time. Weary air travelers may give up a

long-haul flight to Florida for a less stressful trip to an attraction closer to home.

Future Research

This project could be improved on many fronts and also extended to multiple areas. One

major limitation of the dependent variable used in this project is that it does not provide

information about the passenger. It fails to identify the passenger as a person traveling for leisure

or business or a resident returning home. Nevertheless, the variable was useful to identify general

patterns of air travel to Florida. One possible improvement could be to identify the type









passengers to obtain more insights related to their travel behavior: business versus leisure

travelers. Also, are leisure travelers returning to Florida more than once a year because they are

engaging in shorter trips more frequently? Or, are they deciding to split the time in Florida and

somewhere else? If so, what does this other destination offer the customer that has made him

spend less time in Florida?

As shown in Chapter 3, the major tourism companies and the state carry out large

advertising campaigns worth millions of dollars every year. Improvements could be made in

terms of the advertising variable. Since the monthly seasonality variables could have shadowed

the impact of the advertising expenditures, an estimation procedure that could isolate the effect

of each variable could be pursued. Also, other variables could be included in the model. For

example, cost of transport substitutes (e.g., car) and cost of complementary goods (e.g., hotel

accommodations, theme parks) could be relevant factors in determining the demand for air

passenger traffic to Florida.

In regard to international passenger demand, an initial step has been made here by

presenting some descriptive statistics of passengers from different world regions traveling to

Florida. The next logical step would be to investigate the determinants of international air

passenger demand in order to determine their relative importance.

Furthermore, Florida faces greater competition not only from domestic destinations but

also emerging tourist destinations. Major tourist conglomerates have announced plans to build

new attractions in China and Dubai. The potential impact on tourism demand in Florida not only

of international tourists but also adventurous domestic travelers wanting to visit new places

could be analyzed.











APPENDIX A
DESCRIPTION OF DESTINATION CSA AIRPORTS


Airport Airport
CSA Description of CSA Code Name City
1 C ..+1, 71 ;n "-0 A


MOUUIi BrU1cUCa B.- Kn
Miami-Miami Beach-Kendall


Fort Lauderdale-Pompano Beach-
Deerfield Beach
West Palm Beach-Boca Raton-
Boynton Beach



Orlando CSA
Orlando-Kissimmee



Deltona-Daytona Beach-
Ormond Beach


Palm Coast


Tampa-St. Petersburg CSA


Jacksonville CSA


TMB
MIA
OPF
MPB
TNT
HST

FLL
PBI
BCT



ISM
MCO
SFB

DAB
DQN
FL4



PIE
MCF


NZC
NRB
UST
NIP
JAX


Kendall-Tamiami
Miami Intl
Opa Locka
Miami
Miami
Homestead
Fort Lauderdale/
Hollywood Intl
Palm Beach Intl
Boca Raton



Kissimmee Gateway
Orlando Intl
Orlando Sanford Intl

Daytona Beach Intl
New Smyrna Beach
Bunnell


St Petersburg-
Clearwater Intl
Mac Dill AFB


Cecil Field (VQQ)
Mayport
St. Augustine
Jacksonville
Jacksonville Intl


Miami
Miami
Miami
Miami
Miami
Homestead

Fort Lauderdale
West Palm Beach
Boca Raton



Orlando
Orlando
Orlando

Daytona Beach
New Smyrna Beach
Bunnell



St Petersburg-Clearwater
Tampa


Jacksonville
Mayport
St. Augustine
Jacksonville
Jacksonville


14 Fort Myers CSA


RSW Southwest Florida Intl
FMY Fort Myers


Fort Myers
Fort Myers










APPENDIX B
TRANSFORMATION OF VARIABLES: TSP PROGRAMS

OPTIONS DOUBLE MEMORY=200 LIMWMISS=0 LIMWARN=0;
TITLE 'JOSE CAZANOVA PH.D.-DEMAND FOR DOMESTIC AIR PASSENGER TRAFFIC TO FLORIDA';
? AIRFINALMODEL#4.TSP;
IN'C:\ZTOURISM\ZCAZANOVA\TSPPRG\FINAL\FINALAIRMODEL\AIRFINAL FL';


? ADJUSTING TO PASSENGERS PER 100,000 POPULATION

DOT(CHAR=O) 1001 1002 1003 1004;
HZPOP .0 = HZPOP_.0/1000000;
DOT(CHAR=D) 1 5 8 9 14 33;
P.D .0 = P.D .0/1000000;
PP.D_.0 = P.D_.O/(HZPOP_.0/100000);
LPP.D_.0 = LOG(PP.D_.O);
ENDDOT; ENDDOT;


? CREATING AVERAGE PRICES TO DESTINATION I FROM REGION J

DOT(CHAR=D) 1 5 8 9 14;
DOT(CHAR=O,VALUE=J) 1001 1002 1003 1004;
? ROUND TRIP AVERAGE PRICE;
FR.D_.O_2 = (J=1001)*[ FR.D_11_2 + FR.D_12_2 ]/2
+ (J=1002)*[ FR.D_21_2 + FR.D_22_2 ]/2
+ (J=1003)*[ FR.D_31_2 + FR.D_32_2 +FR.D_33_2]/3
+ (J=1004)*[ FR.D_41_2 + FR.D_42_2 +FR.D_43_2]/3;
FR.D .O02 = (FR.D_.O02)*2; ? MAKING THIS A ROUND TRIP FARE;
ENDDOT;ENDDOT;
? CREATING FARE FOR ALL FLORIDA CSA=33 FROM EACH ORIGIN REGION J;
DOT(CHAR=O,VALUE=J) 1001 1002 1003 1004;
FR33 .0 2 = (J=1001)*[FR1_.0 2*P1_.O + FR5_.O02*P5_.O + FR8_.O02*P8_.O + FR9_.O02*P9_. +
FR14 .0 2*P14_.O]/P33_.0 + (J=1002)*[FR1_.0_2*P1_.O + FR5 .O02*P5_.0 + FR8 .O02*P8_.0+
FR9 .O02*P9_.0 + FR14 .0_2*P14_.O]/P33_. + (J=1003)*[FR1_.0 2*P1_.0 + FR5 .O02*P5_.0+
FR8 .O02*P8_.0 + FR9 .O02*P9_.0 + FR14 .0 2*P14_.O]/P33_.0 + (J=1004)*[FR1_.0 2*P1_.0 +
FR5 .O02*P5_.0 + FR8_.O02*P8_.0 + FR9_.O02*P9_.0 + FR14_.02*P14_.O]/P33_.0;
ENDDOT;
? NORMALIZING FARES USING THE NATIONAL CPI AFARE;
DOT(CHAR=D) 1 5 8 9 14 33;
DOT(CHAR=O,VALUE=J) 1001 1002 1003 1004;
AFR.D .O 2=FR.D .O 2/CPI_AFARE;
?AFR.D .0 1=FR.D .O 1/CPIAFARE;
LAFR.D_.O_2=LOG(AFR.D_.0_2);
?LAFR.D_.O_1=LOG(AFR.D_.O_1);
ENDDOT;ENDDOT;


? CREATING THE POPULATION VARIABLE VALUE FOR EACH MONTH
? ADJUSTING PERCAPITAINCOME WITH POPULATION TO LOGLINEAR FORM ;

DOT(CHAR=O,VALUE=J) 1001 1002 1003 1004 1000;
ZPOP .0 = POP .0 MTT;
OLSQ(SILENT) ZPOP_.0 C TT;










FOREST HZPOP .O;
ENDDOT;
PCDI 1001=(PCDI11*POP11 + PCDI12*POP12 + PCDI13*POP13 + PCDI14*POP14 + PCDI15*POP15 +
PCDI16*POP16 + PCDI21*POP21 + PCDI22*POP22 + PCDI23*POP23)/HZPOP_1001;
PCDI_1002=( PCDI41*POP41 + PCDI42*POP42 + PCDI43*POP43 + PCDI45*POP45 + PCDI61*POP61 +
PCDI62*POP62 + PCDI63*POP63 + PCDI64*POP64 + PCDI65*POP65 + PCDI67*POP67 )/HZPOP 1002;
PCDI_1003=( PCDI31*POP31 +PCDI32*POP32 +PCPI33*0 +PCDI34*POP34 +PCDI35*POP35 +
PCDI36*POP36 + PCDI37*POP37 + PCDI38*POP38 + PCDI39*POP39 + PCDI51*POP51 + PCDI52*POP52 +
PCDI53*POP53 + PCDI54*POP54 + PCDI71*POP71 + PCDI72*POP72 + PCDI73*POP73 +
PCDI74*POP74)/HZPOP 1003;
PCDI_1004=( PCDI1 *POP1 + PCDI2 *POP2 + PCDI81*POP81 + PCDI82*POP82 + PCDI83*POP83 +
PCDI84*POP84 + PCDI85*POP85 + PCDI86*POP86 + PCDI87*POP87 + PCDI88*POP88 + PCDI91*POP91 +
PCDI92*POP92 + PCDI93*POP93)/HZPOP 1004;
? CREATING THE INCOME VARIABLE VALUE FOR EACH MONTH
? ADJUSTING PERSONAL INCOME TO LOGLINEAR FORM
DOT(CHAR=O,VALUE=J) 1001 1002 1003 1004;
ZPCDI .0 = PCDI .0 MTT;
OLSQ(SILENT) ZPCDI_.0 C TT;
FOREST HZPCDI_.O;
LPCDI.O =LOG(HZPCDI_.0);
ENDDOT;


? CREATING TEMPERATURE INDEX FROM REGION J

DOT(CHAR=O) 1001 1002 1003 1004;
OLSQ(SILENT) W.OMEAN C MTHS2-MTHS12;
IW.OMEAN=W.OMEAN/(@FIT);
LIW.OMEAN=LOG(IW.OMEAN);
ENDDOT;


? CREATING PRECIPITATION INDEX FROM REGION J

DOT(CHAR=O) 1001 1002 1003 1004;
W.OPCP= (W.OPCP=0)*0.001 + (W.OPCP>0)*W.OPCP;
OLSQ(SILENT) W.OPCP C MTHS2-MTHS12;
IW.OPCP=W.OPCP/(@FIT);
LIW.OPCP=LOG(IW.OPCP);
ENDDOT;


? CREATING FIRE INDEX VARIABLE

FIRELOW=FIRE1+FIRE2;
FIREMED=FIRE3+FIRE4;
FIREHGH=FIRE5+FIRE6+FIRE7;
SIZELOW=SIZE1+SIZE2;
SIZEMED=SIZE3+SIZE4;
SIZEHGH= SIZE5+SIZE6+SIZE7;
WT=(FIRELOW>0); MSD(NOPRINT, WEIGHT=WT) SIZELOW; SET MSIZELOW=@MEAN;
WT=(FIREMED>0); MSD(NOPRINT, WEIGHT=WT) SIZEMED; SET MSIZEMED=@MEAN;
WT=(FIREHGH>0); MSD(NOPRINT, WEIGHT=WT) SIZEHGH; SET MSIZEHGH=@MEAN;
WT=(FIRELOW>0); MSD(NOPRINT, WEIGHT=WT) FIRELOW;
WT=(FIREMED>0); MSD(NOPRINT, WEIGHT=WT) FIREMED;
WT=(FIREHGH>0); MSD(NOPRINT, WEIGHT=WT) FIREHGH;










FIRE = FIRELOW*(MSIZELOW/MSIZELOW) + FIREMED*(MSIZEMED/MSIZELOW) +
FIREHGH*(MSIZEHGH/MSIZELOW);
OLSQ(SILENT) FIRE C MTHS2-MTHS12;
IFIRE=FIRE/(@FIT);
LIFIRE=LOG(IFIRE);
DIFIRE=IFIRE>1;


? ADJUSTING ADVERTISING TO LOGLINEAR FORM
? NOTE THAT ADV$ ARE IN 1000 OF US$

DD1=(MISS(NBC FL)=0);
NBC FL= (DD1=0)*0 + (DD1=1)*NBC FL;
DD2=(MISS(NBC_AMB)=0);
NBCAMB= (DD2=0)*0 + (DD2=1)*NBC_AMB;
DD3=(MISS(WDC FL)=0);
WDC FL= (DD3=0)*0 + (DD3=1)*WDC FL;
DD4=(MISS(WDC AMB)=0);
WDC AMB= (DD4=0)*0 + (DD4=1)*WDC AMB;
DD5=(MISS(BEG FL)=0);
BEG FL= (DD5=0)*0 + (DD5=1)*BEG FL;
DD6=(MISS(BEG AMB)=0);
BEG AMB= (DD6=0)*0 + (DD6=1)*BEG AMB;
DD7=(MISS(GEN TOT)=0);
GEN TOT= (DD7=0)*0 + (DD7=1)*GEN TOT;
DD8=(MISS(BRAD NFL)=0);
BRAD NFL= (DD8=0)*0 + (DD8=1)*BRAD NFL;
SELECT AYR>1994012;
BRGNAD 5 = (NBC FL + NBC AMB + WDC FL + WDC AMB + BEG FL + BEG AMB + GEN TOT);
LBRGNAD 5 = LOG(0.001 + BRGNAD_5);
BRGNAD 8 = BEG FL + BEG AMB + GEN TOT;
LBRGNAD_8 = LOG(0.001 + BRGNAD_8);
SELECT 1;
BRGNAD_1 =(CRNV TOT + GEN TOT);
LBRGNAD_1 = LOG(0.001 + BRGNAD 1);
BRGNAD 33 = BRGNAD TOT BRAD NFL;
LBRGNAD_33 = LOG(0.001 + BRGNAD_33);
BRGNAD 9 = BRGNAD TOT BRAD NFL;
LBRGNAD 9 = LOG(0.001 + BRGNAD 9);
BRGNAD 14 = BRGNADTOT BRAD NFL;
LBRGNAD_14 = LOG(0.001 + BRGNAD_14);


? CREATING 9.11 DUMMY VARIABLES

AYR=YEAR* 1000+MTHS;
TER911_1= [(AYR)>2001008];
TER911_2= [(AYR)<2001009]*1
+[(AYR >2001008) & (AYR <2002009)]*2
+[(AYR >2002008) & (AYR <2003009)]*3
+[(AYR >2003008) & (AYR <2004009)]*4
+[(AYR >2004008) & (AYR <2005009)]*5
+[(AYR >2005008) & (AYR <2006009)]*6
+[(AYR >2006008)]*7;
DUMMY(PREFIX=TER) TER911_2;
?=======================================-- - -










? CREATING HURRICANE VARIABLES

HCATLOW= HCAT6+HCAT7;
HCATMED= HCAT1+HCAT2; HCATHGH= HCAT3+HCAT4+HCAT5;


? CREATING DUMMY FOR SEASONALITY

YEAR=YEAR1;
MTHS=MTH1;
DUMMY MTHS;










APPENDIX C
ESTIMATION OF MODEL: TSP PROGRAMS

OPTIONS DOUBLE MEMORY=200 LIMWMISS=0 LIMWARN=0;
TITLE 'JOSE CAZANOVA PH.D.-DEMAND FOR DOMESTIC AIR PASSENGER TRAFFIC TO FLORIDA';
? AIRFINALMODEL#4.TSP;
IN'C:\ZTOURISM\ZCAZANOVA\TSPPRG\FINAL\FINALAIRMODEL\AIRFINAL FL';

DOT(CHAR=D,VALUE=J)1 5 8 9 14 33;
DOT(CHAR=O,VALUE=L)1001 1002 1003 1004;
W2LPP.D .O=LPP.D_.O(-2) LPP.D_.O(-12);
W4LPP.D .O=LPP.D_.O(-4) LPP.D_.O(-12);

? SUR MODEL MONTHS AND CORRECTING FOR THE AR1 APPROACH;

DOT(CHAR=D,VALUE=J) 1 5 8 9 14;
DOT(CHAR=O,VALUE=K) 1001 1002 1003 1004;

FRML EQ1.D_.O LPP.D_.O = (AO_.D_.O*C + A1_.D_.O*LAFR.D_.0_2 + A2_.D_.O*LPCDI_.0 +
A3 .D .O*LIW.OMEAN + A4 .D .O*LIW.OPCP + A5 .D .O*LIFIRE + A6 .D .O*HCATLOW +
A7 .D .O*HCATMED + A8 .D .O*HCATHGH + A9 .D .O*LBRGNAD .D +
A10 .D .O*TER2 +All .D .O*TER3 +A12 .D .O*TER4 +A13 .D .O*TER5 +A14 .D .O*TER6 +
A15 .D .O*TER7 +A16 .D .O*MTHS2 + A17 .D .O*MTHS3 + A18 .D .O*MTHS4 +
A19 .D .O*MTHS5 + A20 .D .O*MTHS6 + A21 .D .O*MTHS7 + A22 .D .O*MTHS8 +
A23 .D .O*MTHS9 + A24 .D .O*MTHS10+A25 .D .O*MTHS11 +A26 .D .O*MTHS12 +
A27_.D_.O*W2LPP.D_.0 + A28_.D_.O*W4LPP.D_.0 + A29_.D_.O*LPP.D_.O(-12))+
RHO.D_.O*(LPP.D_.O(-1) ((AO_.D_.O*C(-1) + A1_.D_.O*LAFR.D_.0_2(-1) + A2_.D_.O*LPCDI_.O(-1)
+ A3_.D_.O*LIW.OMEAN(-1) + A4_.D_.O*LIW.OPCP(-1) + A5_.D_.O*LIFIRE(-1) +
A6_.D_.O*HCATLOW(-1) + A7_.D_.O*HCATMED(-1) + A8_.D_.O*HCATHGH(-1) +
A9 .D .O*LBRGNAD_.D(-1) + A10_.D_.O*TER2(-1) + A11.D_.O*TER3(-1) + A12_.D_.O*TER4(-
1) + A13 .D .O*TER5(-1)+ A14_.D_.O*TER6(-1)+ A15_.D_.O*TER7(-1) +
A16 .D_.O*MTHS2(-1) + A17 .D_.O*MTHS3(-1) + A18_.D_.O*MTHS4(-1) + A19_.D_.O*MTHS5(-
1)+ A20 .D_.O*MTHS6(-1) + A21 .D .O*MTHS7(-1) + A22_.D_.O*MTHS8(-1) +
A23 .D .O*MTHS9(-1) + A24_.D_.O*MTHS10(-1) + A25_.D_.O*MTHS11(-1) + A26 .D .O*MTHS12(-1)
+ A27_.D_.O*W2LPP.D_.O(-1) + A28_.D_.O*W4LPP.D_.O(-1) + A29_.D_.O*LPP.D_.O(-13))));

PARAM AO .D .0, A1 .D .0, A2 .D .0, A3_.D_.0, A4 .D .0, A5_.D_.0, A6 .D .0, A7 .D .0,
A8_.D_.0, A9 .DO,.., A10 .D .O, A .D., A12 .D .0, A13_.D_.0, A14_.D_.0, A15_.D_.0, A16 .D .0,
A17 .D .O, A18 .D .O, A19 .D .O, A20 .D .O, A21 .D .O, A22_.D_.O, A23_.D_.O, A24_.D_.O, A25_.D_.0,
A26 .D .O, A27 .D .O, A28 .D .O, A29 .D .O, RHO.D.O;

ENDDOT; ENDDOT;
TREND OBS;
SELECT OBS>1;
LSQ(MAXIT=100) EQ11_1001 EQ11_1002 EQ11_1003 EQ11_1004
EQ15_1001 EQ15_1002 EQ15_1003 EQ15_1004
EQ18_1001 EQ18_1002 EQ18_1003 EQ18_1004
EQ19_1001 EQ19_1002 EQ19_1003 EQ19_1004
EQ114_1001 EQ114_1002 EQ114_1003 EQ114_1004;

MAT MCOEFALL=@COEF;
OUT 'C:\ZTOURISM\ZCAZANOVA\TSPPRG\FINAL\FINALAIRMODEL\AIRFINAL FL';
MAT MCOEFALL=MCOEFALL;
OUT;
END;










APPENDIX D
SIMULATION ANALYSIS: TSP PROGRAMS

OPTIONS DOUBLE MEMORY=200 LIMWMISS=0 LIMWARN=0;
TITLE 'JOSE CAZANOVA PH.D.-DEMAND FOR DOMESTIC AIR PASSENGER TRAFFIC TO FLORIDA';
? AIRFINALMODEL#4.TSP;
IN'C:\ZTOURISM\ZCAZANOVA\TSPPRG\FINAL\FINALAIRMODEL\AIRFINAL FL';

? <<<<<<<<<<<<<<<<<<<<<<<<< SIMULATION SECTION >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
? LET THE COEFFICIENT MATRIX BE A K (NUMBER OF VARIABLES) x M (NUMBER OF D TIME O);
? LET XX.D .0 BE THE ORIGINAL DATA IN MARTIX FORM FOR DESTINATION D AND ORIGIN O;
? BB.D .0 IS THE MATRIX OF COEFFICIENT FOR D AND O;
? WE WILL SIMULATE ALL D AND O AT THE SAME TIME;
DOT(CHAR=D) 14; ? 1 5 8 9 14 33;
DOT(CHAR=O) 1001 1002 1003 1004;
LPP=LPP.D_.0; PCPFIT=W.OPFIT; TEMPFIT=W.OTFIT; HZPOP=(HZPOP_.0/100000)*1000000;
XFARE =LAFR.D_.0_2; XINC =LPCDI_.O; XWEA1 =LIW.OMEAN; XWEA2 =LIW.OPCP; XFIRE
=LIFIRE;
XCATL =HCATLOW ; XCATM =HCATMED ; XCATH =HCATHGH; XBRND =LBRGNAD_.D;
XTER2 =TER2; XTER3 =TER3; XTER4 =TER4; XTER5 =TER5; XTER6 =TER6; XTER7 =TER7;
XMTHS2 = MTHS2; XMTHS3 = MTHS3; XMTHS4 = MTHS4; XMTHS5 = MTHS5; XMTHS6 = MTHS6;
XMTHS7 = MTHS7; XMTHS8 = MTHS8; XMTHS9 = MTHS9; XMTHS10= MTHS10; XMTHS11= MTHS11;
XMTHS12= MTHS12;

LIST zZVARZz
XFARE XINC XWEA1 XWEA2 XFIRE XCATL XCATM XCATH XBRND
XTER2 XTER3 XTER4 XTER5 XTER6 XTER7
XMTHS2 XMTHS3 XMTHS4 XMTHS5 XMTHS6 XMTHS7 XMTHS8 XMTHS9 XMTHS10 XMTHS11
XMTHS12;

LIST ZVARZ
QFARE QINC QWEA1 QWEA2 QFIRE XCATL XCATM XCATH QBRND
XTER2 XTER3 XTER4 XTER5 XTER6 XTER7
XMTHS2 XMTHS3 XMTHS4 XMTHS5 XMTHS6 XMTHS7 XMTHS8 XMTHS9 XMTHS10 XMTHS11
XMTHS12;

LIST WVARW
WFARE WINC WWEA1 WWEA2 WFIRE WCATL WCATM WCATH WBRND
WTER2 WTER3 WTER4 WTER5 WTER6 WTER7
WMTHS2 WMTHS3 WMTHS4 WMTHS5 WMTHS6 WMTHS7 WMTHS8 WMTHS9 WMTHS10 WMTHS11
WMTHS12;

LIST SVARS
SFARE SINCE SWEA1 SWEA2 SFIRE SCATL SCATM SCATH SBRND
STER2 STER3 STER4 STER5 STER6 STER7
SMTHS2 SMTHS3 SMTHS4 SMTHS5 SMTHS6 SMTHS7 SMTHS8 SMTHS9 SMTHS10 SMTHS11 SMTHS12;
ENDDOT; ENDDOT;

DOT WVARW SVARS; SET .=0; SET .=0; SET IHH=1; ENDDOT;
SET I=0;

PROC INIT;

DOT WVARW SVARS; SET .=0; SET .=0; SET IHH=1; ENDDOT;
ENDPROC INIT;











MFORM(TYPE=GEN,NROW=70,NCOL=1) MSIMO=0;
MFORM(TYPE=GEN,NROW=70,NCOL=55) MSIM=0;
SET SIMNUM=0;
SET SIMVAR=0;


? ASSUME YOUR COEFFICIENTS ARE BB(I,J);
? CREATING THE MATRIX OF COEFFICIENTS;
SET K=0;
DOT(CHAR=D) 1 5 8 9 14; ? 33; ? <=== MUST SET THE DESTINATION REGION BECAUSE OF THE
SPREADSHEET SIZE;
DOT(CHAR=O) 1001 1002 1003 1004;
MFORM(TYPE=GEN,NROW=31,NCOL=1) BB.D_.O=0;
MFORM(TYPE=GEN,NROW=27,NCOL=1) ABB.D_.O=0;
SET K=K+1;
DO J=l TO 31;
SETKK= J + 31*(K-1);
SET BB.D_.O(J) = MCOEFALL(KK);
ENDDO;
DO J=l TO 27;
MAT ABB.D_.O(J)=BB.D_.O(J);
ENDDO;
SET L1=28;
SET L2=29;
SET L3=30;
ENDDOT; ENDDOT;
?PRINT BB1 1001 BB5 1002 BB9 1003 BB14 1004;
?PRINT ABB1 1002 ABB8 1004;

PROC ZSIMZ;

SELECT 1;
SET I=I+1;
?SET II=0;
IZZ=1; ?INTERCEPT;
SET IWW=0;
?SET I=I+1;
MAKE SX1 IZZ ZVARZ ; ? THIS IS THE ORIGINAL DATA BUT FARE, INC, AND BRAND ARE NOT
LOGS YET;
MAKE SX4 IZZ ZZVARZZ; ? THIS IS THE ORIGINAL DATA IN LOG FORM;
MAKE SX2 IWW SVARS;
MAKE SX3 IWW WVARW;
MAT X2= SX1%(IZZ#SX2');
MAT X3= SX1%(IZZ#SX3');
MAT Q1= SX1 -X2 + X3;
UNMAKE Q1
IQ QQFARE QQINC QQWEA1 QQWEA2 QQFIRE QXCATL QXCATM QXCATH QQBRND
QXTER2 QXTER3 QXTER4 QXTER5 QXTER6 QXTER7
QXMTHS2 QXMTHS3 QXMTHS4 QXMTHS5 QXMTHS6 QXMTHS7 QXMTHS8 QXMTHS9 QXMTHS10
QXMTHS11 QXMTHS12;

QQ1FARE=LOG(QQFARE);
QQ1INC=LOG(QQINC);
QQ1BRND=LOG(QQBRND);
QQ1WEA1=LOG(QQWEA1);
QQ1WEA2=LOG(QQWEA2);










QQ1FIRE=LOG(QQFIRE);


MAKE X1
IQ QQ1FARE QQ1INC QQ1WEA1 QQ1WEA2 QQ1FIRE QXCATL QXCATM QXCATH QQ1BRND
QXTER2 QXTER3 QXTER4 QXTER5 QXTER6 QXTER7
QXMTHS2 QXMTHS3 QXMTHS4 QXMTHS5 QXMTHS6 QXMTHS7 QXMTHS8 QXMTHS9 QXMTHS10
QXMTHS11 QXMTHS12;

MAT NRX1=NROW(X1);
MAT XB=X1*AA; ? THIS IS THE XB WITHOUT THE LAGGED IMPACTS BUT THE CONTROLLED
VARIABLES;

MAT SXB=SX4*AA;
UNMAKE SXB PPP3;
UNMAKE XB PPP1; ? PPP1 BEING THE STRUCTURAL EFFECTS WITHOUT THE HABIT PERSISTENCE;

? WE NEED TO CONTROL FOR THE MISSING VALUES THAT GIVE EXTREME RESULTS;
MISSPPP1=(PPP1<100);
SELECT MISSPPP1=0; MISSPPP1=@ MISS;
SELECT 1;

PPP1=PPP1*MISSPPP1; ? REMOVING THE DATA POINTS THAT WERE MISSING;
PPP3=PPP3*MISSPPP1;

LPPSIM = LPP;
LPPSIM3= LPP;
SELECT YEAR >= 1996;

LPPSIM = PPP1 + LAM1*[LPPSIM(-2) LPPSIM(-12)] + LAM2*[LPPSIM(-4) LPPSIM(-12)]+
LAM3*[LPPSIM(-12)];
LPPSIM3 = PPP3 + LAM1*[LPPSIM3(-2) LPPSIM3(-12)] + LAM2*[LPPSIM3(-4) LPPSIM3(-12)]+
LAM3*[LPPSIM3(-12)];
LPPSIM2 = PPP1 + LAM1*[LPPSIM3(-2) LPPSIM3(-12)] + LAM2*[LPPSIM3(-4) LPPSIM3(-12)]+
LAM3*[LPPSIM3(-12)];
LPPSIM1=LPPSIM;

WWLPP=(LPP>0);
LPPSIM2 = LPPSIM2*WWLPP; ? 2 IS WITH THE ACTUAL LPP;
LPPSIM1 = LPPSIM1*WWLPP; ? 1 IS WITH THE SIMUATATED CHANGES IN LPP;
PRINT LPPSIM1 LPPSIM2 LPPSIM3;

PPASS1= EXP(LPPSIM1);
PPASS2= EXP(LPPSIM2);
MSD(NOPRINT) PPASS1 PPASS2; SET MPPASS1_0=(@MEAN(1)); SET MPPASS2_0=( MEAN(2));
FARE =QFARE*CPIAFARE; INC=QINC; BRD=QBRND;
?FARE =EXP(XFARE)*CPI_AFARE; INC=EXP(XINC); BRD=EXP(XBRND);
TEMP=EXP(XWEA1)*TEMPFIT; RAIN=EXP(XWEA2)*PCPFIT; INFIRE=EXP(XFIRE);
AFIRE=EXP(XFIRE)*FIREFIT;
HHZPOP=HZPOP;
?PRINT FARE INC BRD;
MSD(NOPRINT) FARE INC BRD TEMP RAIN INFIRE AFIRE HHZPOP;
SET MFARE=@MEAN(1); SET MINC=@MEAN(2); SET MBRD=@MEAN(3);
SET MTEMP=@MEAN(4); SET MRAIN=@MEAN(5); SET MINFIRE=@MEAN(6); SET
MAFIRE=@MEAN(7); SET MHHZPOP= MEAN(8);
?PRINT MFARE MINC MBRD;











PRINT PPASS1 PPASS2;
DOT(CHAR=M) 1-12;
WWW=(MTHS.M=1); MSD(NOPRINT,WEIGHT=
WWW=(MTHS.M=1); MSD(NOPRINT,WEIGHT=
WWW=(MTHS.M=1); MSD(NOPRINT,WEIGHT=
WWW=(MTHS.M=1); MSD(NOPRINT,WEIGHT=
WWW=(MTHS.M=1); MSD(NOPRINT,WEIGHT=
WWW=(MTHS.M=1); MSD(NOPRINT,WEIGHT=
WWW=(MTHS.M=1); MSD(NOPRINT,WEIGHT=
WWW=(MTHS.M=1); MSD(NOPRINT,WEIGHT=
WWW=(MTHS.M=1); MSD(NOPRINT,WEIGHT=
WWW=(MTHS.M=1); MSD(NOPRINT,WEIGHT=
ENDDOT;
PRINT MFARE1;


DOT(CHAR=M,VALUE=K) 1996-2007;


=(YEAR=
=(YEAR=
=(YEAR=
=(YEAR=
=(YEAR=
=(YEAR=
=(YEAR=
=(YEAR=
=(YEAR=
=(YEAR=


K); MSD(NOPRINT,WEIGHT=
=K); MSD(NOPRINT,WEIGHT=
=K); MSD(NOPRINT,WEIGHT=
=K); MSD(NOPRINT,WEIGHT=
=K); MSD(NOPRINT,WEIGHT=
=K); MSD(NOPRINT,WEIGHT=
=K); MSD(NOPRINT,WEIGHT=
=K); MSD(NOPRINT,WEIGHT=
=K); MSD(NOPRINT,WEIGHT=
:K); MSD(NOPRINT,WEIGHT=


=WWW) PPASS1; SET MPPASS1_.M=@MEAN;
=WWW) PPASS2; SET MPPASS2_.M=@MEAN;
WWW) FARE; SET MFARE.M=@MEAN;
=WWW) INC; SET MINC.M=@MEAN;
=WWW) BRD; SET MBRD.M=@MEAN;
=WWW) TEMP; SET MTEMP.M=@MEAN;
=WWW) RAIN; SET MRAIN.M=@MEAN;
=WWW) INFIRE; SET MINFIRE.M=@MEAN;
=WWW) AFIRE; SET MAFIRE.M=@MEAN;
=WWW) HHZPOP; SET MHHZPOP.M=@MEAN;


=WWW) PPASS1; SET MPPASS1_.M=@MEAN;
=WWW) PPASS2; SET MPPASS2_.M=@MEAN;
=WWW) FARE; SET MFARE.M=@MEAN;
=WWW) INC; SET MINC.M=@MEAN;
=WWW) BRD; SET MBRD.M=@MEAN;
=WWW) TEMP; SET MTEMP.M=@MEAN;
=WWW) RAIN; SET MRAIN.M=@MEAN;
=WWW) INFIRE; SET MINFIRE.M=@MEAN;
=WWW) AFIRE; SET MAFIRE.M=@MEAN;
=WWW) HHZPOP; SET MHHZPOP.M=@MEAN;


ENDDOT;


SET MSIM(I,J)=I;
SET MSIM(I,J)=SIMNUM;
SET MSIM(I,J)=SIMVAR;
SET MSIM(I,J)=0;


5; SET MSIM(I,J)=
7; SET MSIM(I,J)=
9; SET MSIM(I,J)=
11; SET MSIM(I,J)
13; SET MSIM(I,J):
15; SET MSIM(I,J):
17; SET MSIM(I,J)
19; SET MSIM(I,J)
21; SET MSIM(I,J)
23; SET MSIM(I,J)
25; SET MSIM(I,J)
27; SET MSIM(I,J)
29; SET MSIM(I,J)

31; SET MSIM(I,J)
33; SET MSIM(I,J)
35; SET MSIM(I,J)
37; SET MSIM(I,J)
39; SET MSIM(I,J)
41; SET MSIM(I,J)
43; SET MSIM(I,J)
45; SET MSIM(I,J)


MPPASS1I
MPPASS1I
MPPASS1I
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1

=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1
=MPPASS1


0; SET J=
1; SET J=
2; SET J=
3; SET J
4; SET J
5; SET J
6; SET J
7; SET J
8; SET J
9; SET J
10; SET J
11; SET J
12; SET J


1996; SET J=
1997; SET J=
1998; SET J=
1999; SET J=
2000; SET J=
2001; SET J
2002; SET J=
2003; SET J=


6; SET MSIM(I,J)=(]
8 ; SET MSIM(I,J)=(]
10; SET MSIM(I,J)=(
=12; SET MSIM(I,J)=
=14; SET MSIM(I,J)=
=16; SET MSIM(I,J)=
=18; SET MSIM(I,J)=
=20; SET MSIM(I,J)=
=22; SET MSIM(I,J)=
=24; SET MSIM(I,J)=
=26; SET MSIM(I,J)=
=28; SET MSIM(I,J)
=30; SET MSIM(I,J)


:32; SET MSIM(I,J):
=34; SET MSIM(I,J)=
:36; SET MSIM(I,J)=
38; SET MSIM(I,J)=
40; SET MSIM(I,J)=
42; SET MSIM(I,J)=
44; SET MSIM(I,J)=
=46; SET MSIM(I,J)=


MPPASS2_0
MPPASS2_1
MPPASS2_2
(MPPASS2_
(MPPASS2_
(MPPASS2_
(MPPASS2_
(MPPASS2_
(MPPASS2_
(MPPASS2_
=(MPPASS2
=(MPPASS2
=(MPPASS2


);
);
5);
3);
4);
5);
6);
7);
8);
9);
10);
11);
12);


(MPPASS2
:(MPPASS2_
=(MPPASS2
=(MPPASS2
=(MPPASS2
=(MPPASS2
=(MPPASS2
=(MPPASS2


1996);
1997);
1998);
1999);
2000);
2001);
2002);
2003);


WWW=
www=
WWW=
www=
WWW=
www=
WWW=
www=
WWW
www=
WWW=
WWW=


SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=

SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=
SET J=










SET J=47; SET MSIM(I,J)=MPPASS1_2004; SET J=48; SET MSIM(I,J)=(MPPASS2_2004);
SET J=49; SET MSIM(I,J)=MPPASS1_2005; SET J=50; SET MSIM(I,J)=(MPPASS2_2005);
SET J=51; SET MSIM(I,J)=MPPASS 1_2006; SET J=52; SET MSIM(I,J)=(MPPASS2_2006);
SET J=53; SET MSIM(I,J)=MPPASS1_2007; SET J=54; SET MSIM(I,J)=(MPPASS2_2007);

SET J=55 ;SET MSIM(I,J)=MFARE;
SET J=56 ;SET MSIM(I,J)=WFARE;
SET J=57 ;SET MSIM(I,J)=MINC;
SET J=58 ;SET MSIM(I,J)=WINC;
SET J=59 ;SET MSIM(I,J)=MBRD;
SET J=60 ;SET MSIM(I,J)=WBRND;
SET J=61 ;SET MSIM(I,J)=MTEMP;
SET J=62 ;SET MSIM(I,J)=WWEA1;
SET J=63 ;SET MSIM(I,J)=MRAIN;
SET J=64 ;SET MSIM(I,J)=WWEA2;
SET J=65 ;SET MSIM(I,J)=MINFIRE;
SET J=66 ;SET MSIM(I,J)=MAFIRE;
SET J=67 ;SET MSIM(I,J)=WFIRE;
SET J=68 ;SET MSIM(I,J)=MHHZPOP;

DOT(VALUE=FF) 1-12;
SET EF=FF+68; SET MSIM(I,EF)=MFARE.; ENDDOT;
DOT(VALUE=FF) 1-12;
SET EF=FF+68+12; SET MSIM(I,EF)=MINC.; ENDDOT;
DOT(VALUE=FF) 1-12;
SET EF=FF+68+24; SET MSIM(I,EF)=MBRD.; ENDDOT;
DOT(VALUE=FF) 1-12;
SET EF=FF+68+36; SET MSIM(I,EF)=MTEMP.; ENDDOT;
DOT(VALUE=FF) 1-12;
SET EF=FF+68+48; SET MSIM(I,EF)=MRAIN.; ENDDOT;
DOT(VALUE=FF) 1-12;
SET EF=FF+68+48+12; SET MSIM(I,EF)=MINFIRE.; ENDDOT;
DOT(VALUE=FF) 1-12;
SET EF=FF+68+48+24; SET MSIM(I,EF)=MAFIRE.; ENDDOT;
DOT(VALUE=FF) 1-12;
SET EF=FF+68+48+24+12; SET MSIM(I,EF)=MHHZPOP.; ENDDOT;

DOT(INDEX=FF) 1996-2007;
SET EF=FF+68+96; SET MSIM(I,EF)=MFARE.; ENDDOT;
DOT(INDEX=FF) 1996-2007;
SET EF=FF+68+96+12; SET MSIM(I,EF)=MINC.; ENDDOT;
DOT(INDEX=FF) 1996-2007;
SET EF=FF+68+96+12+12; SET MSIM(I,EF)=MBRD.; ENDDOT;
DOT(INDEX=FF) 1996-2007;
SET EF=FF+68+96+12+12+12; SET MSIM(I,EF)=MTEMP.; ENDDOT;
DOT(INDEX=FF) 1996-2007;
SET EF=FF+68+96+12+12+12+12; SET MSIM(I,EF)=MRAIN.; ENDDOT;
DOT(INDEX=FF) 1996-2007;
SET EF=FF+68+96+12+12+12+12+12; SET MSIM(I,EF)=MINFIRE.; ENDDOT;
DOT(INDEX=FF) 1996-2007;
SET EF=FF+68+96+12+12+12+12+12+12; SET MSIM(I,EF)=MAFIRE.; ENDDOT;
DOT(INDEX=FF) 1996-2007;
SET EF=FF+68+96+12+12+12+12+12+12+12; SET MSIM(I,EF)=MHHZPOP.; ENDDOT;
ENDPROC;











DOT(CHAR=D) 14; ? 1 5 8 9 14 33; ? <=== MUST SET THE DESTINATION REGION BECAUSE OF THE
SPREADSHEET SIZE;
DOT(CHAR=O) 1001 1002 1003 1004;
MAT AA =ABB.D .0;
SET LAM1=BB.D_.O(L1);
SET LAM2=BB.D_.O(L2);
SET LAM3=BB.D_.O(L3);
LPP=LPP.D .0;
XFARE =LAFR.D .O02; XINC =LPCDI.O ; XWEA1 =LIW.OMEAN; XWEA2 =LIW.OPCP;
PCPFIT=W.OPFIT; TEMPFIT=W.OTFIT; HZPOP=(HZPOP_.0/100000)*1000000;
SET I=0;
QFARE =EXP(XFARE);
QINC =EXP(XINC);
QBRND =EXP(XBRND);


QWEA1= EXP(XWEA1);
QWEA2= EXP(XWEA2);
QFIRE= EXP(XFIRE);

?PRINT LAM1 LAM2 LAM3;


? STARTING THE ACTUAL SIMULATIONS;
-======================================= -




? SIMULATION #1 ACTUAL VALUES OF ALL VARIABLES;
-======================================= -
SET SIMNUM=1;
SET SIMVAR=1;INIT;ZSIMZ;

-======================================= -
? SIMULATION #2 STORMS;
-======================================= -
SET SIMNUM=2;
?ACTUAL STORMS;
SET SIMVAR=1;INIT; ZSIMZ;
? NO STORMS;
SET SIMVAR=2;INIT;
SET SCATL=1; SET SCATM=1; SET SCATH=1;
SET WCATL=0; SET WCATM=0; SET WCATH=0; ZSIMZ;
? NO CATEGORY FIVE STORMS;
SET SIMVAR=3; INIT;
SET SCATL=0; SET SCATM=0; SET SCATH=1;
SET WCATL=0; SET WCATM=0; SET WCATH=0; ZSIMZ;
? NO MEDIUM OR HIGH CATEGORY STORMS;
SET SIMVAR=4; INIT;
SET SCATL=0; SET SCATM=1; SET SCATH=1;
SET WCATL=0; SET WCATM=0; SET WCATH=0; ZSIMZ;

-======================================= -
? SIMULATION #3 FARES;

SET SIMNUM=3; ? ACTUAL FARES;
SET SIMVAR=1; INIT; ZSIMZ;










DO HH= .80 TO 1.21 BY .05;
SET SIMVAR=SIMVAR+1; INIT;
SET SFARE=1; SET WFARE=HH; ZSIMZ;
ENDDO;


? SIMULATION #4 INCOME;

SET SIMNUM=4;
? ACTUAL INCOME;
SET SIMVAR=1; INIT; ZSIMZ;
DO HH= .95 TO 1.05 BY .01;
SET SIMVAR=SIMVAR+1; INIT;
SET SINC=1; SET WINC=HH; ZSIMZ;
ENDDO;


? SIMULATION #5- ADVERTISING;

SET SIMNUM=5;
? ACTUAL ADVERTISING;
SET SIMVAR=1; INIT; ZSIMZ;
DO HH= .80 TO 1.21 BY .05;
SET SIMVAR=SIMVAR+1; INIT;
SET SBRND=1; SET WBRND=HH; ZSIMZ;
ENDDO;


? SIMULATION #6 TERROR;

SET SIMNUM=6;
?ACTUAL 9-11;
SET SIMVAR=1;INIT; ZSIMZ;
? NO 9-11;
SET SIMVAR=2;INIT;
SET STER2=1; SET STER3=1; SET STER4=1; SET STER5=1; SET STER6=1; SET STER7=1;
SET WTER2=0; SET WTER3=0; SET WTER4=0; SET WTER5=0; SET WTER6=0; SET WTER7=0; ZSIMZ;


? SIMULATION #7- TEMPERATURES;

SET SIMNUM=7;
? ACTUAL TEMPERATURES;
SET SIMVAR=1; INIT; ZSIMZ;
DO HH= .80 TO 1.21 BY .05;
SET SIMVAR=SIMVAR+1; INIT;
SET SWEA1=1; SET WWEA1=HH; ZSIMZ;
ENDDO;


? SIMULATION #8- RAINFALL;

SET SIMNUM=8;
? ACTUAL RAINFALL;
SET SIMVAR=1; INIT; ZSIMZ;
DO HH= .80 TO 1.21 BY .05;










SET SIMVAR=SIMVAR+1; INIT;
SET SWEA2=1; SET WWEA2=HH; ZSIMZ;
ENDDO;


? SIMULATION #9 FIRES;

SET SIMNUM=9;
?ACTUAL FIRES;
SET SIMVAR=1;INIT; ZSIMZ;
? NO FIRES;
SET SIMVAR=2;INIT;
SET SFIRE=1;
SET WFIRE=0; ZSIMZ;
DO HH= .80 TO 1.21 BY .05;
SET SIMVAR=SIMVAR+1; INIT;
SET SFIRE=1; SET WFIRE=HH; ZSIMZ;
ENDDO;


MAKE MSIMO MSIMO MSIM;
ENDDOT;
ENDDOT;
MAT MSIMT = MSIMO';
?PRINT MSIMO;
WRITE(FORMAT=EXCEL,FILE='C:\ZTOURISM\ZCAZANOVA\TSPPRG\FINAL\FINALAIRMODEL\SIMRE
SULTSCSA1414.XLS') MSIMT;

END;










APPENDIX E
POPULATION FACTORS BY U.S. REGION

Table E-1. Northeast region (1001): population factors by month and by year between 1996 and
2006.
Year-

Monthl 3 3 3 ^ r^ ^ ^ ^ AVG by Month
January 525 528 530 533 535 538 540 543 546 548 551 538
February 525 528 531 533 536 538 541 543 546 548 551 538
March 526 528 531 533 536 538 541 543 546 549 551 538
April 526 528 531 533 536 539 541 544 546 549 551 539
May 526 529 531 534 536 539 541 544 546 549 552 539
June 526 529 531 534 536 539 542 544 547 549 552 539
July 526 529 532 534 537 539 542 544 547 549 552 539
August 527 529 532 534 537 539 542 545 547 550 552 539
September 527 529 532 535 537 540 542 545 547 550 552 540
October 527 530 532 535 537 540 542 545 548 550 553 540
November 527 530 532 535 538 540 543 545 548 550 553 540
December 528 530 533 535 538 540 543 545 548 550 553 540
AVG byYear 526 529 531 534 537 539 542 544 547 549 552 540*
*Overall average for the entire 11-year period.

Table E-2. Midwest region (1002): population factors by month and by year between 1996 and
2006.
Year-

Monthu 5 5 5 5 5 52 5 5 5 AVG by Month
January 508 512 516 519 523 526 530 534 537 541 544 526
February 509 512 516 519 523 527 530 534 538 541 545 527
March 509 513 516 520 523 527 531 534 538 541 545 527
April 509 513 516 520 524 527 531 535 538 542 545 527
May 509 513 517 520 524 528 531 535 538 542 546 528
June 510 513 517 521 524 528 532 535 539 542 546 528
July 510 514 517 521 525 528 532 535 539 543 546 528
August 510 514 518 521 525 528 532 536 539 543 547 528
September 511 514 518 522 525 529 532 536 540 543 547 529
October 511 515 518 522 525 529 533 536 540 544 547 529
November 511 515 519 522 526 529 533 537 540 544 547 529
December 512 515 519 522 526 530 533 537 541 544 548 530
AVGby Year 510 514 517 521 524 528 532 535 539 543 546 528*
*Overall average for the entire 11-year period.










Table E-3. South region (1003): population factors by month and by year between 1996 and
2006.
Year-

Monthj 3 3 3 r AVG by Month
January 797 809 820 832 843 854 866 877 888 900 911 854
February 798 810 821 833 844 855 867 878 889 901 912 855
March 799 811 822 833 845 856 868 879 890 902 913 856
April 800 812 823 834 846 857 868 880 891 903 914 857
May 801 813 824 835 847 858 869 881 892 904 915 858
June 802 814 825 836 848 859 870 882 893 904 916 859
July 803 815 826 837 849 860 871 883 894 905 917 860
August 804 815 827 838 850 861 872 884 895 906 918 861
September 805 816 828 839 851 862 873 885 896 907 919 862
October 806 817 829 840 851 863 874 886 897 908 920 863
November 807 818 830 841 852 864 875 886 898 909 921 864
December 808 819 831 842 853 865 876 887 899 910 922 865
AVGby Year 803 814 825 837 848 859 871 882 894 905 916 859*
*Overall average for the entire 11-year period

Table E-4. West region (1004): population factors by month and by year between 1996 and
2006.
Year-

Month}j 3 3 3 r AVG by Month
January 594 604 614 624 634 644 654 664 674 684 695 644
February 594 604 615 625 635 645 655 665 675 685 695 645
March 595 605 615 625 636 646 656 666 676 686 696 646
April 596 606 616 626 636 647 657 667 677 687 697 647
May 597 607 617 627 637 647 657 668 678 688 698 647
June 598 608 618 628 638 648 658 668 679 689 699 648
July 599 609 619 629 639 649 659 669 679 689 700 649
August 599 609 620 630 640 650 660 670 680 690 700 650
September 600 610 620 631 641 651 661 671 681 691 701 651
October 601 611 621 631 641 652 662 672 682 692 702 652
November 602 612 622 632 642 652 663 673 683 693 703 652
December 603 613 623 633 643 653 663 673 684 694 704 653
AVG by Year 598 608 618 628 639 649 659 669 679 689 699 649*
*Overall average for the entire 11-year period.









LIST OF REFERENCES


Adamowicz, W.L. 1994. "Habit Formation and Variety Seeking in a Discrete Choice Model of
Recreation Demand." Journal ofAgricultural and Resource Economics 19(1): 19-31.

Archer, B.H. 1994. "Demand Forecasting and Estimation." In J.R.B. Ritchie and C. Goeldner,
ed. Travel Tourism and Hospitality Research: A Handbook for Managers and
Researchers. New York: JohnWiley and Sons Ltd. Pp. 105-114.

Bhadra D. and J. Kee. 2008. "Structure and Dynamics of the Core U.S. Air Travel Markets: A
Basic Empirical Analysis of Domestic Passenger Demand." Journal ofAir Transport
Management 14:27-39.

Blunk, S.S., D.E. Clark, and J.M. McGibany. 2006. "Evaluating the Long Run Impacts of the
9/11 Terrorist Attacks on the U.S. Domestic Airline Travel." Applied Economics 38:363-
370.

Brons, M., E. Pels, P. Nijkamp, and P. Rietveld. 2002. "Price Elasticities of Demand for
Passenger Air Travel: A Meta-Analysis." Journal ofAir Transport Management 8:165-
175.

Brueckner, J.K. 2004. "Network Structure and Airline Scheduling." The Journal of Industrial
Economics L2(2):291-312.

Chamera, W.W. and D.F. Deadman 1992. New Directions in Econometric Practice: General to
Specific Modelling, Cointegration and Vector Autoregression. Aldershot: Edward Elgar
Publishing Limited.

Dargay, J. and M. Hanly 2001. "The Determinants of the Demand for International Air Travel to
and from the UK." ESRC Transport Studies Unit.

De Mello M. M. and N. Fortuna. 2005. "Testing Alternative Dynamic Systems for Modelling
Tourism Demand." Research Center on Industrial, Labour and Managerial Economics
DP2005-01, University of Porto.

Eugenio-Martin, J.L., N. Martin Morales, and R. Scarpa. 2004. "Tourism and Economic Growth
in Latin American Countries: A Panel Data Approach." Social Science Research Network
Electronic Paper Collection, Nota Di Lavoro 26.

Flannery, M.J. and K.P. Rangan. 2006. "Partial Adjustment toward Target Capital Structures."
Journal ofFinancial Economics 79:469-506.

Fritz, R.G., C. Brandon, and J. Xander. 1984. "Combining Time-Series and Econometric
Forecast of Tourism Activity." Annals of Tourism Research 11:219-229.

Fyall, A. and B. Garrod. 2005. Aspects of Tourism: Tourism Marketing: A Collaborative
Approach. New York: Channel View Publications.









Grais, R.F., J.H. Ellis, and G.E. Glass. 2003. "Assessing the Impact of Airline Travel on the
Geographical Spread of Pandemic Influenza." European Journal ofEpidemiology
18(11):1065-1072.

Greene, W.H. 2003. Econometric Analysis. New Jersey: Prentice Hall.

Harvey, A.C. and P.H.J. Todd. 1983. "Forecasting Economic Time Series with Structural and
Box-Jenkins Models: A Case Study." Journal of Business and Economic Statistics
1(4):299-307.

Hodges, A. and D. Mulkey. 2001. "Terrorist Attacks in New York City and Washington, D.C.:
Implications for the Florida Tourism Industry." University of Florida Cooperative
Extension Service, IFAS EDIS No.316, October.

Ito, H. and D. Lee. 2005. "Assessing the Impact of the September 11 terrorist Attacks on U.S.
Airline Demand." Journal of Economics and Business 57:75-95.

Jones, C., M. Munday, and A. Roberts. 2003. "Regional Tourism Satellite Accounts: A Useful
Policy Tool?" Urban Studies 40(13):2777-2794.

Kulendaran, N. and K.K.F. Wong. 2005. "Modeling Seasonality in Tourism Forecasting."
Journal of Travel Research 44:163-170.

Li, G. and H. Song. 2006. "New Forecasting Models." Journal of Travel and Tourism Marketing
21(4):3-13.

Li, G., H. Song and S. Witt. 2005. "Recent Developments in Econometrics Modeling and
Forecasting." Journal of Travel Research 44:82-99.

Lu, C.H.Y. and P. Morrell. 2001. "Evaluation and Implications of Environmental Charges on
Commercial Flights." Transport Reviews 21(3):377-395.

Lumsdon, L. and S.J. Page. 2004. Tourism and Transport: Issues andAgenda for the New
Millennium. Oxford: Elsevier Ltd.

Lundtorp, S. 2001. "Measuring Tourism Seasonality." In T. Baum, and S. Lundtrop, ed.
Seasonality in Tourism. Oxford, UK: Elsevier Science Ltd. Pp. 23-50.

Maekawa, K. and H. Hisamatsu. 2002. "SUR Models with Integrated Regressors." In A. Ullah,
A.T.K. Wan, and A. Chaturvedi, ed. Handbook ofApplied Econometrics and Statistical
Inference. New York: M. Dekker Inc. Pp. 469-490.

Martin, C.A. and S.F. Witt. 1989. "Forecasting Tourism Demand: A Comparison of the
Accuracy of Several Quantitative Methods." International Journal ofForecasting 5:7-19.

Naude, W.A. and A. Saayman. 2004. "The Determinants of Tourist Arrivals in Africa: A Panel
Data Regression Analysis." Paper presented at International Conference, Centre for the
Study of African Economies, Oxford UK, 21-22 March.









Nerlove, M. 1958. "Distributed Lags and Estimation of Long Run Supply and Demand
Elasticities: Theoretical Considerations." Journal of Farm Economics 40(2):301-311.

Nerlove, M. 1972. "Lags in Economic Behavior." Econometrica 40(2):221-246.

Nielsen, S.K. 2001. "Determinants of Air Travel Growth." World Transport Policy and Practice
7(2):28-37.

Oh, C. and B.J. Morzuch. 2005. "Evaluation Time Series Models to Forecast the Demand for
Tourism in Singapore: Comparing Within Sample and Post-Sample Results." Journal of
Travel Research 43:404-413.

Phakdisoth, L. and D. Kim. 2007. "The Determinants of Inbound Tourism in Laos." ASEAN
Economic Bulletin. Volume 24, No. 2. Pp. 225-237. August.

Pollak, R.A. 1970. "Habit Formation and Dynamic Demand Functions." The Journal of Political
Economy 78(4):745-763.

Proenca, S.A. and E. Soukiazis. 2005. "Demand for Tourism in Portugal: A Panel Data
Approach." Discussion Paper No. 29, Coimbra PT, February.

Sheldon, P.J. and T. Var. 1985. "Tourism Forecasting: A Review of Empirical Research."
Journal ofForecasting 4(2): 183-195.

Smeral, E. and A. Weber. 2000. "Forecasting International Tourism Trends to 2010." Annals of
Tourism Research 27(4):982-1006.

Song, H. and S.F. Witt. 2006. "Forecasting International Tourist Flows to Macau." Tourism
Management 27:214-224.

Syriopuolos, T. 1995. "A Dynamic Model of Demand for Mediterranean Tourism." International
Review ofApplied Economics 9:318-336.

Tan, A.Y.F., C. McCahon, and J. Miller. 2002. "Modeling Tourist Flows to Indonesia and
Malaysia." Journal of Travel and Tourism Marketing 13(1/2):63-84.

Thapa, B., S.M. Holland, and J. Absher. 2003. "Wildfires and Tourist Behaviors in Florida."
RWU 4902 Technical Report submitted to USDA-Forest Service. March.

Uysal, M. and M.S. El Roubi. 1999. "Artificial Neural Networks versus Multiple Regression in
Tourism Demand Analysis." Journal of Travel Research 38:111-118.

Vanhove, N. 2005. The Economics of Tourism Destinations. Boston: Elsevier Butterworth-
Heinemann.

Visit FloridaC Research. 2006. Historic Visitor Numbers. Tallahassee.

Ward, R.W. and J.E. Davis. 1978. "A Pooled Cross-Section Time Series Model of Coupon
Promotions." American Journal ofAgricultural Economics 60(3):393-401.

296









Witt, S.F., H. Song, and P. Louvieris. 2003. "Statistical Testing in Forecasting Model Selection."
Journal of Travel Research 42:151-158.

Witt, S.F., G.D. Newbould, and A.J. Watkins. 1992. "Forecasting Domestic Tourism Demand:
Application to Las Vega Arrivals Data." Journal of Travel Research 31:36-41.

Witt, S.F. and C.A. Witt. 1995. "Forecasting Tourism Demand: A Review of Empirical
Research." International Journal ofForecasting 11:447-475.

World Travel and Tourism Council. 2007. World Travel and Tourism: Navigating the Path
Ahead: The 2007 Travel & Tourism Economic Research. London.

Young, K.H. 1972. "A Synthesis of Time Series and Cross-Section Analyses: Demand for Air
Transportation Services." Journal of the American Statistical Association 67(339):550-
566.

Zhang, A. and Y. Zhang. 2006. "Airport Capacity and Congestion when Carriers have Market
Power." Journal of Urban Economics 60:229-247.









BIOGRAPHICAL SKETCH

Jose Cazanova is originally from San Pedro Sula, Honduras. He obtained his bachelor's

degree with a specialization in agribusiness, from Zamorano University in 2001, graduating at

the top of his class. He completed his master's degree in business administration in the

Warrington College of Business at the University of Florida in 2003. He then pursued a Ph.D.

degree in food and resource economics.

While working on his Ph.D., Jose was admitted to the World Citizenship Program and had

the opportunity of a lifetime to travel to Ethiopia in the summer of 2006. There, he volunteered

for Project Concern International (PCI), a non-profit organization. During the three months he

spent in Addis Ababa, Jose conducted a value chain analysis for the silk industry in Ethiopia.

The analysis included an assessment of the production chain and distribution channels and a

marketing study aimed to identify potential international markets for silk products. The final

report provided recommendations for policy implementation that were aimed to improve the

value of silk, and ultimately, the livelihood of hundreds of Ethiopian families.

Jose also completed a study abroad in Madrid, Spain during the fall of 2006. His

coursework at the Universidad Politecnica de Madrid included econometrics, international

development, and marketing. He also conducted research in generic advertising of various

Spanish agricultural products with his colleague Paul Jaramillo at the Ministerio de Agricultura,

Pesca, y Alimentacion.

At every stage of his education career, Jose has served in a leadership capacity through

student councils where he has applied his organizational and communication skills to help

improve the academic experience of his peers. He has been elected to serve as class president

multiple times since high school and have actively participated in faculty-student discussion

boards. Jose was the president of the FRED Graduate Student Organization in 2006 and has

298









remained actively involved for the past four years. After graduation, he plans to work in projects

with an international scope where he can apply his agribusiness and economics knowledge.





PAGE 1

1 DEMAND FOR AIR PASSENGER T RAFFIC AND ITS IMPACT ON THE TOURISM INDUSTRY OF FLORIDA By JOSE CAZANOVA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS F OR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Jose Cazanova

PAGE 3

3 To my Mom and Dad.

PAGE 4

4 ACKNOWLEDGMENTS First and foremost, I thank God for the opportunity to pursue and complete this d octoral program. I thank my family for their undying love and patience during this long academic journey that started on a sunny day in Zamorano, Honduras ten years ago I would like to thank Mom and Dad for their unwavering support and great sacrifices ma de throughout I could not have asked for better parents I also extend my deepest appreciation to my brothers, Mauricio and Levi who ha ve also made sacrifices to help me achieve this goal. I also thank Nini, my little booboo, for making the days more enjo yable. There are no word s to express my gratitude and overall admiration to my chair, Dr. Ronald W. Ward a great mentor and even better person He has been the best mentor and source of inspiration someone could ask for during this immense task. I remembe r when we first talk about the research idea and its complexity. He told me that it would be a lot of work that will require great dedication. I took th e challenging task because I knew that with his help, the project was going to be very rewarding I gre atly appreciate his hard work, excellent advice, and well timed guidance during my research I am very grateful for his words of encouragement every time I felt overwhelmed by the immensity of this research All my struggles were lessen ed by his patience a nd help. I would also like to thank him for giving me the opportunity to pursue my internship in Ethiopia and also for encourag ing me to study abroad in Spain. It has bee n a great honor to work with him I also thank Dr. Richard Kilmer, who always seemed to have a comment of support when he saw m e struggle in the office I am grateful for his valuable feedback given in the dissertation I have also enjoyed our great conversations on Gator sports. I would also like to thank Dr. Robert Emerson and Dr. Stephe n Holland Their time and effort spent on my dissertation are greatly

PAGE 5

5 appreciated. My appreciation also extends to Dr. Antonio Flores for his feedback on my dissertation that has enabled me to improve its quality. I also thank Dr. Burkhardt. I cannot thank him enough for his great support and advice to pursue the internship in Ethiopia and the study abroad in Spain. When things seemed to be impossible, he found a way to make sure I had the se great opportunities to travel the world. My appreciation also is e xtended to Dr. Jane Luzar. Without her help, I might never have had the opportunity to pursue the PhD. I am thankful for her constant support and dedication she ha s offered to every Zamorano who has knocked on her door. I would like to thank all of my frie nds who have helped me while writing my dissertation. These include my dear friends from the Food and Resource Economics Department and Gainesville and to those who are living across the United States. My appreciation also extends to my friends living in H onduras, Bolivia, Ethiopia and Spain who have always sent me e mails expressing their great support. I would like to extend a special thank you to Marcos G. who has al ways expressed nothing but sincere words of encouragement to me I could not finish this acknowledgement without expressing how proud I am to be a Fl orida Gator. Go Gators!

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ......................... 10 LIST OF FIGURES ................................ ................................ ................................ ....................... 12 ABSTRACT ................................ ................................ ................................ ................................ ... 22 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ................... 23 Airline Transportation and Tourism ................................ ................................ ....................... 24 Problem Statement ................................ ................................ ................................ .................. 26 Research Objectives ................................ ................................ ................................ ................ 26 Research Methodology ................................ ................................ ................................ ........... 27 Airline Passenger Traffic Partial Adjustment Model ................................ ...................... 28 Data and Scope ................................ ................................ ................................ ....................... 29 2 LITERATURE REVIEW ................................ ................................ ................................ ........ 32 Forecasting Tourism Demand ................................ ................................ ................................ 32 Scope of Recent Tourism Demand Studies ................................ ................................ ............ 35 Econometric Techniques Applied to Tourism Demand Analysis ................................ .......... 38 Determinants of Tourism Demand ................................ ................................ ......................... 41 Static versus Dynamic Regression Models ................................ ................................ ............. 45 Challenges in Tourism Demand Analysis ................................ ................................ .............. 47 Recent Developments in Tourism Demand Analysis ................................ ............................. 51 Chapter Summary ................................ ................................ ................................ ................... 52 3 DESCRIPTIVE STATISTIC S ON THE AIRLINE PASSENGER TRAFFIC TRAVELING TO FLORIDA AND OTHER INDICATORS ................................ ................. 53 Airline Passenger Traffic ................................ ................................ ................................ ........ 54 Description, Select ion, and Aggregation of Domestic Airline Passenger Traffic Data ................................ ................................ ................................ .............................. 54 Description, Selection, and Aggregation of International Airline Passenger Traffic Data ................................ ................................ ................................ .............................. 56 Domestic and International Airline Passenger Traffic Traveling to Florida ................... 57 Domestic airline passenger traffic by U.S. region ................................ .................... 58 Domestic airline passenger traffic by destination CSA ................................ ........... 61 South Florida CSA domestic airline passenger traffic ................................ ............. 62 Orlando CSA domestic airline passenger traffic ................................ ...................... 65 Tampa St. Petersburg CSA domestic airline passenger traffic ................................ 67

PAGE 7

7 Jacksonville CSA domestic airline passenger traffic ................................ ............... 70 Fort Myers CSA domestic airline passenger traffic ................................ ................. 72 International airl ine passenger traffic by world region ................................ ............ 75 International airline passenger traffic by destination CSA ................................ ...... 77 South Florida CSA interna tional airline passenger traffic ................................ ....... 78 Orlando CSA international airline passenger traffic ................................ ................ 81 Tampa St. Petersburg CSA internationa l airline passenger traffic .......................... 83 Airline Ticket Prices ................................ ................................ ................................ ............... 86 Description, Selection, and Aggregation of Airline Ticket Price Data for Domestic Flights ................................ ................................ ................................ .......................... 86 Description, Selection, and Aggregation of Airline Ticket Price Data for International Flights ................................ ................................ ................................ ..... 88 Domestic Airline Ticket Prices by U.S. Region ................................ .............................. 89 Domestic Airline Ticket Prices by Destination CSA ................................ ...................... 92 South Florida CSA domestic airline ticket prices ................................ .................... 93 Orlando CSA domestic airline ticket prices ................................ ............................. 96 Tampa St. Petersburg CSA domestic airline ticket prices ................................ ....... 99 Jacksonville CSA domestic airline ticket prices ................................ .................... 102 Fort Myers CSA domestic airline ticket prices ................................ ...................... 105 International Airline Round Trip Airline Ticket Prices by World Region .................... 108 International Airline Round Trip Airline Ticket Prices by Destination CSA ............... 110 South Florida CSA international airline ticket prices ................................ ............ 111 Orlando CSA international airline ticket prices ................................ ..................... 113 Tampa St. Petersburg CSA international airline ticket prices ............................... 114 Freight and Mail Transported via Commercial Passenger Airlines to Florida ..................... 116 Economic, Social, and Weather Indicators ................................ ................................ ........... 127 Gross Domestic Product, Personal Disposable Income, and Population ...................... 127 Brand and Generic Advertising Expenditures ................................ ............................... 131 Description, selection, and aggregation of advertising expenditures data ............. 131 Brand advertising expenditures ................................ ................................ .............. 132 Generic advertising expenditures ................................ ................................ ........... 136 Foreign Exchange Rate: Euro to U.S. Dollar ................................ ................................ 138 Historic Jet Fuel Prices ................................ ................................ ................................ .. 139 Hurricanes and Wildfires Affecting Florida ................................ ................................ .. 140 Average Temperatures in Origin Regions and Destination CSAs ................................ 144 Precipitation in Florida ................................ ................................ ................................ .. 145 Crime Rates in Florida ................................ ................................ ................................ ... 146 Chapter Summary ................................ ................................ ................................ ................. 147 4 THEORETICAL FRAMEWORK ................................ ................................ ......................... 149 Motivation ................................ ................................ ................................ ............................. 149 The Partial Adjustment Model ................................ ................................ .............................. 152 Domestic Air Passenger Traffic Partial Adjustment Model ................................ ................. 154 Construction of Empirical Domestic Air Passenger Model ................................ ................. 156 Description of Variables in the Static Component of the DAP PAM ........................... 157 Identification of the Variables in the Dynamic Component of the DAP PAM ............ 161

PAGE 8

8 Estimation Possibilities ................................ ................................ ................................ ......... 166 Chapter Summary ................................ ................................ ................................ ................. 169 5 RESULTS ................................ ................................ ................................ .............................. 170 Comparison of the Estimation Alternatives ................................ ................................ .......... 170 Results for Demand for Passengers by CSA ................................ ................................ ........ 171 Florida Results ................................ ................................ ................................ .............. 173 Analysis of c oefficients in dynamic component ................................ .................... 173 Analysis of coefficients in static component ................................ .......................... 173 South Florida CSA Results ................................ ................................ ........................... 174 Analysis of coefficients in dynamic component ................................ .................... 175 Analysis of coefficients in static component ................................ .......................... 175 Orlando CSA Results ................................ ................................ ................................ .... 176 Analysis of coefficients in dynamic component ................................ .................... 176 Analysis of coefficients in sta tic component ................................ .......................... 177 Tampa St. Petersburg CSA Results ................................ ................................ .............. 178 Analysis of coefficients in dynamic component ................................ .................... 178 Analysis of coefficients in static component ................................ .......................... 178 Jacksonville CSA Results ................................ ................................ ............................. 179 Analysis of coefficients in dynamic component ................................ .................... 180 Analysis of coefficients in static component ................................ .......................... 180 Fort Myers CSA Results ................................ ................................ ............................... 181 Analysis of coefficients in dynamic component ................................ .................... 181 Analysis of coefficients in static component ................................ .......................... 182 Chapter Summary ................................ ................................ ................................ ................. 183 6 SIMULATION ANALYSIS ................................ ................................ ................................ .. 194 Introduction ................................ ................................ ................................ ........................... 194 Simulations for Income ................................ ................................ ................................ ......... 197 Florida CSA Income Simulations ................................ ................................ ................. 197 South Florida CSA Income Simulations ................................ ................................ ....... 201 Orlando CSA Income Simulations ................................ ................................ ................ 204 Tampa St. Petersburg CSA Income Simulations ................................ .......................... 207 Jacksonville CSA Income Simulations ................................ ................................ ......... 210 Fort Myers CSA Income Simulations ................................ ................................ ........... 213 Simulations for Airline Ticket Pric es ................................ ................................ ................... 216 Florida CSA Airline Ticket Price Simulations ................................ .............................. 217 South Florida CSA Airline Ticket Price Simulations ................................ ................... 219 Tampa St. Petersburg CSA Airline Ticket Price Simulations ................................ ...... 221 Jacksonville CSA Airline Ticket Price Simulations ................................ ..................... 223 Fort Myers CSA Airline Ticket Price Simulations ................................ ....................... 225 Simulations for Terror ................................ ................................ ................................ .......... 226 Florida CSA Te rror Simulations ................................ ................................ ................... 227 South Florida CSA Terror Simulations ................................ ................................ ......... 229 Orlando CSA Terror Simulations ................................ ................................ .................. 23 1

PAGE 9

9 Tampa St. Petersburg CSA Terror Simulations ................................ ............................ 233 Jacksonville CSA Terror Simulations ................................ ................................ ........... 235 Fort My ers CSA Terror Simulations ................................ ................................ ............. 237 Simulations for Hurricanes ................................ ................................ ................................ ... 239 Florida CSA Hurricane Simulations ................................ ................................ ............. 239 South Florida CSA Hurricane Simulations ................................ ................................ ... 242 Orlando CSA Hurricane Simulations ................................ ................................ ............ 244 Tampa St. Petersburg CSA Hurricane Simulations ................................ ...................... 246 Jacksonville CSA Hurricane Simulations ................................ ................................ ..... 248 Fort Myers CSA Hurricane Simulations ................................ ................................ ....... 249 Simulations for Seasonality ................................ ................................ ................................ .. 250 Florida CSA Seasonality ................................ ................................ ............................... 250 South F lorida CSA Seasonality ................................ ................................ ..................... 252 Orlando CSA Seasonality ................................ ................................ ............................. 254 Tampa St. Petersburg CSA Seasonality ................................ ................................ ........ 256 Jacksonville CSA Seasonality ................................ ................................ ....................... 258 Fort Myers CSA Seasonality ................................ ................................ ......................... 260 Simulations for Fire, Rainfall, and Temperature ................................ ................................ .. 262 Chapter Summary ................................ ................................ ................................ ................. 262 7 SUMMARY, CONCLUSIONS, IMPLICATIONS, AND FUTURE RESEARCH ............. 268 Conclusions ................................ ................................ ................................ ........................... 269 Implications ................................ ................................ ................................ .......................... 274 Future Research ................................ ................................ ................................ .................... 276 APPENDIX A DESCRIPTION OF DESTINATION CSA AIRPORTS ................................ ....................... 278 B TRANSFORMATION OF VARIABLES: TSP PROGRAMS ................................ ............. 279 C ESTIMATION OF MODEL: TSP PROGRAMS ................................ ................................ .. 283 D SIMULATION ANALYSIS: TSP PROGRAMS ................................ ................................ .. 284 E POPULATION FACTORS BY U.S. REGION ................................ ................................ ..... 292 LIST OF REFERENCES ................................ ................................ ................................ ............. 294 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 298

PAGE 10

10 LIST OF TABLES Table page 3 1 Geographic region sche me as defined by the United States Census. ................................ 55 4 1 Summary of the variables included in the DAP PAM. ................................ .................... 163 4 2 Identification of destination CSAs and origin U.S. regions. ................................ ............ 165 5 1 29 and their corresponding t value using five different estimation approaches. ................................ ................................ ................................ ..... 187 5 2 Florida CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U.S. regions using the SUR AR1 ALL approach. ................................ ................................ ................................ .......................... 188 5 3 South Florida CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U. S. regions using the SUR AR1 ALL approach. ................................ ................................ ................................ .......................... 189 5 4 Orlando CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U.S. regions using the SUR AR1 ALL approach. ................................ ................................ ................................ .......................... 190 5 5 Tampa St. Petersburg CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U.S. regions using the SUR AR1 ALL approach. ................................ ................................ ................................ ........ 191 5 6 Jacksonville CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U.S. regions using the SUR AR1 ALL approach. ................................ ................................ ................................ .......................... 192 5 7 Fort Myers CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U.S. regions using the SUR AR1 ALL approach. ................................ ................................ ................................ .......................... 193 6 1 Comparison between the simulated demand in the presence of the 9 11 terrorist attacks and the simulate d demand in the absence of 9 11 terrorist attacks. .................... 266 6 2 Comparison between the simulated demand in the presence of hurricanes and the simulated demand in the absence hurricanes during the hurricane season (June November). ................................ ................................ ................................ ...................... 267 E 1 Northeast region (1001): population factors by month and by year between 1996 and 2006. ................................ ................................ ................................ ................................ 292

PAGE 11

11 E 2 Midwest region (1002): population factors by month and by year between 1996 and 2006. ................................ ................................ ................................ ................................ 292 E 3 South region (1003): population factors by month and by year between 1996 and 2006. ................................ ................................ ................................ ................................ 293 E 4 West region (1004): population factors by month and by year between 1996 and 2006. ................................ ................................ ................................ ................................ 293

PAGE 12

12 LIST OF FIGURES Figure page 1 1 Share of U.S. domestic airline passenger traffic by state in 2006. ................................ .... 25 3 1 Monthly seasonal pattern of domestic and international airline passenger traffic traveling to Florida between 1990 and 2006. ................................ ................................ ..... 58 3 2 Florida CSA: total domestic airline passenger traffic from five U.S. regions between 1990 and 2006. ................................ ................................ ................................ ................... 60 3 3 Florida CSA: share of total domestic airline passenger traffic from five U.S. regions in 1990, 1998, and 2006. ................................ ................................ ................................ .... 60 3 4 Florida CSA: monthly seasonal patte rn of domestic airline passenger traffic from four U.S. regions between 1990 and 2006. ................................ ................................ ........ 61 3 5 Total domestic airline passenger traffic traveling to the destination CSAs in Florida between 1990 and 2006. ................................ ................................ ................................ .... 62 3 6 South Florida CSA: total domestic airline passenger traffic from five U.S. regions between 1990 and 2006. ................................ ................................ ................................ .... 63 3 7 South Florida CSA: share of total domestic airline passenger traffic from five U.S. regions in 1990, 1998, and 2006. ................................ ................................ ....................... 64 3 8 South Florida CSA: monthly seasonal pattern of domestic airline passenger traffic from four U.S regions between 1990 and 2006. ................................ ............................... 64 3 9 Orlando CSA: total domestic airline p assenger traffic from five U.S. regions between 1990 and 2006. ................................ ................................ ................................ ................... 66 3 10 Orland o CSA: share of total domestic airline passenger traffic from five U.S. regions in 1990, 1998, and 2006. ................................ ................................ ................................ .... 66 3 11 Orlando CSA: monthly seasonal pattern of domestic airline passenger traffic from four U.S. regions between 1990 and 2006. ................................ ................................ ........ 67 3 12 Tampa St. Petersburg CSA: total airline passenger traffic from five U.S. regions between 1990 and 2006. ................................ ................................ ................................ .... 68 3 13 Tampa St. Petersburg CSA: share of total domestic airline passenger traffic from five U.S. regions in 1990 1998, and 2006. ................................ ................................ ........ 69 3 14 Tampa St. Petersburg CSA: monthly seasonal pattern of domestic airline passenger traffic from four U.S. regions between 1990 and 2006. ................................ .................... 69

PAGE 13

13 3 15 Jacksonville CSA: total domestic airline passenger traffic from five U.S. regions between 1990 and 2006. ................................ ................................ ................................ .... 71 3 16 Jacksonville CSA: share of total domestic airline passenger traffic from four U.S. regions in 1990, 1998, and 2006. ................................ ................................ ....................... 71 3 17 Jacksonville CSA: monthly seasonal pattern of domestic airline passenger traffic from four U.S. regions between 1990 and 2006. ................................ ............................... 72 3 18 Fort Myers CSA: total domestic airline passenger traffic from five U.S. regions between 1990 and 2 006. ................................ ................................ ................................ .... 73 3 19 Fort Myers CSA: share of total domestic airli ne passenger traffic from four U.S. regions in 1990, 1998, and 2006. ................................ ................................ ....................... 74 3 20 Fort Myers CSA: monthly seasonal pattern of domestic airline passenger traffic from four U.S. regions between 1990 and 2006. ................................ ................................ ........ 74 3 21 Florida CSA: total international airline passenger traffic from four world regions between 1990 and 2006. ................................ ................................ ................................ .... 76 3 2 2 Florida CSA: share of total international airline passenger traffic from fou r world regions in 1990, 1998, and 2006. ................................ ................................ ....................... 76 3 2 3 Florida CSA: monthly seasona l pattern of total international airline passenger traffic from four world regions between 1990 and 2006. ................................ ............................. 77 3 24 Total international airline passenger traffic traveling to the top three destination CSAs in Florida between 1990 and 2006. ................................ ................................ .................... 78 3 2 5 South Florida CSA: total international airline passenger traffic from four world regions between 1990 and 2006. ................................ ................................ ........................ 79 3 2 6 South Florida CSA: share of total international airline passenger traffic from four world regions in 1990, 1998, and 2006. ................................ ................................ ............. 80 3 2 7 South Florida CSA: monthly seasonal pattern of total international airline passenger from four world regions between 1990 and 2006. ................................ ............................. 80 3 2 8 Orlando CSA: total international airline passenger traffic from four world regions between 1990 and 2006. ................................ ................................ ................................ .... 82 3 2 9 Orlando CSA: share of total international airline passenger traffic from four world regions in 1990, 1998, and 2006. ................................ ................................ ....................... 82 3 3 0 Orlando CSA: monthly seasonal pattern of total international airline passenger traffic from four world regions between 199 0 and 2006. ................................ ............................. 83

PAGE 14

14 3 3 1 Tampa St. Petersburg CSA: total international airline pa ssenger traffic from four world regions between 1990 and 2006. ................................ ................................ ............. 84 3 3 2 Tampa St. Petersburg CSA: share of total international airline passenger traffic from four world regions in 1990, 1998, and 2006. ................................ ................................ ..... 85 3 3 3 Tampa St. Petersburg CSA: monthly seasonal pattern of total international airline passenger traffic from four world regions between 1990 and 2006. ................................ 85 3 3 4 Florida CSA: average one way airline ticket prices for domestic flights from four U.S. regio ns between 1993 and 2006. ................................ ................................ ................ 90 3 3 5 Florida CSA: average round trip airline ti cket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ................ 91 3 3 6 Florida CSA: quarterly seasonal pattern of average one way (OW) and round trip (RT) airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ................................ ............................ 92 3 3 7 South Florida CSA: average one way airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006 ................................ ................................ ........ 94 3 3 8 South Florida CSA: average round trip air ticket prices for domest ic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ................ 95 3 3 9 South Florid a CSA: quarterly seasonal pattern of average one way (OW) and round trip (RT) airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ................................ ................... 96 3 4 0 Orlando CSA: average one way airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ................ 97 3 4 1 Orlando CSA: average round trip air ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ........................ 98 3 4 2 Orlando CSA: quarterly seasonal pattern of average one way (OW) and round trip (RT) airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ................................ ............................ 99 3 4 3 Tampa St. Petersburg CSA: average one way airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ............................. 100 3 4 4 Tampa St. Petersburg CSA: average round trip air ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ............................. 101 3 4 5 Tampa St. Petersburg CSA: quarterly seasonal pattern of average one way (OW) and round trip (RT) airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ................................ .. 102

PAGE 15

15 3 4 6 Jacksonville CSA: average one way airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ...... 103 3 4 7 Jacksonville CSA: average round trip air ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ .............. 104 3 4 8 Jacksonville CSA: quarterly seasonal pattern of average one way (OW) and round trip (RT) airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ................................ ................. 105 3 4 9 Fort Myers CSA: average one way airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ .............. 106 3 5 0 Fort Myers CSA: average round trip air ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ .............. 107 3 5 1 Fort Myers CSA: quarterly seasonal pattern of average one way (OW) a nd round trip (RT) airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. ................................ ................................ ................................ .......................... 108 3 5 2 Florida CSA: average international round trip airline ticket prices from three world regions between 1995 and 2006. ................................ ................................ ...................... 109 3 5 3 Florida CSA: monthly seasonal pattern of relative change from the average round trip airline ticket prices from three world regions between 1995 and 2006. ................... 11 0 3 5 4 South Florida CSA: average international round trip airline ticket prices from three world regions between 1995 and 2006. ................................ ................................ ........... 111 3 5 5 South Florida CSA: m onthly seasonal pattern of relative change from the average round trip airline ticket prices from three world regions between 1995 and 2006. ......... 112 3 5 6 Orlando CSA: average international round trip airline ticket prices from three world regions between 1995 and 2006. ................................ ................................ ...................... 113 3 5 7 Orlando CSA: monthly seasonal pattern of relative change from the average round trip airline ticket prices from three world regions betw een 1995 and 2006. ................... 114 3 5 8 Tampa St. Petersburg CSA: average international ro und trip airline ticket prices from three world regions between 1995 and 2006 ................................ ................................ ... 115 3 5 9 Tampa St. Petersburg CSA: monthly seasonal pattern of the relative change from the average round trip airline ticket prices from three world regions between 1995 and 2006 ................................ ................................ ................................ ................................ 116 3 6 0 Florida CSA: total domestic freight transported from four U.S. reg ions using commercial passenger airlines between 1990 and 2006. ................................ ................. 118

PAGE 16

16 3 6 1 Flor ida CSA: monthly seasonal pattern of total freight transported from four U.S. regions using commercial passenger airlines between 1990 and 2006. ........................... 118 3 6 2 Total domestic freight transported to six destination CSAs in Florida using commercial passenger airlines between 1990 and 2006. ................................ ................. 119 3 6 3 Florida CSA: total domestic mail transported from four U.S. regions using commercial passenger airlines between 1990 and 2006. ................................ ................. 120 3 6 4 Florida CSA: monthly seasonal pattern of total mail transporte d from four U.S. regions using commercial passenger airlines between 1990 and 2006. ........................... 121 3 6 5 Total domestic mail transported to six destination CSAs in Florida using commercial passenger airlines between 1990 and 2006. ................................ ................................ ..... 122 3 6 6 Florida CSA: total international freight transported from three world regions using commercial passenger airlines between 1990 and 2006. ................................ ................. 123 3 6 7 Florida CSA: monthly seasonal pattern of total freight transported from three world regions using commercial passenger airlines between 1990 and 2006. ........................... 123 3 6 8 Total international freight transported to six destination CSAs in Florida using commercial passenger airlines between 1990 and 2006. ................................ ................. 124 3 6 9 Florida CSA: total international mail transported from four world regions using commercial passenger airlines between 1990 and 2006. ................................ ................. 125 3 7 0 Florida CSA: monthly seasonal pattern of total mail transported from three world regions using commercial pass enger airlines between 1990 and 2006. ........................... 126 3 7 1 Total international mail trans ported to six destination CSAs in Florida using commercial passenger airlines between 1990 and 2006. ................................ ................. 127 3 7 2 Annual gross domestic product from four U.S. regions between 1990 and 2006. .......... 129 3 7 3 Annual per capita personal disposable income from four U.S. regions between 1990 and 2006. ................................ ................................ ................................ .......................... 130 3 7 4 Annual population estimates from four U.S. regions between 1990 and 2006. .............. 130 3 7 5 Total brand advertising expenditures from four tourism related com panies in Florida between 1995 and 2006. ................................ ................................ ................................ .. 133 3 7 6 Florida attractions: monthly seasonal pattern of total brand adverting expenditures from three tourism related companies between 1995 and 2006. ................................ ..... 134 3 7 7 Florida and non Florida attractions: monthly seasonal pattern of total brand adverting expenditures from three tourism related companies between 1995 and 2006. ................ 135

PAGE 17

17 3 7 8 Total generic advertising expenditures spent by government entities from Florida between 2002 and 2006. ................................ ................................ ................................ .. 136 3 7 9 Monthly seasonal pattern of total generic adverting expenditures spent by city, county, and state government to promote Florida between 2002 and 2006. ................... 137 3 8 0 Annual air passenger traffic traveling from Europe to Florida and annual exchange rate (Euro to USD) between 1990 and 2006. ................................ ................................ ... 138 3 8 1 Historic average kerosene type jet fuel prices from five worldwide locations between 1990 and 2006. ................................ ................................ ................................ ................. 139 3 8 2 Monthly seasonal pattern of historic average kerosene type jet fuel prices from five worldwide locations between 1990 and 2006 ................................ ................................ .. 140 3 8 3 Number of tropical storms (by category) affecting Florida between 1990 and 2006. ..... 141 3 8 4 Number of wildfires (by category) affectin g Florida between 1990 and 2006 ............... 143 3 8 5 Percentage of wildfires and acreage burned in Florida (by month) between 1 990 and 200 6. ................................ ................................ ................................ ................................ 143 3 8 6 Difference between average temperatures in Florida and each U.S. region betwee n 1990 and 2006 ................................ ................................ ................................ ................. 145 3 8 7 Monthly average rainfall in each of the five destination CSAs in Florida between 1990 and 2006. ................................ ................................ ................................ ................. 146 3 8 8 Annual crime index in F lorida between 1990 and 2006. ................................ ................. 147 4 1 Power spectrum of domestic air passenger traffic data for the Florida Northeast (33 1001) pair. ................................ ................................ ................................ ........................ 162 5 1 Florida CSA: effect of terrorism dummies on demand for airline passengers from four U.S. regions. ................................ ................................ ................................ ............ 184 5 2 South Florida CSA: effect of terrorism dummies on demand for airline passengers from four U.S. regions. ................................ ................................ ................................ ... 184 5 3 Orlando CSA: effect of terrorism dummies on demand for airline passengers from four U.S. regions. ................................ ................................ ................................ ............. 185 5 4 Tampa St. Petersburg CSA: effect of terrorism dummies on demand for airline passengers from four U.S. regions. ................................ ................................ ................. 185 5 5 Jacksonville CSA: effect of terro rism dummies on demand for airline passengers from four U.S. regions. ................................ ................................ ................................ .... 186

PAGE 18

18 5 6 Fort Myers CSA: effect of terrorism dummies o n demand for airline passengers from four U.S. regions. ................................ ................................ ................................ ............. 186 6 1 Florida CSA Income simulations: relationship between differ ent levels of personal disposable income and domestic demand for air passengers from four U.S. regions. .... 199 6 2 Florida CSA Income simulations: short and long run changes in domestic demand for air passengers from four U.S. regions at di fferent levels of per capita personal disposable income. ................................ ................................ ................................ ........... 200 6 3 South Florida CSA Income simulations: relationship between different levels of personal disposable income and number of passengers for each of the four U.S. regions. ................................ ................................ ................................ ............................. 202 6 4 South Florida CSA Income simulations: short and long run changes in the number of passengers at different levels of per capita personal disposable income for each of the four U.S. regions. ................................ ................................ ................................ ............. 203 6 5 Orlando CSA Income simulations: relationship between different levels of personal disposable income and number of passengers for each of the four U.S. regions. ........... 205 6 6 Orlando CSA Income simulations: short and long run changes in the number of passengers at different levels of per capita personal disposable income for each of the four U.S. regions. ................................ ................................ ................................ ............. 206 6 7 Tampa St. Petersburg CSA Income simulations: r elationship between different levels of personal disposable income and number of passengers for each of the four U.S. regions. ................................ ................................ ................................ ..................... 208 6 8 Tampa St. Petersburg CSA Income simulations: short and long run changes in the number of passengers at different levels of per capita personal disposable income for each of the four U.S. regions. ................................ ................................ .......................... 209 6 9 Jacksonville CSA Income simulations: relationship between different levels of personal disposable income and number of passengers for eac h of the four U.S. regions. ................................ ................................ ................................ ............................. 211 6 1 0 Jacksonville CSA Income simulations: short and long run changes in the number of passengers at different levels of per capita personal disposable income for each of the four U.S. regions. ................................ ................................ ................................ ............. 212 6 1 1 Fort Myers CSA Income simulations: relationship between different levels of personal disposable income and number of passengers for each of the four U.S. regions. ................................ ................................ ................................ ............................. 214 6 1 2 Fort Myers CSA Income simulations: short and long run changes in the number of passengers at different levels of per capita personal disposable income for each of the four U.S. regions. ................................ ................................ ................................ ............. 215

PAGE 19

19 6 1 3 Florida CSA Airline ticket price simulations: A) Relationship between different levels of ai rline ticket prices and number of passengers; and B) Short and long run changes in the number of passengers at different levels of airline ticket prices for two U.S. regions. ................................ ................................ ................................ ..................... 218 6 1 4 South Florida CSA Airline ticket price simulations: A) Relationship between different levels of airline ticket prices and number of passengers; and B) Short and long run changes in the number of passengers at different levels of airline ticket prices for three U.S. regions. ................................ ................................ ........................... 220 6 1 5 Tampa St Petersburg CSA Airline ticket price simulations: A) Relationship between different levels of airline ticket prices and number of passengers; and B) Short and long run changes in the number of passengers at different levels of airline ticket prices for tw o U.S. regions. ................................ ................................ ............................. 222 6 1 6 Jacksonville CSA Airline ticket price simulations: A) Relationship between different levels of airline ticket prices and number of passengers; and B) Short and long run c hanges in the number of passengers at different levels of airline ticket prices for two U.S. regions. ................................ ................................ ................................ ..................... 224 6 1 7 Fort Myers CSA Airline ticket price simulations: A) Relationship between different levels of airline ticket prices and number of passengers; and B) Short and long run changes in the number of passengers at different levels of airline ticket prices for the Northeast region. ................................ ................................ ................................ .............. 225 6 1 8 Florida CSA Terror simulations: comparison between the simulated results of dema nd for air passengers from four U.S. regions in the presence and absence of the 9 11 terrorist attacks. ................................ ................................ ................................ ....... 228 6 1 9 South Flor ida CSA Terror simulations: comparison between the simulated results of demand for air passengers from four U.S. regions in the presence and absence of the 9 11 terrorist attacks. ................................ ................................ ................................ ....... 230 6 2 0 Orlando CSA Terror simulations: comparison between the simulated results of demand for air passengers from four U.S. regions in the presence and absence of the 9 11 terrorist attacks. ................................ ................................ ................................ ....... 232 6 2 1 Tampa St. Petersburg CSA Terror simulations: comparison between the simulated results of the demand for passengers fro m four U.S. regions in the presence and absence of the 9 11 terrorist attacks. ................................ ................................ ................ 234 6 2 2 Jacksonville CSA Terror simulation s: comparison between the simulated results of the demand for air passengers from four U.S. regions in the presence and absence of the 9 11 terrorist attacks. ................................ ................................ ................................ .. 236 6 2 3 Fort Myers CSA Terror simulations: comparison between the simulated results of the demand for passengers from four U.S. regions in the presence and absence of the 9 11 terrorist attacks. ................................ ................................ ................................ ....... 238

PAGE 20

20 6 2 4 Florida CSA Hurricane simulations: short and long run changes in the number of passengers from four U.S. regions in the absence of hurricanes in Florida. .................... 241 6 2 5 South Florida CSA Hurricane simulations: short and long run changes in the number of passengers from four U.S. regions in the absence of hurricanes in Florida. ............... 243 6 2 6 Or lando CSA Hurricane simulations: short run and long run changes in the number of passengers from four U.S. regions in the absence of hurricanes in Florida. ............... 245 6 2 7 Tampa St. Petersburg CSA Hurricane simulations: short and long run chang es in the number of passengers from four U.S. regions in the absence of hurricanes in Florida. .. 247 6 2 8 Jacksonville CSA Hurricane simulations: short and long run changes in the number of passengers from four U.S. regions in the absence o f hurricanes in Florida. ............... 248 6 2 9 Fort Myers CSA Hurricane simulations: short and long run changes in the number of passengers from four U.S. regions in the absence of hurricanes in Florida. .................... 249 6 3 0 Florida CSA Seasonality simulations: A) Monthly seasonal pattern of domestic demand for air passengers; and B) Percentage change from the monthly average of domestic demand for air passengers from four U.S. regions. ................................ .......... 251 6 3 1 South Florida CSA Seasonality simulations: monthly seasonal pattern of domestic demand for air passengers from four U.S. regions. ................................ ......................... 252 6 3 2 South Florida CSA Seasonality simulations: percentage change from the monthly average demand for passengers from four U.S. r egions. ................................ ................. 253 6 3 3 Orlando CSA Seasonality simulations: monthly seasonal pattern of domestic demand for a ir passengers from four U.S. regions. ................................ ................................ ....... 254 6 3 4 Orlando CSA Seasonality simulations: percentage change from the monthly aver age domestic demand for air passengers from four U.S. regions. ................................ .......... 255 6 3 5 Tampa St. Petersburg CSA Seasonality simulations: monthly seasonal pattern of domestic demand for air passengers from four U.S. regions. ................................ .......... 256 6 3 6 Tampa St. Petersburg CSA S easonality simulations: percentage change from the monthly average domestic demand for air passengers from four U.S. regions. .............. 257 6 3 7 Jacksonville CSA Seasonality simulations: monthly seasonal pattern of domestic demand for air passengers from four U.S. regions. ................................ ......................... 258 6 38 Jacksonville CSA Seasonality simulations: percentage change from the monthly average domestic demand for air passengers from four U.S. regions. ............................ 259 6 39 Fort Myers CSA Seasonality simulations: monthly seasonal pattern of domestic demand for air passengers from four U.S. regions. ................................ ......................... 260

PAGE 21

21 6 40 Fort Myers CSA Seasonality simulations: percentage change from the monthly average domestic demand for air passengers from four U.S. regions. ............................ 261 6 4 1 Comparison of the demand response from each U.S. region traveling to Florida to a three percent increase in income and airline ticket prices. ................................ .............. 263 6 4 2 Comparison of the demand response in each CSA ORG pair to a three percent increase in income and airline ticket prices. ................................ ................................ .... 264

PAGE 22

22 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DEMAND FOR AIR PASSENGER TRAFFIC AND IT S IMPACT ON THE TOURISM INDUSTRY OF FLORIDA By Jose Cazanova August 2008 Chair: Ronald W. Ward Major: Food and Resource Economics The airline industry plays a key role in the tourism industry, serving as a vital link between consumers and the tourism in dustry. Since nearly one half of the tourists coming to Florida use air transportation, having quantitative methods for both explaining and forecasting air passenger traffic are vital to planning and infrastructure development for the tourist sector and th e state in general. Also, knowing how air travel demand responds to unexpected shocks, such as hurricane s, provides guidelines for emergency planning, risk assessment, and prevention management as well as impacts on the state economy. The primary objecti ve of this research was to develop an understanding of the factors influencing domestic demand for airline travel to Florida. Factors such as prices, income, terrorism, seasonality, storms, wildfires and advertising expenditures were included in the analy sis. Estimation results from the partial adjustment model indicates that overall air passengers respond immediately up to a certain extent to changes in some demand drivers and that the full response to such change is realized in subsequent periods. Also, results indicate that demand for air passenger travel ha s not fully recovered from the terrorist attacks o n September 11, 2001. Simulation analysis indicates that overall demand for air passenger travel to Florida is more sensitive to changes in income and that wild fires have no significant impact o n demand for air passenger travel to Florida.

PAGE 23

23 CHAPTER 1 INTRODUCTION Tourism has mul tiple effects o n various areas such as the economy, society, culture and environment of a particular area or country. It has been used as a tool for economic development in both rich and i mpoverished nations. The economic impact of tourism has been widely documented throughout the years. Tourism has been a viable source of income and employment for individuals and a source of re venue for local state and national governments. According to the World Travel and Tourism Council (WTTC), the tourism industry directly employed 72 million people worldwide, accounting for 3.8 % g ross d omestic p roduct (GDP) in 2005. In the United States, direct tourism employment totaled 5.7 million people generating US D 195.08 billion in compensation in 2005. 1 Moreover, 923,064 people worked in Florida representing nearly 16 % of the national tourism workforce in that year. In 2006 s economy, with a real GDP of USD 609.9 billion 2 ranked fourth in the country behind California, New York and Texa s F urthermore, if analyzed as an independent state, Florida would be the 20 th largest economy in the world above countries such as Australi a, Netherlands and Russia. T USD 37.29 billion for that year and is expected to incre ase in coming years. In fact, the World Travel Tourism Council ( WTTC ) predicts that by the year 2010 international to urists will exceed one billion. More and more destinations will be available to tourists and could represent a threat or Despite the positive economic effects derived from tourism (e.g., urban regeneratio n, foreign exchange), some negative effects can be attributed to the industry Fyall and Garrod 1 Travel and Tourism Satellite Accounts. BEA, 2005 2 Chained dollars (2000)

PAGE 24

24 (2005) stated that t ouris m expansion will bring considerable pre ssure on the already fragile natural and cultural environments upon which tourism relies. New ch allenges arise on how to manage growth given the limited resources in area s such as transportation network, energy, water and waste management, and health facilities. Tourists bring traffic congestion, which mea ns more gas consumption due to delays on high ways. Also, airline travelers contribute to noise pollution created by planes affecting neighborhoods close t o airports. Tourists also attract crime to the area since thieves may perceive that travelers carry valuable items and are less cautious and hence more vul nerable when vacationing. There is a cost associated to each of these negative effects of tourism but the exceptional growth of the industry suggests that the benefits far outweigh the costs Airline Transportation and Tourism The tourism industry can be divided into five main sectors : th e attraction sector, accommodation sector, transport ation sector, travel organizer sector and destination organization sector (Vanhove 2005). The t ransportation sector plays an important role in the tourism system as it serves a s a vital link between consumer s and the tourism industry. A irlines, railways, bus and coach operators, car rental operators and shipping lines are included in the transportation sector Both the tourism industry and the transportation secto r show a high level of interdependence which is especially true when describing the airline transport ation sector Personal mobility has increased substantially worldwide over the past decades and airline travel continues to become an increasingly importa nt transp ort mode (Dargay and Hanly 2001 ). Air line transportation has contributed to such increase in mobi lity and therefore, to the economic growth of many countries around the world. Airline transport at ion is the means used by millions of individuals t o get to their touristic destination. Tourists have access to more destinations today than 10 years ago and their

PAGE 25

25 spending has improved the economy of many countries. Africa, the fastest growing destination worldwide, has experienced an expansion of the ir domestic economy and employment due to tourism. A n estimated 675,000 people directly employed in tourism are supported by overseas visitors arriving by air, representing 20 % of all tourism jobs in Africa ( WTTC 200 7 ). According to the WTTC 40 % of internat ional tourists now travel by air up from 35 % in 1990. Surveys from Visit Florida indicate that 42.6 million tourists arrived by air in Florida during 2006 accounting for 50.8 % Florida airline passenger traffic is composed by domestic airline passenger traffic and international passenger airline traffic. Both Domestic airline passenger traffic in Florida totaled 56,059,319 p eople in 2006 which accounts for 8 5 % of total U.S. domestic airline passenger traffic, as shown in Figure 1 1. Figure 1 1 S hare of U.S. domestic airline passenger traffic by state in 2006 Source: Bureau of Transportation Statistics. Int ernational airline passenger traffic coming to Florida totaled 10,499,493 passengers in 2006 which accounts for 13.7 % of t otal international airline passenger traffic arriving in the United States. 11.2 9.6 8.5 6.2 5.8 4.9 4.0 0.0 2.0 4.0 6.0 8.0 10.0 12.0 California Texas Florida Illinois Georgia New York 45 states Share (in percentage) State

PAGE 26

26 Problem Statement The tourism industry plays an importan totaled USD 32.96 billion in 2006 making the tourism industry a major source of revenue and the largest employer in the state. Moreover Florida is one of the premier tourist destinations for domestic and international visitors alike. More than one half of the domestic visitors that traveled to Florida used airline transportation ( Visit Florida 2006). Additionally, close to 10.5 million international passengers arrived in Florida using airline transportati on in 2006. In order for Florida to continue to be a top destination in the world, an understanding of the factors that affect demand is v ital A ccurate understanding of the number of passengers traveling to Florida is critical in order to achieve proper development. Measures of tourism demand could be used as a guide that will help state officials design a strategic plan to meet the long run needs of the state. A lso, a n effective allocation of resources is critical for a strategic plan that aims to d evelo p new infrastructure and renovate and expand current infrastructure (e.g., highways, airports) secure resources such as water and energy prevent crime and lessen the negative effects of noise and air pollution and transformation of land use. Research O bjectives The general objective of this study is to d evelop an understanding of the factors influencing domestic demand for airline travel to Florida In order to achieve this general objective, three specific objectives were s et and are outlined as follo ws: Develop an econometric model that helps explain th e driving forces of demand for domestic airline travel to Florida and to five combined statistical areas ( CSA s) in Florida Examine the factors that co ntribute to increasing domestic airline passenger tr affic to Florida and to five destination CSAs in F lorida

PAGE 27

27 Use the results from the econometric model ( i.e., partial adjustment model) to conduct simulations on specific shocks to the system to determine their effect on the demand for domestic airline passen ger traffic. Research Methodology Passenger d ata used in this study were collected by the Bureau of Transportation Statistics and are available online. Data present number of passengers from each state (domestic) and country (international) traveling to th eir ultimate destination, Florida. D ata also include month and year of travel which could be used to measure any seasonal patterns in the airline passenger traffic traveling to Florida. Given the size of the research project, this work was limited to esti mating an econometric model for the domestic airline passenger traffic traveling to Florida. Nevertheless, d ata related to international passengers traveling to Florida and freight and mail sent via commercial passenger airlines to Florida were analyzed an d some general descriptive statistics are illustrated in Chapter 3. Such descriptive statistics give a general overview of international passenger traffic that will allow the reader have an understanding of the size and importance of the market. Furthermor e, freight and mail are included to provide a complete overview of the industry. Theoretically one could expect that freight and mail help defray the cost of a flight incurred by airlines and it is expected to be reflected in the airfare variable included in the model. Chapter 4 present s the theoretical framework of the partial adjustment model and also the empirical application to air passenger demand traveling to Florida. The model, which includes a static and a dynamic component, is an attractive approac h since it yields a speed of adjustment coefficient that shows how fast passengers adjust to events that affect their decision to travel to Florida. An empirical model was introduced to show how domestic demand for air passenger traffic could be represente d and estimated. Economic theory was used to build the static component and a spec tral analysis was performed to identify the dynamic component of the

PAGE 28

28 model. Preliminary diagnostics were conducted to determine whether the data included in the model are st ationary (augmented Dickey Fuller test). Also, the Durbin h statistic hypothesis test was performed to identify if the errors had an au to regressive scheme of order 1 These diagnostics helped in determining whether ordinary least squares ( OLS ) estimates a re consistent or whether other estimation procedures were needed. Air line Passenger Traffic Partial Adjustment Model The partial adjustment model developed by Nerlove (1958) introduces explanatory variables with the argume nt that current demand is not onl y driven from its past values but also by consumer responses to exogenous factors such as income, crime and terrorism, advertising, and weather. This model has been used to model demand in areas such as tourism, finance, and agriculture. For example, Flann ery and Rangan (2006) specified a partial adjustment model to identify target capital structures in firms Also, Ward and James (1978) used the partial adjustment framework to determine the effect of coupons on consumption of frozen orange juice. T he part ial adjustment model attempt s to identify any factors that influence the demand for air line passenger traffic t raveling t o Florida. As described by Pollak (1970), past consumption patterns are an important determinant of present consumption patterns. Adamo wicz (1994) visitation occurs. The partial adjustment model i s used to determine whether rigidity exists in the demand by domestic air passengers traveling to Fl orida. Assume that there is a desired airline pa ssenger traffic demand and it is a function of airline ticket prices, personal disposable income, gross d omestic product by U.S. region and dummies for hurricanes wildfires and terror. If the partial adjust ment parameter ( ) equals zero there is no partial adjustment and demand for airline passenger traffic is purely determined by its past values. If equals one, demand for airline

PAGE 29

29 passenger traffic is determined by some explanatory variables, and the ARIMA model is no longer relevant. It is hypothesized that the adjustment coefficient of the demand for air passenger traffic to Florida lies between zero and one and that it varies between origin (U.S. region) and destination (CSA). Thus s eparate equations were specified for domestic airline passenger demand according to its originating U.S. region: Northeast, Midwest, South, or West. There were a total of 20 equations estimated under the Seemingly Unrelated Regression ( SUR ) approach. The S UR modeling accounts f or any correlation across equations and should be the appropriate approach If any partial adjustment is present in airline passengers traveling to Florida, individuals will adjust somewhat to changes in demand drivers in the short run but the ultimate ful l effect will be realized in subsequent periods. Finally, estimates of the partial adjustment models were used to simulate different shocks such as hurricanes and wildfires and changes in prices and income These simulations are helpful when making recomm endations for the tourist industry in an attempt to increase overall demand for tourism in Florida. Data and Scope Passenger d ata used in this study were collected by the Bureau of Transportation Statistics and are available online. Nearly 4.3 million obse rvations were collected from the database T 100 Domestic 3 Market and T 100 International Market between January 1990 and December 2006. A market is defined by the first departure airport on a ticket and the ultimate arrival airport. The T 100 Domestic Mark et database included all passengers enplaned at a U.S. origin airport and deplaned at the ultimate U.S. destination airport. It also provides monthly statistics on freight and 3 Data include 50 States, District of Columbia, Puerto Rico, and other U.S. t erritories and p ossessions.

PAGE 30

30 mail specified by their origin airport, city and state, carrier (airline) identi fication number, destination airport, and miles between airports. Since the scope of the study only covers Florida, all observations reporting Florida as destination state were then selected. Due to the high volume of data collected, aggregation was perfo rmed in order to make the data more manageable. Such aggregation was made at two levels: at the origin grouping states into U.S. regions and at the destination grouping cities into 2 2 CSAs, both following U.S. Census guidelines. The T 100 International Ma rket database includes all passengers enplaned at an international origin airport and deplaned at the ultimate U.S. destination airport and all passengers enplaned at a U.S. origin airport and deplaned at the ultimate international destination airport. It also provides monthly statistics on freight and mail specified by their origin airport, city and country, carrier (airline) identification number, destination airport, city and country and miles between airports. Again, all observations reporting Florida as ultimate destination state were selected. I nternational passenger data underwent a level of aggregation similar to that performed on the domestic passenger data. Data on airline ticket prices for domestic flights used in this study were collected by th e Bureau of Trans portation Statistics (BTS) and are available online. The Origin and Destination Survey (DB1B) Market was used to collect one way and round trip airline ticket prices from several domestic origins traveling to Florida. According to the BTS, the DB1B Market contains (directional) origin and destination markets, which is a 10 % sample of airline tickets from reporting carriers. It includes items such as passengers, air fares, and distances for each directional market, as well as information abou t whether the market was domestic or international. Nearly 46.8 million observations reported o n a quarterly basis were retrieved from

PAGE 31

31 the DB1B Market database between Quarter I, 1993 and Quarter IV, 2006. Description, selection, and aggregation of airline ticket prices are detailed in Chapter 3. Additionally, the following data were collected from various sources and included in this study: annual gross domestic product by state and per capita disposable income by state monthly statistics on number of wi ldfires including total of acres burned a cross F lorida hurricane data, average temperatures and rainfall from selected cities, brand advertising expenditures from selected private firms, generic advertising expenditures, monthly average kerosene type jet fuel prices annual population estimates by state, and annual c rime statistics from Florida Th is study focused on the specification of a partial adjustment model for domestic airline passenger traffic traveling to Florida between January 199 5 and Decem ber 2006 only. Passenger d ata include all passengers traveling to Florida without specifying the reason of travel. These passengers could be residents coming back to Florida, individuals on business trips or tourists. Therefore, all inferences are restricted to total passengers rather than tourists.

PAGE 32

32 CHAPTER 2 LITERATURE REVIEW Over the past few decades, demand for tourism has become a focal point in several empirical studies. Many industries depend on accurate forecasts of tourism demand. Unlike manufacturing products that can be stored, tourism products such as airline seats and hotel rooms cannot be stockpiled (Archer 1987). Any under used tourism product has essentially no value since it does not have alternative uses. Airlines, tour operators, and hotel ch ains incorporate t hese forecasts into their every day operational decision making, risk manag ement, and strategic planning. As Witt, Song and Louvieris (2003) state, failure to predict major downturns or upswings in tourism demand could have serious financ ial consequences. Forecasting Tourism Demand Forecasting literature regarding tourism demand is to a large extent dominated by a relatively limited number of writers: Archer, Frecht ling, Smeral, and Witt (Vanhove 2005). Moreover, international tourism dema nd at a national level (due to data availability) dominates the scope of most studies and only a few have focused on domestic demand (e.g ., Witt, Newbould, and Watkins 1992). In this study, the se authors compared three models (exponential smoothing, nave I, and nave II) to generate forecasts of visitor arrivals in Las Vegas, Nevada. M ean absolute percentage error and root mean square percent age error were used to measure forecasting accuracy of the three models. Also, data used in the study include monthl y visitor arrivals to Las Vegas from 1973 to 1986 reported by the Marketing Bulletin and the 1990 Ten Year Summary. R esults suggest that real improvements in forecast accuracy can be achieved using the exponential smoothing technique, given the nature of t heir dataset. Both t he exponential smoothing and nave II models outperformed the nave I model in terms of mean absolute percentage error. Also, the se authors conclude that domestic tourism demand tends to

PAGE 33

33 be less volatile than international tourism deman d, since external influences such as exchange rate fluctuations and international political events have much less impact on domestic tourism than international tourism. Moreover, a weak U.S. dollar and political instability in foreign countries may increas e domestic demand because consumers would rather choose to travel to cheaper and safer domestic destinations. They agree that the methods used to forecast international demand may not be appropriate to estimate domestic demand. Hence, further research is n eeded to determine the relative accuracy of various forecasting techniques applied to estimate domestic tourism demand. Studies on international tourism demand are found more frequently in the literature, with an increased interest in developing countries. As mention ed in Chapter 1, tourism has become a vital source of income and a major contributor in the balance of payment s for developing nations. Countries in Africa, Asia, and Latin America have greatly benefited from flow s of visitors t raveling to their sites. In addition researchers have used econometric forecasting models to gain some insights in to the forces driving tourism demand in these countries and also to estimat e future flows of visitors to these destinations The objective of these studies h as been to forecast flows of tourists in order to plan for necessary investments in infrastructure and labor and to determine the environmental and social impact in their communities. Empirical studies have been conducted in Macau (Song and Wi t t 2006), Ind onesia and Mal aysia (Tan, McCahon, and Miller 2002), Laos (Phakdisoth and K im 2007), Latin America ( Eugenio Martn, Martn Morales, and Scarpa 2004), and Africa (Naud and A. Saayman 2004). Studies have also shown special interest in analyzing air travel f lows to Latin America, Southeast Asia, and Africa since most of the visitors use this mode of transport to

PAGE 34

34 travel to these destinations. For instance, air travel in Europe competes with othe r modes of transportation ( i.e., car and train) because distances between origin s and destinations are shorter. Interestingly, their conclusions have notably varied from those made when analyzing tourism demand in developed countries. Several studies have shown that the primary demand driver of tourism in developed count ries is income, whereas political and social stability and accessibility drive demand for tourism in developing countries. Phakdisoth and Kim (2007) present a static and a dynamic econometric model that attempts to identify key determinants of inbound tou rism in Laos. Their model includes both demand and supply factors in a single equation model. Demand drivers such as gross national income from countries that had bilateral trade with Laos ; Laos GDP, exports, and imports ; cost of living in Laos ; and distan ce between origin country and Laos are included in the analysis. In addition, road infrastructure at Laos, communication infrastructure at Laos, and the rule of law indicator published by the World Bank, were included as supply factors in the model. The ra ndom effects and fixed effects model and ordinary least squares ( OLS ) were used as static model specifications, while the partial adjustment model was used to specify the dynamic model. The se authors con clude that supply factors ( i.e., roads, communication and stability in Laos) In developed countries, concern s that tourism is no longer a luxury good ha ve prompted more empirical studies to determine the effect of this trend Smeral and Weber (2000) su ggests that there is an expectation that tourism will gradually lose the character of a luxury good and that at least in some segments there will be a trend towards saturation. As a result, long term growth in tourism expenditures will level off. Proena a nd Soukiazis (2005) conducted a study that analyzed tourism demand in Portugal. They concluded that income is the most important

PAGE 35

35 determinant of demand in their static model. The se authors also specify a dynamic model which identifies accommodation capacity as the most imp ortant supply determinant in tourism demand in Portugal. Similarly to the previous study conducted in Laos, this paper specifies a random effects and fixed effects model for the static model and a partial adjustment model for the dynamic mo del. This study is also limited to the country level and include s four countries rigidity exists in tourism inflows. It is unclear though whether this is relat ed to the maturation of the markets. Nevertheless, the authors conclude that Portugal must develop new policies to reduce the dependence from the U nited K ingdom Spain, Germany and France an d explore alternative markets. Scope of Recent Tourism Demand Stu dies Several studies with a domestic scope explore other issues related to air transportation but that could affect tourism demand. For example, some studies focused on determining the impact o n optimal ai rport capacity (Zhang and Zhang 2006), analyzing productivity and efficiency of hub and spoke models ( Brueckner 2004), and estimating price, income, and distance elasticities of air transportation demand (Bhadra and Kee 2008). These studies, although not related directly to touris m demand, provide valuable insight since any policies assessed to the transportation industry may have an effect o n the tourism industry. Recently, other concerns such as environmental issues (e.g., noise and air pollution), health scares (e.g., influenza SARS) and security incidents ( e.g., 9 11 terrorist attacks) have been subject to discussion and analysis related to air transportation and tourism. L u and Morrell (2001) analyze d the implications of an environment charge mechanism assessed to airlines, s pecifically its impact on airline costs, airfares, and passenger demand. Since externalities such as aircraft noise and engine emissions generate profound impacts on human beings and on the

PAGE 36

36 environment, environmental charges are one of the economic instrum ents used to control them. Government agencies have assessed these fees to commercial flights and proceeds are used to improve the environment at airports and surrounding communities. The study concluded that there is a need for an assessment of environmen tal charge mechanisms to encourage sustainable development. Brons et al. (2002) state, however, that if airlines can charge all extra costs to passengers without decreasing demand, the environment policy has no other effect than enues Grais, Ellis, and Glass (2003) conduct ed a simulation analysis in which they explore d potential ramifications of the influenza pandemic strain of 1968 1969 as if it had r esurfaced in Hong Kong in 2000. They suggest ed that since air travel ha d increased since the late 1960s, the disease could be spread faster (111 days earlier than it had been forecast ed in 1968 ) and to more people (176% greater than the 1958 pandemic). Moreover travelers sacrifice not only cash co sts but also the opportunity of using the time in an alternative activity (i.e., opportunity cost of travel). Studies by Blunk, Clark, and McGibany (2006) and Ito and Lee (2005) have evaluated the economic impact of the 9 11 terrorist attacks o n New York C ity, Washington D.C., and Pennsylvania. Both studies agreed that the stri cter security requirements implemented as a result of the terrorist attack have increased the opportunity cost of travel. Blunk, Clark, and McGibany (2006) used a modal choice model applied to communal behavior to estimate air travel demand in terms of revenue passenger miles. They tested the hypothesis that anything that permanently raises the opportunity cost of travel, holding benef its constant, should reduce travel volume The ir s tudy concluded that the 9 11 terrorist attacks

PAGE 37

37 increased the opportunity cost of travel and that the detrimental impacts of the attacks were not temporary. In their study, Ito and Lee (2005) attempt ed to separate the effect of the 9 11 terrorist attacks in to temporary and ongoing components. Results suggest ed that negative impacts on both quantity and price indicate that the 9 11 terrorist attacks resulted in a negative demand shift, rather than a supply contraction. Such negative demand shift was estimated to be at 30 % in its transitory stage in addition to a 7.4 % ongoing negative demand shift. They also agree d that such a catastrophic event could require a long recovery process. Finally, they state that technological innovations in security screening might eliminate some of the waiting time in airports, reducing the hassle factor and making air travel more convenient, and hence, increase demand for air travel. Hodges and Mulkey (2001) also assess ed the potential impact of the 9 11 terrorist attacks on Flori ed a model using IMPLAN economic impact modeling system to reflect linkages between industries, employees, institutions, and consumers. The objective was to simulate both direct and indirect impacts triggered by the 9 11 terror ist attacks. Results suggest ed that if there is a 10 % decrease in overall tourism expenditures, the economic impact in Florida would increase to USD 11.7 billion, including indirect effects ( USD 1.3 billion) and induced effects ( USD 3.54 billion). The se au thors conclude d that the total economic impact s o n tourism are substantially greater than the direct value of visitor spending, due to the d that a modest change in vi sitor sp ending can dramatically affect state government fiscal balances, which is largely funded by sales tax revenues.

PAGE 38

38 Econometric Techniques Applied to Tourism Demand Analysis Several techniques have been used to model and forecast tourism demand. Econom etric li terature presents techniques that range from pure qualitative studies such as surveys to rigorous quantitative analyses (e.g., multiple regression models). Qualitative studies include the use of surveys, Delphi model s and, to a lesser extent, judgm ent aided models. Within the quantitativ e approach, studies can be divided into trip generation models, gravity models, time series methods, and regression analysis. The following discussion will focus on quantitative techniques used to estimate tourism de mand. Trip generation models are the stepping stone of a four step transportation forecasting process followed by trip distribution mode choice, and route assignment T rip generation model s ha ve been used to estimate the number of trips that each area wi ll generate and attract. There are two kinds of trip generation models: production models and attraction models. Trip production models estimate the number of home based trips to and from areas where trip makers reside. Trip attraction models estimate the number of home based trips to and from each area at the non home end of the trip. Trip production models are commonly used to estimate truck and taxi trips, whereas trip attraction models are used to estimate tourism demand. Gravity models are based on New with the idea that as the size of one or both areas increases, movement between them will also increase. These models have been used to predict the degree of interaction between two places. These studies attempt to determine at which p oint consumers will prefer to travel to one area instead of the other. Explanatory variables most frequently used include distance, time, and expenses. Applications of gravity model s have extended to international trade where it is used to estimate bilater al trade flows between two areas. The model is based on the economic size of each area ( e.g., gross domestic product ) and the d istance between them. Since gravity models rely on distance as its primary

PAGE 39

39 explanatory variable, they have been subject to critic ism for their lack of theoretical foundation Sheldon and Var (1985) argue that gravity models can only be used to forecast number of tourists, but not expenditures, occupancy rates, and other important variables. They also add that origin zones are diffic ult to identify and that measures of travel time and cost are neither constant nor accurate, since they can vary by means of travel or season. Bhadra and Kee (2008) specify a gravity model to estimate airline travel demand in which they express the explan atory variables in their log linear form Such variables include average one way airfare between origin and destination, personal income and population at origin and destination, and distance between origin and destination. Demand is expressed in number of passengers between 192 metro areas in the United States. M etro areas are grouped into four categories based on the level of traffic : super thin, thin, thick, and super thick markets Multicollinearity was mentioned as a potential problem becaus e large pop ulations tend to be associated with higher levels of economic activities. A formal test was conducted and standard errors were small and stable across sub samples, but condition indices appear to be larger than usually suggested. Results suggest that passe nger flows between origin and destination markets exhibit growth in recent years. In addition demand is elastic with respect to airfares in thick markets, supporting the hypothesis that there is high competition in these markets. Passenger flows in semi t hick and super thin markets were found to be distance inelastic. Several studies on international tourism demand focused only on forecasting tourism demand using time series models where the sole objective is to predict future flows of tourists to a parti cular country. Time series models, or non causal models, assume that any variable can be forecasted without taking into account any factor that influenced the level of the variable. Exponential smoothing ; sine wave regression ; and the Box Jenkins family of time series models:

PAGE 40

40 auto regressive (AR), auto regressive moving average (ARMA), auto regressive integrated moving average (ARIMA), and seasonal auto regressive integrated moving average (SARIMA) are some models specified to forecast tourism demand. Some of these models have superior predicting ability ( e.g., Box Jenkins method) and are now widely used to forecast tourism demand. In fact, the Box Jenkins approach is widely applied in the airline industry and has been labeled as the airline model Time se ries models have been widely used to predict tourism demand because they tend to perfor m better than regression models. However, a few empirical studies have concluded the opposite. Several time series models and adaptations can be found in the literature. Most of the focus has been to compare several time series models to determine which model generates the lowest error magnitudes. C ommon performance measures used in these studies include mean average percentage error (MAPE), root mean square error (MRSE), and mean absolute error (MAE). Oh and Morzuch (2005) present a comparison of eight time series models. Flight data from a t Singapore were used in the study. The ir study splits the da ta into two groups: a within sample period from July 1977 to December 1988 and an out of sample period from January 1989 to July 1990. Each model was used to generate data for a 3 month, 15 month and a 19 month horizon. Dickey Fuller tests were performed t o test for stationarity. The se authors then chose six performance criteria: one to determine whether the estimate is unbiased and the remaining five, MAPE, MRSE, MAE, ss due to forecast error. Results suggest that in the three month horizon the ARIMA (3,1,0) specification ranked first on four of the five performance statistics when applied to the post

PAGE 41

41 sample data, but its accuracy decreases as the forecast horizon incre ases. Also, within sample may not show any structural changes during t he post sample period and hence provide p oor forecasts. This study conclude d that ARIMA models provide a reliable and consistent forecasting performance across different time horizons, w ith the caveat that such models shed little light on issues related to policy. As mentioned earlier, not all empirical studies have agreed that time series models perform better than econometric models. Fritz, Brandon and Xander (1984) present ed an econom etric model aimed to forecast air arrivals into Florida from domestic points of departure. Quarterly data w ere collected from the State of Florida reported from 1960 to the fourth quarter of 1980. The model includes variables describing economic characteri stics such as disposable personal income and the composite of leading and coincident indicators published by the Bureau of Economic Analysis. In addition the se authors explain that all variables were lagged three quarters behind to provide more timely for ecasts. The y used the Box Jenkins method as a pure time series model, an econometric model, and a combination of both. Results suggest ed that econometric forecasts were considerably more accurate than the Box Jenkins but that there is an improvement in acc uracy available with the combination of forecasts. Also, forecasting errors increase d as the time horizon increase d The study conclude d that forecasting accuracy provides important benefits to state and local government since their revenues derive from to urist expenditures. Determinants of Tourism Demand Understanding causal relationships between variables proved to be vital in the design of policy and planning strategies. Since pure time series models lack explanatory power, they are used exclusively to f orecast demand with no regard into determine which forces drive such demand for tourism. Regression analysis, on the other hand tries to fill the gap left by time series

PAGE 42

42 models. Regression analysis ( i.e., causal methods, econometric models) attempts to ex plain the behavior of the dependent variable due to some ex planatory variables, as well as to forecast future levels of demand. Under regression analysis, the literature offers several methods used to model tourism demand. Single linear regression methods use one explanatory variable only and are the simplest v ersion found in the literature. M ultiple regression method s which include more than one explanatory variable in a single equation ha ve also bee n used to model tourism demand. Mo st frequen tly used exp lanatory variables include income, commercial ties between countries, price, price substitutes, transportation cost, distance, travel time, exchange rate, promotions efforts, population growth, supply capacities, business cycles, trend factors, qualitative factors, dummy variables for special events, natural disasters, war, terrorism, and lagg ed dependent variables (Vanhove 2005). Air travel demand shares some of these determinants with tourism demand. Nielsen (2001) classifies the determinants of air trav el growth in two categories: drivers and impeders. He identifies increased personal incomes with reduced real airfares as the major drivers of air travel demand. Building up the aviation socio technological system, technological innovation, globalization, competition, incentives to the industry, economic growth policies, population growth, and advertising are also mention ed as drivers of air travel. Impeders of air travel growth include financial and time constraints, airport congestion, environmental polic ies, and alternative lifestyles He conclude d that most air travel relates to globalization, changing geography, and population dynamics. Air travel ha s benefited from increased commercial relations between countries. Yet environmental pressures are arisi ng as countries increase their trade of products, services, and

PAGE 43

43 human and financial capital and pose an important threat to air travel growth. Figure 2 1 presents the physical, social, and political determinants of air travel growth as defined by Nielsen ( 2001) Market Forces Liberalization of markets Competition between airlines Competition between travel agencies Marketing strategies Impeders Drivers Economic Factors Reduced air fares Increased income Economic growth Aircraft Technology Longer ran ge Improved capacity Increased speed Reduced operation costs Political Factors Maintain growth Maintain employment in aviation socio technical system Competition among nations Subsidies Infrastructure Enlarged airport capacity Reduced rail capacity Psychological Factors Individual needs, wants, and desires Geography changing Population growth Migration Internationalization of family structures Globalization Globalization of market Globalization of companies Globalization of political system More international relations Free time availability Work structures Holiday structures Rich and ag ing population People taking a year off Social Factors Social norms and values Air travel as status maker Trends: further away, deeper into the forest, highe r up in the mountains Experience other cultures Traveling cultures Figure 2 1 Determinants of air travel growth. Source: Nielsen 2001. L og linear specification s ha ve been the primary function al form used in most analyses. It has been shown that the log linear specification tends to model the demand data better and it also conveniently provides demand elasticities since the estimated parameters represent the elasticities of the variables. These demand elasticities are then used to formulate policies and examine how consumers respond to changes in demand variables (Uysal and El Roubi 1999). Air Travel Alternat ive Society Regulate market forces Impede globalization Economic satiation Environmental Policies Jet fuel tax Emission quotas Greenhouse gas budgets Planning Stop enlarging airport capacity Improve rail capacity Alternative Lifestyles People choosing lower income and more free time and closer destinations

PAGE 44

44 E lasticities provide valuable information by quantifying how much the demand for a product or service chang es with a change in its determinants or drivers. For example, elasticities can show how much a chang e in air ticket prices affects demand for air travel to a specific area. Some authors have challenged the forecasting accuracy of econometric models and hav e compared them to time series models. Most of them agreed that time series methods are more accurate that ec onometric models. Nonetheless, authors have also stated that econometric models play an important role in policy implementation and strategic plann ing. There is a need to understand the determinants of demand and to assess the impact of any changes in these determinants on tourism demand. I t is also true however that there is a clear tradeoff between forecasting accuracy and identification of causa l relationships. The use of one model over the o th er will depend on whether the interest is pure forecasts or identification of the determinants of demand. Martin and Witt (1989) evaluated seven quantitative forecasting methods, six time series methods an d one econometric forecasting method. Th is analysis used annual tourism flows data at a country to country level ( e.g., U nited K ingdom to Austria). They concluded that, although its accuracy was poor compared to time series models, an econometric forecasti ng system does allow exploration of the consequences of alternative future policies on tourism demand. Econometric forecasting models, unlike time series models, provide helpful insight due to their ability to identify and quantify the determining forces d riving demand. Such insight represents a great value at several time horizons. Witt and Witt (1995) argue that short term forecasts are required for schedu ling and staffing needs, medium term forecasts for planning t our operator brochures, and long term fo recasts for investment in aircraft and infrastructure.

PAGE 45

45 Witt, Song, and Louvieris (2003) reviewed past demand forecasting studies and compared the performance of six econometric models and two univariate time series models on international tourism demand i n Denmark. The se authors evaluate d the f orecasting performance of four econometrics and two time series models, their ability to forecast direction of change, and the unbiasedness of their estimates. E conometric models used in this study include a static m odel ; two error correction models: Wickens and Breusch procedure (WB) and Johansen maximum likelihood method (JM); a reduced autoregressive distributed lag model (ADLM) ; a tim e varying parameter model (TVP); and an unrestricted vector autoregressive model (VAR). The two univariate time series models were an ARIMA model based on the Box Jenkins procedure and a simple nave or no change model. Results suggest ed that for a one year horizon, the TVP and the reduced ADLM generate d the most accurate, unbiased fo recasts when accuracy was defined in terms of error magnitude. When forecasting unbiasedness is taken into account in conjunction with directional change error, the static and TVP models outperform ed the other four models in forecasting one year ahead. Mor eover, econometric models outperform ed the nave model in terms of generating accurate forecasts of directional change, which suggest ed the importance of taking structural instability into consideration when generating short term forecasts. The authors con clude that, even though, econometric models require larger data, considerable user understanding and more expertise than univariate time series models, tourism researchers should give serious consideration to using them to generate forecasts of internatio nal tourism demand. Static versus Dynamic Regression Models Other authors have combined both time series and econometric modeling into one general pooled model. The study by Young (1972) compared three models: a time series m odel with a partial adjustment specification, a cross

PAGE 46

46 model incorporates both temporal and contemporaneous interdependences in order to establish relationships between macro time series parameters and micro cross section parameters I ndependent variable s included in all three models we re permanent income, ticket price, and journey time by the airplane. Results suggest ed that better results can be obtained by pooling time series and cross section data. In addition Young was able to report short run and long run elasticities of price, time, and income and concluded that short run elasticities were not very elast ic while long run elasticities we re highly elastic. These conclusions are validated by economic theory that shows that consum ers and firms are able to adjust in the long run. Under regression analysis, the literature also presents static and dynamic models that attempt to model tourism demand. A particular challenge arises when using a static model, also known as long term model to estimate tourism demand. Consumers tend to adjust better to price and security signals in the long run than to sudden changes in costs or unforeseen security incidents in the short run. This is particularly true in the tourism industry since consumers typically book their flight and hotel reservations in advance limiting their ability to adapt instantaneously. Therefore, dynamic models have become popular and are considered to estimate demand more accurately becaus e the se models account for rigidities that could occur as consumers adjust to changes in their demand drivers, especially in the short run. D ifferences between static and dynamic models are fully explained in Chapter 4. F orecast of tourism demand to Portugal using dynamic modeling was the main focus on an empirical study by De Mello and Fortuna (2005). The se authors argue d that habit persistence, adjustment costs, or imperfect information prevent consumers from automatically fully adjust ing every period. Therefore, an explicit dynamic structure is required to explain demand behavior and to account for the short run adjustment process. They develop ed several dynamic stochastic

PAGE 47

47 specifications including a partial adjustment model and a dynamic almost ideal demand system ( D AIDS) model. The study con cluded that the partial adjustment model wa s consistent with the postulates of economic theory and provide d robust and empirically plausible estimates. Challenges in Tourism Demand Analysis Other studies have focused on the difficulties of collecting rele vant data that could be used to assess the real economic contribution of tourism related activities to a particular area. P roblem s arise because the national accounts classification system does not identify tourism as a separate component of the GDP. The t ourism satellite account (TSA) program aims to fill that According to the Office of Travel and Tourism Industries web site, these data do not exist from any federal government agency o r private sector source. A prototype account is being established by the Bureau of Economic Analysis in conjunction with the Tourism Industries office which will incorporate as much data as is available to the "core" account consisting of the traditional i ndustries for tourism. Jones, Mund ay, and Roberts (2003) explore methodological difficulties in the construction of a TSA at a regional level and the implications of deriving an effective tourism policy. They mention that data from tourism surveys are sel dom detailed enough to account for incidental purchases made by tourists. Moreover, the oversimplification of the labor account presents a problem in the construction of the TSA. U se of labor is highly uneven across space and time, and even with in a region Therefore, these accounts do not illustrate the fact that the labor market in the tourism industry is high ly seasonal. They agree that the disaggregation of existing input output data to account for technical coefficients, local sourcing patterns, and la bor use is needed. T hey also realize however, that this process may be cost prohibited and that there is no real evidence that the construction of the TSA is worth the effort.

PAGE 48

48 Despite the overlapping worlds of tourism and transport, there has only been l imited progress evidenced in the literature (Lumsdon and Page 2004). Many of the shortcomings in research have stemmed from the fact that transport and tourism data are inconsistent. Key elements such as tourist or passenger, trip or tour are not clear ly d efined. Currently, efforts are being made to standardize survey designs, procedures, and sampling. Also, a common challenge faced by most studies has been to determine which dependent variable to use in order to accurately estimate tourism demand. For exam p le, there is no apparent consensus among researchers on which dependent variable is the most suitable to measure tourism demand. Although total tourist arrivals for specific regions or countries ha ve been widely used real expenditures on tourism goods an d services have also served as a measure of tourism demand. The complexity of the industry also hampers the ability to disaggregate data to estimate pure effects on demand. In addition more challenges arise because crises and disasters, which affect touri sm greatly when they occur, are impossible to forecast. Researchers face constraints in terms of accurate and relevant data that are difficult to collect for some variables (Sheldon and Var 1985). However, they have used several techniques and measures to improve the estimation of tourism demand with the information available. Multicollinearity has been mentioned as a potential problem in both time series and cross section estimates. For example, gravity models have encountered high correlation between air f ares and distance. This problem intensifies in time series models given that price and income tend to be strongly associated with a time trend. Also, within the econometrics models, multicollinearity is often present among explanatory variables, especially in income and cost of travel variable s when data are cross sectional. Another limitation found in almost all models is

PAGE 49

49 the forecast horizon. These models m ay only be applicable to short t erm analysis, since estimates change with time. Forecasting models t end to perform better on short term horizons, while forecasting error increases as the time horizon increases. Lagged dependent variables create a problem in dynamic models. Studies conducted by Phakdisoth and Kim (2007) and Proena and Soukiazis (2005) e ncountered statistical problems since there was a correlation between the dependent variable and the error term arising from the lagged dependent variable used as an explanatory variable in the partial adjustment model. Estimates are biased and inconsisten t under ordinary least squares estimation (OLS) or general least squares estimation (GLS), especially in smaller samples. Instrumental variables (IV) estimation techniques were used to solve such problem. The idea behind IV is to identify instruments that are highly correlated with the endogenous regressor but uncorrelated with the error term. Researchers have also dealt with the phenomenon of seasonality. Seasonality in tourism demand has been well documented in the literature and has been regarded as a p roblem. Seasonality can be identified as fluctuations of demand at tourist destinations. It affects tourism suppliers, tourism employees, residents, and tourists alike In the off season, defined as low flow of tourists to a particular area, suppliers stru ggle to fill capacity (e.g., hotel rooms, aircraft seats, attraction tickets), employees have no access to sustainable long term employment, tourists encounter limited services (although can benefit due to discounts during this season), and residents who o wn businesses in the area experience a sales decrease Lundtorp (2001) discusses four main reasons for the existence of seasonality: 1 ) weather (variation in temperature, rainfall, snowfall, and daylight) ; 2 ) institutional patterns (school holidays, indust rial holidays, and

PAGE 50

50 calendar holidays) ; 3 ) destination characteristics (winter versus summer resorts) ; and 4 ) marketing efforts (special events, conferences). A ccurate identification of seasonality and its drivers reduce uncertainty and therefore, risks in tourism operations. It also could lead to a successful marketing strategy that will aim to increase flow of tourists during the off season. It helps the operations of t ourism firms to plan accordingly. For example, it is not uncommon to find some hotels a t summer holiday destinations (main season) closed during winter (off season). Kulendran and Wong (2005) present a time series model aimed to measure seasonal variations in holiday, visit to friends and relatives (VFR), business, and total visits. Conventi onal unit root tests were conducted to determine the nature of the seasonality and order of integration. Two competing models were considered: an ARIMA model with first differences (ARIMA1) and an ARIMA model with first and fourth differences (ARIMA14). Th e se two models were fitted to the inbound tourism quarterly data series of eight European countries and the United States between the period of 1978 and 1998. Results suggest ed that seasonal variation varies according to the type of visit. Holiday tourism was found to have the highest R 2 which implies that it has strong seasonal variation, followed by VFR and business visits In addition the forecasting accuracy of the two models was measured using the MAPE. The se authors concluded that the forecasting pe rformance of ARIMA1 and ARIMA14 depends on the nature of the seasonal variation in tourist arrivals time series and that correct identification of seasonal variation can improve the forecasting performance of the ARIMA model. The ARIMA1 provides better for ecasts for time series with less seasonal variation and ARIMA14 provides better forecasts for time series with strong seasonal variation.

PAGE 51

51 Recent Developments in Tourism Demand Analysis Recent developments in econometric modeling and forecasting show that n ew and more advanced models are being used in the tourism demand literature. Some of these models include time varying parameter (TVP) estimation, almost ideal demand system (AIDS) models, the error correction model (ECM), vector autoregressive (VAR) appro ach, and time series augmented with explanatory variables such as structural time series models (STSM). Li and Song (2006) describe some of these models. The first approach is related to the TVP estimation approach. The TVP single equation models have gain ed popularity du e to their flexibility. Elasticities derived from log linear regressions are constant through time, a condition quite restrictive and that often leads to failure of the dynamic analysis of tourism demand (Li, Song, and Witt 2005). Parameter s in the TVP single equation model vary over time and allow the model to identify any short run changes in consumer preferences. The second approach found in the literature is related to the AIDS models with some refined specifications such as static linea r AIDS (LAIDS) and dynamic LAIDS. D ynamic LAIDS models are somewhat similar to the partial adjustment model in that they account for cons umer behavior in the short run. Overall, AIDS models allow testing for other demand theories such as symmetry and addin g up hypotheses. It also has a flexible functional form and does not impose a priori restrictions on elasticities, satisfies the axioms of consumers without invoking to Engel curves, and largely avoids the need for non linear estimation. The authors sugges t that, even though the Rotterdam and translog models hold some of these features, only the AIDS model contains them simultaneously and hence, it is more suitable for tourism demand analysis. The third approach relates to a combin ation of t he flexibility o f TVP models and the multiple equation specification from the LAIDS models. The authors also indicate that these models have only been applied to annual data and have not model ed seasonality. They conclude that future research should focus

PAGE 52

52 on seasonal patt erns of tourism demand and that such seasonal components need to be incorporated to these models. Moreover, Li, So ng and Witt (2005) stated that investigation of the forecasting performance of advanced econometric models in dealing with seasonality in tour ism demand and comparison of the abilities of alternative models to forecast tourism demand changes are areas of interest and directions for further research. Chapter Summary This chapter presented a summary of the literature related to tourism demand an alysis. Most of tourism demand analyses have focused on forecasting rather than identifying the determinants of demand. Such studies have concluded that time series analysis provides more accurate forecasts of tourism demand. Nevertheless, researchers agre e that econometric models are vital for policy and strategic planning and risk management. However, recent approaches aim to combine time series with causal models to improve accuracy of the estimates. The complex interaction between the tourism and trans portation industries has hindered the ability to accurately estimate tourism demand. Data availability and the lack of a standardized system of accounting between both industries have also been mentioned as a limitation. Most of the studies use annual data and do not model seasonality, an intrinsic characteristic of tourism demand. This study develops a model that attempts to fill the gap by addressing those issues and hence, contribute to the tourism research literature. A partial adjustment model is devel oped in order to identify the demand drivers of air travel to Florida and its relationship to tourism. The model incorporates seasonal components as well as economic, social, and weather factors related to air travel demand. Chapter 4 presents a detailed d erivation of the partial adjustment model.

PAGE 53

53 CHAPTER 3 DESCRIPTIVE STATISTI CS ON THE AIRLINE PA SSENGER TRAFFIC TRAV ELING TO FLORIDA AND OTHER IN DICATORS This chapter consists of descriptive statistics for the passenger, freight, and mail traffic coming to Florida by means of airline transportation, as well as economic and social indicators related to four U.S. origins and Florida. It also includes weather statistics from Florida, such as hurricanes, wildfires and average temperatures and rainfall. C hap ter 3 is organized in to four major sections, with a number of subsections in each section. The four major sections are defined as a) airline passenger traffic, b ) airline ticket prices, c ) freight and mail transport ed by commercial passenger airlines and d ) economic, social, and weather indicators. The first section, airline passenger traffic, contains five subsections: a ) d escription of the collection and selection of airline passenger traffic data; b ) t otal domestic airline traffic in terms of number of passengers coming to Florida by U .S. region from 1990 to 2006; c ) t otal domestic airline traffic in terms of number of passengers coming to five Combined Statistical Areas ( CSAs ) in Florida from 1990 to 2006; d ) t otal international airline passenger traff ic traveling to Florida by world region from 1990 to 2006; and e ) t otal international airline passenger traffic traveling to three destination CSAs in Florida from 1990 to 2006. The second section, airline ticket prices, contains six subsections: a ) d escri ption of the collection and selection of domestic airline ticket price data; b ) o ne way and round trip airline prices of domestic flights coming to Florida by U .S. region, from 1993 to 2006; c ) o ne way and round trip airline prices of domestic flights trav elin g to five destination CSA s in Florida by U.S. region, from 1993 to 2006; d ) d escription of the collection and selection of the international airline ticket price data; e ) round trip airline prices of international flights coming to Florida, by world re gion from 199 5 to 2006; and f ) round trip airline prices of international flights coming to three destination CSAs in Florida, by world region from 199 5 to 2006.

PAGE 54

54 The third section includes freight and mail transported by commercial passenger airlines An analysis by U.S. region and by destination CSA is performed to domestic freight and mail traffic transported to Florida. An analysis by world region and by destination CSA is performed to international freight and mail traffic transported to Florida. The f ourth section, economic, weather and social indicato rs, contains five subsections: a ) GDP by state ; b) b rand and generic advertising expenditures ; c) f oreign exchange rate: Euro to U.S. dollar; d) historic oil prices; e ) h urricanes and wildfires af fecting Florida; f) a verage temperatures in origin U.S. regions and destination CSAs including rainfall in each of the f ive destination CSAs; and g ) c rime rates in Florida. Airline Passenger Traffic Description, Selection, and Aggregation of Domestic Airline Pass enger Traffic Data Passenger d ata used in this study were collected by the Bureau of Transportation Statistics and are available online. Nearly 4.3 million observations were retrieved from the database T 100 Domestic Market between January 1990 and May 200 7 and T 100 International Market between January 1990 and February 2007. A market is defined by the first departure airport on a ticket and the ultimate arrival airport. The T 100 Domestic Market database includes all passengers enplaned at a U.S. origin a irport and deplaned at the ulti mate U.S. destination airport. For example, a flight from Rochester, New York to Miami, Florida with a stop in Atlanta, Georgia identifies the origin state as New York and the ultimate destination state as Florida. Also, t he data set provides monthly statistics on freight and mail specified by their origin airport, city and state, carrier (airline) identification number, destination airport, city and state, and miles between airports. Since the scope of the study only covers F lorida, all observations reporting Flori da as destination state were selected.

PAGE 55

55 Due to the high volume of data collected, aggregation was performed in order to make the data more manageable. Such aggregation was made at two levels: at the origin and destin ation. At the origin level, origin states were aggregated according to the U.S. Census geographic regions scheme as shown in Table 3 1. All domestic statistics from the 50 states and D.C. were aggregated into four different geographical originating U.S. re gions : Northeast, Midwest, South, and West. D omestic statistics for Puerto Rico, U.S. Virgin Islands, and U.S. Pacific Trust Territories and Possessions were aggregated into a category named Other Table 3 1 Geographic region scheme as defined by the United States Census. Northeast Midwest South West 1. Connecticut 2. Maine 3. Massachusetts 4. New Hampshire 5. New Jersey 6. New York 7. Pennsylvania 8. Rhode Island 9. Vermont 1. Illinois 2. Indiana 3. Iowa 4. Kansas 5. Michigan 6. Minnesota 7. Missouri 8. Nebraska 9. North Dakota 10. Ohio 11. South Dakota 12. Wisconsin 1. Alabama 2. Arkansas 3. Delaware 4. D.C. 5. Georgia 6. Kentucky 7. Louisiana 8. Maryland 9. Mississippi 10. North Carolina 11. Oklahoma 12. South Carolina 13. Tennessee 14. Texas 15. Virginia 16. West Virginia 1. Alaska 2. Arizona 3. California 4. Colorado 5. Hawaii 6. Idaho 7. Montana 8. Nevada 9. New Mex ico 10. Oregon 11. Utah 12. Washington 13. Wyoming At the destination level, all Florida airports within the same destination city were grouped. For example, all passengers, freight, and mail destined to Orlando International Airport, Kissimmee Gateway, Orlando Sanford International Airport Daytona International Airport, New Smyrna Beach Airport, and Bunnell Airport were aggregated to the Orlando CSA Then, destination cities were aggregated according to U.S. Census combined statistical areas (CSA) assigned to Florida. For example, the South Florida CSA includes the cities of Miami, Fort Lauderdale, and West Palm Beach. All Florida cities listed on the data were assigned to their

PAGE 56

56 respective U.S. Census CSA. There were 22 CSAs in Florida. The top five destination CSA s in terms of number of passengers received were South Florida, Orlando, Tampa St. Petersburg, Jacksonville, and Fort Myers. D estination CSA airports aggregated to each of the top five destination CSA s are described in Appendix A. Description, Selection, and A ggregation of International Airline Passenger Traffic Data The T 100 International Market database includes all passengers enplaned at an international origin airport and deplaned at the ultimate U.S. destination airport and all passengers enplaned at a U. S. origin airport and deplaned at the ultimate international destination airport. It also provides monthly statistics on freight and mail specified by their origin airport, city and country, carrier (airline) identification number, destination airport, cit y and country and miles between airports. Similar to the aggregation of the domestic data, only observations reporting Florida as ultimate destination state were selected. I nternational passenger data underwent a level of aggregation similar to that perfo rmed on the domestic passenger data. A ggregation of the international passenger data was made at two levels: at the origin and at the destination. At the origin level, origin countries were aggregated according to their respective origin geographical regio n and it was performed as follows. All international statistics were aggregated into four world origin regions: Canada, Latin America, Europe, and Other. Latin America region includes all countries in Central America, the Caribbean, and South America. The Other region includes all countries in Asia, Africa, Oceania and the Middle East. At the destination level, all Florida airports within the same destination city were aggregated. For example, all international passengers, freight and mail destined to Orl ando International Airport, Kissimmee Gateway Orlando Sanford International Airport Daytona International Airport, New Smyrna Beach Airport, and Bunnell Airport were aggregated to the

PAGE 57

57 Orlando CSA Then, destination cities were aggregated according to U.S Census combined statistical areas (CSA) assigned to Florida. For example, the South Florida CSA includes the cities of Miami, Fort Lauderdale and West Palm Beach. All Florida cities listed on the data were assigned to their respective U.S. Census CSA. T here were 22 CSAs in Florida. The top thre e destination CSA s in terms of number of passengers received were South Florida, Orlando, and Tampa St. Petersbu rg Domestic and International Airline Passenger Traffic Traveling to Florida Domestic airline passeng er traffic dominates the commercial airline market traveling to Florida. Approximately 82 % of the total airline passenger traffic traveling to Florida has a domestic origin, while 18 % originates abroad. More than 701.6 million domestic airline passengers, including intrastate traffic, have traveled to Florida from 1990 to 2006, with an average of approximately 41.27 million per year over the 17 year period In 2006, domestic airline passenger traffic to Florida totaled 56,059,319 people which accounted for 7.61 % of total domestic airline passenger traffic and ma de Florida the third largest domestic market behind California and Texas. Intrastate traffic defined as flights originated at a Florida airport destined to another Florida airport accounted for 8 % of total domestic airline passenger traffic traveling to Florida during the same time period. Intrastate traffic has increased 40 % from 3.11 million passengers in 1990 to 4.35 million passengers in 2006. For the purpose of this study, intrastate traffic was not included in the analysis. Therefore, here after, domestic airline passenger traffic traveling to Florida will refer only to domestic flights originated out of the state of Florida destined to Florida. Approximately 155.4 million international airline p assengers traveled to Florida from 1990 to 2006, with an annual average of approximately 9.1 million during the 17 year period. International airline passenger traffic coming to Florida totaled 10,499,493 passengers in 2006

PAGE 58

58 which represents a 61 % increase from 1990. More than 10.8 million international airline passengers traveled to Florida in 2000, accounting for the highest level during the 17 year % of total international airli ne passenger traffic arriving to the United States in 2006. Domestic airline passengers traveled to Florida more frequently during March than any other month while international airline passengers did so in July. September registered the lowest level of a irline passenger traffic for both domestic and international flights. Figure 3 1 shows the seasonal pattern of domesti c and international airline pass enger traffic traveling to Florida. Figure 3 1 Monthly s easonal pattern of domestic and international airline passenger traffic traveling to Florida between 1990 and 2006. Source: Bureau of Transportation Statistics. Domestic a irline passenger traffic by U.S. r egion Domestic airline passenger traffic from the South region increas ed 71 % from 12.48 million in 1990 to 21.4 million in 2006 The South region also registers the largest share with 43 % of the Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Domestic 8.1% 8.6% 10.1% 9.0% 8.1% 8.0% 8.6% 8.1% 6.2% 7.8% 8.3% 9.0% International 8.9% 7.8% 8.8% 8.2% 7.5% 7.9% 10.1% 9.7% 7.1% 7.5% 7.7% 8.8% 5.0% 6.0% 7.0% 8.0% 9.0% 10.0% 11.0% 12.0% Percentage of Total Passengers Domestic International

PAGE 59

59 domestic airline passenger traffic traveling to Florida. The Northeast region (29%) ranks second followed by the Midwest region (19 %). The West region accounts for 6 % of total domestic airline passenger traffic and the Other region accounted for remaining 3 % Despite having the smallest share of the four major U.S. regions the West region experienced the greatest relative growth (2 11%) in the 17 year period from 1.23 million airline passengers in 1990 to approximately 3.82 million airline passengers in 2006. Figure 3 2 illustrates total airline passenger traffic from five U.S. regions traveling to Florida from 1990 to 2006. There h ave not been substantial changes in the shares between U.S. regions from 1990 to 2006. The South region has kept its dominance throughout the 17 year period but its share has experienced a decrease from 44 % in 1990 to 41 % in 2006. Meanwhile the West region has increased its share from 4 % in 1990 to 7 % in 2006. Figure 3 3 illustrates the share of domestic airline passenger traffic from five U.S. regions traveling to Florida in 1990, 1998, and 2006. Domestic airline passengers from every U.S. region traveled to Florida more frequently during March than any other month September registered the lowest level of airline passenger traffic from all U.S. regions. H igh airline traffic level s in March could be attributed to unfavorable weather in the other U.S. regio ns enticing passengers to travel to Florida. Likewise, low airline traffic levels in September could be attributed to favorable weather in the other U.S. regions dissua ding passengers from travel ing to Florida. Figure 3 4 shows the seasonal pattern of dome stic airline passenger traffic traveling to Florida.

PAGE 60

60 Figure 3 2 Florida CSA: t otal domestic airline passenger traffic from f ive U.S. regions between 1990 and 2006. Source: Bureau of Transportation Statistics. Figure 3 3 Florida CSA: s hare of total domestic airline passenger traffic from f ive U.S. regions in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics. 0.0 5.0 10.0 15.0 20.0 25.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Other West Midwest Northeast South 41% 42% 44% 31% 27% 32% 17% 21% 18% 7% 6% 4% 3% 4% 2% Market Share in Percentage 1990 Year 1998 2006 South Northeast Midwest West Other

PAGE 61

61 Figure 3 4 Florida CSA: m onthly s easonal pattern of domestic airline passenger traffic from four U.S. regions between 1990 and 2006. Source: Bureau of Transportation Statistics. Domestic a irline p assenger t raffic by d estination CSA Nearly 648 million domestic airline passengers traveled to Florid a from 1990 to 2006, with an annual average of more than 38 million airline passengers during the 17 year period. Also, Florida experienced growth in domestic airline passengers of 83 % from 28.27 million in 1990 to 51.71 million in 2006. The destination CS As of Orlando, South Florida, Tampa St. Petersburg, Jacksonville, and F or t Myers accounted for 92 % of the total domestic airline passenger traffic market traveling to Florida. The Other category includes the remaining 17 destination CSAs in the state. M oreover, the South Florida CSA received 38 % of the total domestic passenger traffic, followed by the Orlando CSA (28%), Tampa St. Petersburg CSA (16%), Fort Myers CSA (6%), and Jacksonville CSA (5%). The Jacksonville CSA experienced the largest relative gr owth Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec West 8.4% 8.0% 9.5% 8.5% 8.0% 8.5% 8.8% 8.5% 6.7% 7.8% 8.1% 9.2% Midwest 8.5% 9.7% 11.7% 9.3% 7.6% 7.7% 7.7% 7.1% 5.6% 7.6% 8.1% 9.4% Northeast 8.2% 9.0% 10.1% 9.4% 7.8% 7.4% 8.4% 8.5% 6.0% 7.7% 8.4% 9.0% South 7.6% 8.1% 9.6% 8.6% 8.4% 8.6% 8.7% 8.2% 6.6% 8.2% 8.3% 9.0% 5% 6% 7% 8% 9% 10% 11% 12% 13% Percentage of Total Passengers West Midwest Northeast South

PAGE 62

62 (130%) during the 17 year period, followed by the Fort Myers CSA (121%). The South Florida CSA reported the smallest relative growth (59%) during the same time period, as shown in Figure 3 5. Figure 3 5 Total domestic ai rline passenger traffic traveling to the destination CSA s in Florida between 1990 and 2006. Source: Bureau of Transport ation Statistics. South Florida CSA domestic airline passenger traffic Total domestic airline passenger traffic to the South Florida CSA increased 58 % from nearly 12.0 million in 1990 to approximately 19.0 million in 2006 The South region accounts for the largest share with 38 % of the domestic airline passenger traffic traveling to the South Florida CSA. The Northeast region (36%) ranks se cond followed by the Midwest (15%) and West (7%) region s In spite of having the smallest share of the four major U.S. regions the West region experienced the greatest relative growth (177%) in the 17 year period from 0.61 million airline passengers in 19 90 to approximately 1.69 million airline passengers in 2006. The South region experienced the second largest relative growth (57%) follo wed by the Northeast (51%) and 0.0 5.0 10.0 15.0 20.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Jacksonville Ft. Myers Other Tampa St. Petersburg Orlando South Florida

PAGE 63

63 Midwest (42%) regions Figure 3 6 illustrates total domestic airline passenger traffic fr om five U.S. regions traveling to the South Florida CSA from 1990 to 2006. The South region has kept the largest share (38%) during the 17 year period. Meanwhile the West region has increased its share from 5 % in 1990 to 9 % in 2006, whereas the Northeast % in 1990 t o 37 % in 2006. Figure 3 7 illustrates the share of domestic airline passenger traffic from five U.S. regions traveling to the South Florida CSA in the years 1990, 1998, and 2006. Do mestic airline passengers fr om all U.S. regions traveled to the South Florida CSA more frequently during March than any other month. September registered the lowest level of airline passenger traffic f rom all U.S. regions. Figure 3 8 shows the seasonal pattern of domestic airline pas senger traffic traveling to the South Florida CSA. Figure 3 6 South Florida CSA: t otal d omestic airline passenger traffic from f ive U.S. regions between 1990 and 2006. Source: Bureau of Transportation Statistics. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Other West Midwest Northeast South

PAGE 64

64 Figure 3 7 South Florida CSA : share of t otal domestic airline passenger traffic from f ive U.S. regions in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics. Figure 3 8 South Florida CSA : mon thly s easonal pattern of domestic airline passenger traffic from four U.S. regions between 1990 and 2006. Source: Bureau of Transportation Statistics. 37% 37% 38% 37% 32% 39% 13% 16% 14% 9% 6% 5% 4% 9% 4% Market Share in Percentage 1990 Year 1998 2006 South Northeast Midwest West Other Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec West 9.1% 8.4% 9.9% 9.1% 8.4% 8.0% 8.5% 8.5% 6.2% 7.1% 7.9% 9.0% Midwest 9.4% 10.4% 12.2% 9.6% 7.6% 7.0% 7.1% 6.8% 5.2% 6.9% 8.0% 9.8% Northeast 9.2% 9.6% 10.5% 9.7% 7.9% 6.8% 7.9% 8.1% 5.6% 7.2% 8.2% 9.3% South 8.5% 8.7% 10.1% 9.0% 8.7% 8.0% 8.2% 7.8% 6.2% 7.5% 8.1% 9.1% 3.0% 5.0% 7.0% 9.0% 11.0% 13.0% Percentage of Total Passengers West Midwest Northeast South

PAGE 65

65 Orlando CSA domestic airline passenger traffic Nearly 187 million domestic airline passengers traveled to the Orlando CSA from 1990 to 2006, with an annual average of 11.0 million domestic airline passengers during the 17 year period Domestic airline passenger traffic to the Orlando CSA more than doubled from 7.45 million in 1990 to approximately 15.62 milli on in 2006 Among the major U.S. regions, the South region accounted for the largest share with 37 % of domestic airline passenger traffic traveling to the Orlando CSA. The Northeast region (30%) ranked second fol lowed by the Midwest (22%) and West (8%) reg ions In spite of having the smallest share of the four major U.S. regions, the West region experienced a three fold increase in the number of airline passengers during the 17 year period from 0.45 million in 1990 to approximately 1.37 million in 2006. Th e Northeast region experienced the second largest relative growth (124%) followed by the Midw est (109%) and South (78%) regions Figure 3 9 illustrates total domestic airline passenger traffic from five U.S. regions traveling to the Orlando CSA from 1990 t o 2006. Even though the South region registered the largest share (37%) in 2006, its share has declined since 1990 when it accounted for 42 % of total domestic airline passengers traveling to the Orlando CSA. On the other hand, two other major U.S. regions have experienced growth in their shares. The Midwest region has increased its share from 29 % in 1990 to 31 % in 2006. Figure 3 10 illustrates the share of domestic airline passenger traffic from five U.S. regions traveling to the Orlando CSA in the years 19 90, 1998, and 2006. Domestic airline passengers from four major U.S. regions traveled to the Orlando CSA more frequently during March than any other month. September registered the lowest level of airline passenger traffic from all U.S. regio ns, as illust rated in Figure 3 11

PAGE 66

66 Figure 3 9 Orlando CSA: t otal d omestic airline passenger traffic from five U.S. region s between 1990 and 2006. Source: Bureau of Transportation Statistics. Figure 3 10 Orl ando CSA: s hare of t otal domestic airline pas senger traffic from five U.S. region s in 1 990, 1998, and 2006. Source: Bureau of Transportation Statistics. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Other West Midwest Northeast South 36% 37% 42% 31% 29% 29% 21% 24% 21% 9% 8% 6% 3% 2% 2% Market Share in Percentage 1990 Year 1998 2006 South Northeast Midwest West Other

PAGE 67

67 Figure 3 11 Orlando CSA : monthly s easonal pattern of domestic airline p assenger traffic from four U.S. regions between 1990 and 2006. Source: Bureau of Transportation Statistics Tampa St. Petersburg CS A d omestic a irline p assenger t raffic The Tampa St. Petersburg CSA accounts for 16 % of the total domestic airline passenger tr affic traveling to Florida making it the third largest destination market in the state. Its airline passenger traffic more than doubled from 4.07 million in 1990 to approximately 8.25 million in 2006 The South region held the largest share with 43 % of the domestic airline passenger traffic traveling to the Tampa St. Petersburg CSA. The Northeast region (26%) ranked second fol lowed by the Midwest (24%) and West (6%) regions In spite of having the smallest share of the four U.S. regions the West region ex perienced the greatest relative growth (357%) in the 17 year period from 0.15 million airline passengers in 1990 to approximately 0.69 million airline passengers in 2006. The Northeast region experienced Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec West 8.3% 8.0% 9.4% 8.5% 8.3% 8.7% 8.7% 8.3% 6.9% 8.2% 7.9% 8.9% Midwest 7.8% 8.9% 10.5% 8.9% 8.1% 8.7% 8.5% 7.7% 6.1% 8.3% 8.0% 8.6% Northeast 7.4% 8.3% 9.7% 9.4% 8.3% 7.9% 8.9% 9.1% 6.2% 8.1% 8.3% 8.2% South 7.4% 8.1% 9.5% 8.6% 8.6% 8.9% 8.9% 8.1% 6.6% 8.5% 8.2% 8.5% 5% 6% 7% 8% 9% 10% 11% Percentage of Total Passengers West Midwest Northeast South

PAGE 68

68 the second largest relative growth (113%) followed b y the Midwest (85%) and South (84%) regions as illustrated in Figure 3 12. Both the South and Midwest regions have experienced a decrease in the share of domestic airline passenger traffic to the Tampa St. Petersburg CSA between 1990 and 2006. In contrast, the Northeast and West regions have increased their share during the same time period. Figure 3 13 shows the evolution of the share of domestic airline passenger traffic traveling to the Tampa St. Petersburg CSA. Domestic airline passengers from every U. S. region traveled to the Tampa St. Petersburg CSA more frequently during March than any other month September registered the lowest level of airline passenger traffic from the four major U.S. regions. Figure 3 14 shows the seasonal pattern of domestic ai rline passenger traffic traveling to the Tampa St. Petersburg CSA. Figure 3 12 Tampa St. Petersburg CSA : t otal a irline passenger traffic from five U.S. region s between 1990 and 2006. Source: Bureau of Transportation Statistic s. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Other West Midwest Northeast South

PAGE 69

69 Figure 3 13 Tampa St. Petersburg CSA : s hare of total domestic airline pas senger traffic from five U.S. reg ions in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics. Figure 3 14 Tampa St. Petersburg CSA : monthly s easonal pattern of domestic airline passenger traffic from four U.S. regions between 1990 and 2006. Source: Bureau of Transportation Statistics 40% 44% 44% 30% 24% 28% 21% 26% 23% 8% 5% 4% 1% 1% 1% Market Share in Percentage 1990 Year 1998 2006 South Northeast Midwest West Other Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec West 8% 8% 10% 8% 8% 8% 9% 8% 7% 8% 8% 9% Midwest 8% 9% 12% 9% 8% 8% 8% 7% 6% 8% 8% 9% Northeast 8% 9% 10% 9% 8% 8% 9% 9% 6% 8% 8% 9% South 7% 8% 10% 9% 8% 9% 9% 8% 7% 8% 8% 9% 5% 6% 7% 8% 9% 10% 11% 12% Percentage of Total Passengers West Midwest Northeast South

PAGE 70

70 Jacksonville CSA domestic airline passenger traffic Total domestic airline passenger traffic traveling to the Jacksonville CSA increased 130 % from 1.16 million in 1990 to approximately 2.66 million in 2006 The South region has the largest share with almost 71 % of the total domestic airline passenger traffic traveling to the Jac ksonville CSA. The Northeast (14%) and Midwest (14%) regions ranked second followed by the West region (<1%). Figure 3 15 shows total domestic airline passenger traffic from each U.S. region traveling to the Jacksonville CSA from 1990 to 2006. The South r egion has kept the largest market share despite experiencing a reduction from 73 % to 69 % market share from 1990 to 2006. In contrast, share from the Midwest region has increased since 1990. Also, the Northeast region recorded an increase from 16 % to 17 % in its market share of domestic airline passengers traveling to the Jacksonville CSA during the same time period. Figure 3 16 shows the evolution of the share of domestic airline passenger traffic traveling to the Jacksonville CSA since 1990. Total domestic airline passenger traffic from the South, Northeast, and Midwest regions traveled more frequently to the Jacksonville CSA during March, while passengers from the West region did so in July. January experienced the lowest levels of domestic airline passeng er traffic originating from the South, Northeast, and Midwest regions The West region traveled less frequently during September to the Jacksonville CSA than any other month Figure 3 17 illustrates monthly seasonal patterns of domestic airline passenger t raffic traveling to the Jacksonville CSA from 1990 to 2006.

PAGE 71

71 Figure 3 15 Jacksonville CSA: t otal domestic airline passenger traffic from five U.S. reg ion s between 1990 and 2006. Source: Bureau of Transportation Statistics. Figure 3 16 Jacksonville CSA: s h are of total domestic airline passenger traffic from four U.S. region s in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics. 0.0 0.5 1.0 1.5 2.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Thousands Year Other West Midwest Northeast South 69% 70% 73% 17% 13% 16% 12% 16% 11% 2% 0% 1% Market Share in Percentage 1990 Year 1998 2006 South Northeast Midwest West

PAGE 72

72 Figure 3 17 Jacksonvill e CSA: monthly s easonal pattern of domestic airline passenger traffic from four U.S. region s between 1990 and 2006. Source: Bureau of Transportation Statistics. Fort Myers CSA domestic airline passenger traffic Total domestic airline passenger traffic trav eling to the Fort Myers CSA increased 121 % from 1.57 million in 1990 to approximately 3.47 million in 2006 The South region recorded the largest share with 38 % of the total domestic airline passenger traffic traveling to the Fort Myers CSA. The Midwest re gion (35%) ranked second, followed by the Northeast (26%) and West (<1%) region s F our major U.S. regions experienced growth in terms of number of airline passengers traveling to Fort Myers CSA during the 17 year period. The West region registered the larg est increase at 606 % followed by the Northeast (164%), Midwest (158%), and S outh (67%) regions Figure 3 1 8 shows t otal domestic airline passenger traffic from f ive U.S. regions traveling to the Fort Myers CSA from 1990 to 2006. Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec West 7% 7% 9% 8% 9% 9% 10% 10% 7% 8% 7% 9% Midwest 6% 7% 10% 9% 8% 9% 9% 9% 7% 9% 8% 8% Northeast 6% 7% 9% 9% 8% 8% 9% 9% 7% 9% 9% 9% South 7% 7% 9% 9% 9% 9% 9% 9% 7% 8% 8% 9% 5.0% 6.0% 7.0% 8.0% 9.0% 10.0% 11.0% Percentage of Total Passengers West Midwest Northeast South

PAGE 73

73 The South region experienc ed the largest drop of market share of total domestic airline passenger traffic traveling to the Fort Myers CSA, decreasing from 45 % in 1990 to 34 % in 2006 By 2006, the Northeast, Midwest, and South regions held one third of the total share each. The Midw % to 33 % during the same time period. The Northeast region also increased its share from 27 % in 1990 to 32 % in 2006. Figure 3 19 shows the evolution of the share of domestic airline passenger traffic traveling to the For t Myers CSA since 1990. Total domestic airline passenger traffic from the South, Northeast, and Midwest regions traveled more frequently to the Fort Myers CSA during March, while passengers from the West region did so in December. September experienced the lowest levels of domestic airline passenger traffic originating from f our major U.S. regions. Figure 3 20 illustrates monthly seasonal patterns of domestic airline passenger traffic traveling to the Fort Myers CSA from 1990 to 2006. Figure 3 18 Fort Myers CSA: t otal domestic airline passenger traffic from five U.S. region s between 1990 and 2006. Source: Bureau of Transportation Statistics. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Other West Northeast Midwest South

PAGE 74

74 Figure 3 19 Fort Myers CSA: s hare of total domestic airli ne passenger traffic from four U.S. region s in 1 990, 1998, and 2006. Source: Bureau of Transportation Statistics. Figure 3 20 Fort Myers CSA: monthly s easonal pattern of domestic airline pa ssenger traffic from four U.S. regio ns between 1990 and 2006. Source: Bureau of Transportation Statistics. 34% 36% 45% 32% 23% 27% 33% 41% 28% 1% 0% 0% Market Share in Percentage 1990 Year 1998 2006 South Northeast Midwest West Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec West 12% 9% 14% 11% 6% 5% 5% 4% 3% 5% 11% 15% Midwest 10% 12% 16% 11% 6% 6% 5% 5% 4% 7% 8% 10% Northeast 9% 11% 13% 11% 7% 6% 7% 7% 5% 7% 9% 9% South 9% 10% 12% 10% 8% 7% 7% 7% 5% 8% 8% 9% 0.0% 2.0% 4.0% 6.0% 8.0% 10.0% 12.0% 14.0% 16.0% 18.0% Percentage of Total Passengers West Midwest Northeast South

PAGE 75

75 International a irline p assenger t raffic by w orld r egion The Latin America region accounted for the largest share with 65 % of the total international airline passenger traffic traveling to Florida from 1990 to 2006, followed by Europe (23%) and Canada (11%) Also, Latin America has experienced the largest increase (79%) in airline passengers since 1990, followed by Europe (63%). In contrast, airline passenger traffic from Canada has drop ped 5 % from 1990 to 2006. In 2006, the Latin America and Europe region almost matched its previous highest level of airline passenger traffic reached in 2000, while Canada failed to improve its highest airline passenger level (approximately 1.14 million ai rline passengers) reac hed in 1990. Figure 3 21 shows total international airline passenger traffic from four w orld regions traveling to Florida from 1990 to 2006. Latin America increased its share of the international airline passenger traffic traveling to Florida from 61 % in 1990 to 68 % % during the same time period. Europe has kept its share at 23 % during the 17 year period as presented in Figure 3 22. International airline passengers from Canada traveled to Flo rida more frequently during March, while September registered the lowest level. H igh airline traffic level s in March could be attributed to unfavorable weather in Canada enticing airline passengers to travel to Florida. Likewise, low airline traffic levels in September could be attributed to favorable weather in Canada dissuadin g airline passengers from travel to Florida. The Latin America and Europe regions registered the highest airline passenger traffic during July. Airline passenger levels from Europe t raveled less in January, while September registered the lowest level of airline passengers traveling from Latin America as shown in Figure 3 23.

PAGE 76

76 Figure 3 21 Florida CSA: t otal international airline passenger traffic from fou r world regi on s between 1990 and 2006. Source: Bureau of Transportation Statistics. Figure 3 22 Florida CSA: s hare of total international airline passenger traffic from four world regions in 1990, 1998, and 2006. Source: Bur eau of Transportation Statistics. 0.0 2.0 4.0 6.0 8.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Other Canada Europe Latin America 67% 66% 61% 22% 23% 22% 10% 10% 18% 0% 1% 0% Market Share in Percentage 1990 Year 1998 2006 Latin America Europe Canada Other

PAGE 77

77 Figure 3 23 Florida CSA: m onthly s easonal pattern of total international airline passenger traffic from four world regions between 1990 and 2006. Source: Bureau of Transportation Statistics. International a irline passenger t raffic by d estination CSA More than 155 million international airline passengers traveled to Florida from 1990 to 2006, with an annual average of approximately 9.14 million airline passengers during the 17 year period. Als o, Florida experienced growth in international airline passenger traffic of 61 % from 6.53 million in 1990 to 10.5 million in 2006. The destination CSAs of South Florida, Orlando, and Tampa St. Petersburg accounted for 99 % of total international airline pas senger traffic market traveling to Florida. Moreover, the South Florida CSA received 81 % of the total international passenger traffic, followed by the Orlando CSA (15%), and the Tampa St. Petersburg CSA (3%). The South Florida CSA experienced the largest r elative growth (93%) during the 17 year period, followed by the Orlando CSA (63%). The Tampa St. Petersburg CSA reported a 45 % decline in number of international airline passengers during the same time Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Other 8% 8% 7% 6% 8% 8% 7% 7% 7% 10% 13% 9% Canada 13% 14% 14% 10% 5% 4% 5% 5% 4% 5% 9% 11% Europe 7% 7% 8% 8% 8% 9% 11% 10% 9% 10% 7% 7% Latin America 10% 8% 8% 8% 7% 8% 10% 10% 7% 7% 7% 9% 0% 2% 4% 6% 8% 10% 12% 14% 16% Percentage of Total Passengers Other Canada Europe Latin America

PAGE 78

78 period. Figure 3 24 presents to tal i nternational airli ne passenger traffic traveling to the top three CSAs in Florida between 1990 and 2006. Figure 3 24 Total international airline passenger traffic traveling to the top th ree destination CSAs in Florida between 1990 and 2006. Sou rce: Bureau of Transportation Statistics. South Florida CSA i nternational a irline passenger t raffic International airline passenger traffic to the South Florida CSA increased 63 % from 5.21 million in 1990 to nearly 8.46 million in 2006 The Latin America r egion accounts for the largest share with 77 % of the international airline passenger traffic traveling to the South Florida CSA. The Europe region (15%) ranks second, followed by Canada (7%). Also, Latin America recorded the largest relative growth (80%) i n the 17 year period from 3.76 million in 1990 to approximately 6.77 million airline passengers in 2006. The Europe region experienced the second largest relative growth at 35 % Figure 3 25 illustrates total international airline passenger traffic travelin g to the South Florida CSA from 1990 to 2006. 0.0 2.0 4.0 6.0 8.0 10.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Tampa Orlando South Florida

PAGE 79

79 Latin America increased its share of i nternational airline passenger traffic traveling to the South Florida CSA from 72 % in 1990 to 80 % 12 % to 7 % during the same % to 13 % during the 17 year period. Figure 3 26 illustrates the evolution of the share of international airline passenger traffic from four world regions traveling to the South Florida CSA from 1990 t o 2006. International airline passengers from Canada traveled to the South Florida CSA more frequently during March, while September registered the lowest level. Latin America and Europe registered the highest levels of airline passenger traffic during Jul y. Airline passenger from Europe traveled less frequently in May, while September registered the lowest level of airline passengers traveling from Latin America. Figure 3 27 shows monthly seasonal pattern of international airline passenger traffic travelin g to the South Florida CSA. Figure 3 25 South Florida CSA: t otal international airline passenger traffic from four world regions between 1990 and 2006. Source: Bureau of Transportation Statistics. 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Other Canada Europe Latin America

PAGE 80

80 Figure 3 26 South Florida CSA: s hare of total international airline passenger traffic from four world region s in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics. Figure 3 27 South Florida CSA: monthly s e asonal pattern of total international airline passenger from four world regions between 1990 and 2006. Source: Bureau of Transportation Statistics. 80% 78% 72% 13% 14% 16% 7% 7% 12% 0% 1% 0% Market Share in Percentage 1990 Year 1998 2006 Latin America Europe Canada Other Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Other 8.3% 7.0% 8.5% 7.9% 7.7% 8.4% 9.4% 9.2% 7.3% 8.2% 8.2% 9.8% Canada 12.4% 12.9% 13.1% 9.4% 5.2% 4.2% 5.6% 5.6% 4.1% 5.6% 9.7% 12.1% Europe 7.8% 7.8% 9.1% 8.4% 7.2% 7.6% 10.1% 8.9% 7.3% 8.9% 8.2% 8.8% Latin America 9.4% 7.4% 8.3% 7.9% 7.6% 8.2% 10.3% 10.2% 7.0% 7.1% 7.6% 8.8% 3% 5% 7% 9% 11% 13% 15% Percentage of Total Passengers Other Canada Europe Latin America

PAGE 81

81 Orlando CSA i nternational airline p assenger t raffic International airline passenger traffic to the Orlando C SA almost double d from 0.86 million in 1990 to nearly 1.66 million in 2006 The Europe region dominated the Orlando CSA market with a 68 % share of t otal international airline passenger traffic. The Canada region (16%) ranks second, followed by Latin Americ a (15%). Also, Latin America recorded the largest relative growth (152%) in the 17 year period from 0.10 million in 1990 to approximately 0.26 million airline passengers in 2006. The Europe region experienced the second largest relative growth at 92 % follo wed by Canada (64%). Figure 3 2 8 illustrates total international airline passenger traffic traveling to the Orlando CSA from 1990 to 2006. The Europe region recorded a market share of 65 % of total international airline passenger traffic traveling to the Or lando CSA in 1990. This U.S. region showed an increase of 70 % in 1998, but it decreased to 65 % by 2006. In contrast Latin America increased its share from 12 % in 1990 to 16 % % to 19 % during the same time period. Figure 3 2 9 illustrates the evolution of the share of international airline passenger traffic from four world regions traveling to the Orlando CSA from 1990 to 2006. International airline passengers from Europe traveled to the Orlando CSA more fre quently during July, while January registered the lowest level. The Latin America region registered the highest level of airline passenger traffic during July, while Canada did so in March. Airline passenger from Latin America and Canada traveled less freq uently in September as shown in Figure 3 30

PAGE 82

82 Figure 3 28 Orlando CSA: t otal international airline passenger traffic from four world regions between 1990 and 2006. Source: Bureau of Transportation Statistics. Figure 3 29 Orlando CSA: s hare of total international airline passenger traffic from four world regions in 1 990, 1998, and 2006. Source: Bureau of Transportation Statistics. 0 200 400 600 800 1,000 1,200 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Other Latin America Canada Europe 66% 69% 66% 19% 13% 22% 16% 15% 12% 0% 2% 0% Market Share in Percentage 1990 Year 1998 2006 Europe Canada Latin America Other

PAGE 83

83 Figure 3 30 Orlando CSA: monthly s e asonal pattern of total international airline passenger traffic from four world region s between 1990 and 2006. Source: Bureau of Transportation Statistics. Tampa St. Petersburg CSA i nternational a irline p assenger t raffic International airline passenger tr affic to the Tampa St. Petersburg CSA decreased from 0.43 million in 1990 to nearly 0.24 million in 2006. The Canada region dominated the Tampa St. Petersburg CSA market with 60 % share of total international airline passenger traffic. The Latin America reg ion (21%) ranks second, followed by Europe (21%). Europe is the only region that recorded a positive relative growth (78%) in the 17 year period. Both Canada and Latin America experienced a decrease of 62 % and 45 % respectively. Figure 3 3 1 illustrates tota l international airline passenger traffic traveling to the Tampa St. Petersburg CSA from 1990 to 2006. Canada has experienced a drastic decrease in its share of international airline passenger traffic traveling to the Tampa St. Petersburg CSA from 70 % in 1 990 to 47 % in 2006. In contrast, Europe increased its share from 10 % in 1990 to 32 % in 2006. Meanwhile, Latin America Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Canada 12% 14% 13.9% 10.8% 5.8% 4.7% 6.2% 6.2% 3.9% 5.5% 7.8% 9.6% Europe 5% 5% 6.1% 7.2% 9.9% 10.1% 12.3% 11.8% 10.9% 10.6% 5.8% 5.6% Latin America 9.05% 6.56% 7.2% 7.1% 6.4% 8.5% 15.4% 11.0% 6.0% 6.2% 6.0% 10.6% 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% Percentage of Total Passengers Canada Europe Latin America

PAGE 84

84 increased its 1990 share of approximately 20 % to 21 % in 2006. Figure 3 3 2 illustrates the evolution of the share of international airline passenger traffic from the world regions traveling to the Tampa St. Petersburg CSA from 1990 to 2006. International airline passengers from Europe and Canada traveled to the Tampa St. Petersburg CSA more frequently during March, while passengers from Latin America did so in July. Airline passengers from Canada and Latin America traveled less frequently in September, while airline passengers from Europe did so in January. Figure 3 3 3 shows monthly seasonal pattern s of international airline passenger traffic traveling to the Tampa St. Petersburg CSA from 1990 to 2006 Figure 3 31 Tampa St. Petersburg CSA: t otal international airline passenger traffic from four world regions between 1990 and 2006. Source: Bureau of Transportation Statistics. 0 50 100 150 200 250 300 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Passengers in Millions Year Other Latin America Europe Canada

PAGE 85

85 Figure 3 32 Tampa St. Petersburg CSA: s hare of total international airline passenger traffic from four world region s in 1990, 1998, and 2006. Source: Bureau of Transportation Statistics. Figure 3 33 Tampa St. Petersburg CSA : monthly s easonal pattern o f total international airline passenger traffic from four world region s between 1990 and 2006. Source: Bureau of Transportation Statistics. 47% 55% 70% 32% 24% 10% 21% 21% 21% 0% 0% 0% Market Share in Percentage 1990 Year 1998 2006 Canada Europe Latin America Other Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Canada 11.3% 14.0% 15.8% 12.2% 5.9% 3.7% 4.4% 4.6% 3.5% 6.0% 9.1% 9.5% Europe 6.4% 7.3% 10.9% 8.6% 7.2% 8.0% 10.1% 8.7% 8.1% 9.4% 7.8% 7.6% Latin America 8.6% 7.7% 9.2% 8.7% 8.9% 9.8% 10.5% 9.5% 5.3% 6.8% 7.8% 7.3% 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% Percentage of Total Passengers Canada Europe Latin America

PAGE 86

86 Airline Ticket Prices Description, Selection, and Aggregation of Airline Ticket Price Data for Domestic Flights Data on airline ticket prices 1 for domestic flights used in this study were collected by the Bureau of Trans portation Statistics (BTS) and are available online. The Origin and Destination S urvey (DB1B) Market was used to collect one way and round trip airline ticket prices from several domestic origins traveling to Florida. According to the BTS, the DB1B Market contains (directional) origin and destination markets, which is a 10 % sample of a irline tickets from reporting carriers. It includes items such as passengers, air fares, and distances for each directional market, as well as information about whether the market was domestic or international. Nearly 46.8 million observations reported in a quarterly basis were retrieved from the DB1B Market database between Quarter I, 1993 and Quarter IV, 2006. A market is defined by the first departure airport on a ticket and the ultimate arrival airport. The DB1B Market database includes a market ID that identifies the itinerary of a passenger and it is unique to each itinerary. If the market ID appears once in the data, it represents that the itinerary was a one way trip. If the market ID appears more than once, the itinerary was a round trip and may inc lude multiple stops. For example, if a passenger traveled from New York City to Jacksonville then to Miami and finally back to New York City, the New York Ci ty to Jacksonville, (2) Jacksonville to Miami, (3) Miami to New York City For simplification, all observations with more than two market IDs were dropped from the analysis. That is, only itineraries involving two legs, origin to destin ation and destinatio n to origin, were selected to calculate average round trip airline ticket prices. For market IDs appearing twice in 1 The terms airline ticket prices and airfare are equivalent in this study and both are used interchangeably.

PAGE 87

87 the database, the first entry represents the first leg of the itinerary and the second entry represents the s econd (or returning) leg This information is useful to determine the origin and the ultimate destination of a flight Also, itineraries with one market ID were selected to calculate one way airline ticket prices. Within the DB1B M arket database, only those itineraries that included a city in Florida were selected. Then, a few airports were selected to represent the average airline ticket price of each U.S. region to follow the airline passenger data scheme. Itineraries from La Guardia International Airport and John F. Kennedy Internati onal Airport from New York City and Logan International Airport from Boston were selected as a proxy for the average airline ticket price for and Lambert Saint Louis International Airport from Saint Louis were selected to calculate the average airline ticket price for the Midwest region. Hartsfield Jackson Atlanta International Airport from Atlanta, Dallas/ Fort Worth International Airport and Fort Worth Alliance from Dallas, and George Bush Intercontinental Airport and William P. Hobby Airport in Houston were selected to calculate the average airline ticket price for the South region. Finally, itineraries from Los Angeles International Airport in Los Angeles, Denver I nternational Airport from Denver, and Seattle Tacoma International Airport in Seattle were selected to calculate the average airline ticket price for the West region. Then, destination airports in Florida chosen for the airline ticket price analysis are th e following: Miami International Airport for the South Florida CSA, Orlando International Airport for the Orlando CSA, Tampa International Airport for the Tampa St. Petersburg CSA, Jacksonville International Airport for the Jacksonville CSA, and the Southw est Florida Regional

PAGE 88

88 Airport for the Fort Myers CSA. Also, an average of the airline ticket prices of the top five destination CSAs was calculated to represent the average airline ticket price for the Florida CSA Once the airports from each origin region and destination CSA were chosen, averages were calculated for each origin region traveling to each destination CSA. In other words, there were two average airline ticket prices: ( 1) one way price and ( 2) round trip price from each originating region trave ling to each destination CSA for every quarter from 1993 to 2006. For example, all ticket price entries from New York and Boston to Miami were used to calculate the airline ticket price from the Northeast region to the South Florida CSA for every quarter f rom 1993 to 2006 Description, Selection, and Aggregation of Airline Ticket Price Data for International Flights Data on ticket prices for international flights were estimated using the passenger yield reported by the Air Transport Association Group (ATA G ). According to the ATA G web site, "passenger yield" is the average price someone pays to fly one mile (excluding government taxes and fees, which often constitute a substantial portion of an airline ticket). Passenger yield data, which ATA G reports mo nth ly by geographic region, use reports from seven U.S. airlines and results are not adjusted for inflation or trip length. D ata are based on 100 % of scheduled service and reflects all "revenue" passengers, including those redeeming frequent flyer miles ( USD 0 air fare) for award travel. P assenger yield data are often used as one indicator of recent U.S. airline market/pricing trends. The ATA G reports the passenger yields of four different regions: ( 1) Domestic (U.S.) ( 2) Atlantic (Europe and Africa) ( 3) Lati n (Latin America) and ( 4) Pacific (Asia). The passenger yield was then multiplied by the average distance from its originating world region. A verage distance was calculated using the T 100 International Market database which

PAGE 89

89 reports the miles between airp orts. For example, in order to get an average distance between Europe and South Florida CSA, an average distance from all the flights originated from Europe to Miami International Airport in Miami was calculated. Then, this average distance was multiplied by two to denote a round trip airline ticket and finally the passenger yield reported for A tlantic flights was multiplied by the average distance to get the average airline ticket price of a round trip flight from Europe to the South Florida CSA. Similar c alculations were performed for the Canada region and the Latin America region. Note that the passenger yield for domestic flights was used to estimate an average ticket price of a round trip flight from Canada to Florida. The following sections discuss nom inal domestic airline ticket prices by U.S. r egion and nominal international airline ticket prices by world region traveling to Florida and each destination CSA. Domestic Airline Ticket Prices by U.S. Region One way airline ticket prices on flights to Flor ida have de creased since 1993. Over the 15 year period the Midwest region has experienced the largest decrease in one way airline ticket prices (16%) while the South region records the smallest decrease at 1% Passengers from the West region paid more for one way airline ticket s to Florida than any other region. This fact could explain why this U.S. region recorded the lowest level of airline traffic among U.S. regions traveling to Florida as discussed in the previous section. Distance could be attributed t o the higher prices in the West region Note that the year 2000 recorded the highest price levels on one way airline ticket s on three of the four U.S. regions Since 2001 prices have been decreasing until 2005 where one way airline ticket prices started to increase. Higher oil prices could be responsible for this recent upward trend. Figure 3 3 4 shows average one way trip airline ticket prices from each U.S. region traveling to Florida from 1993 to 2006.

PAGE 90

90 Figure 3 34 Florida C SA: a verage one way airline ticket prices for domestic flights from four U.S. regions between 1993 and 2006. Source: Bureau of Transportation Statistics. Round trip airline ticket prices to Florida have decreased 2% overall since 1993. But airline passenge rs traveling from the Northeast region paid 3% more for round trip airline ticket s in 200 6 compared to 1993 In contrast, the other three U.S. regions have experienced round trip airline ticket price decreases over the same time period. The South region re cords the largest decrease in round trip airline ticket prices (8%) while the West region records the smallest decrease at 1%. As with one way airline ticket prices, airline passengers from the West region pa id more for round trip airline ticket s to Florid a than any other region. This fact could explain why this U.S. region recorded the lowest level of airline traffic among U.S. regions traveling to Florida. Distance could be attributed to the higher prices in the West region Figure 3 3 5 shows average roun d trip airline ticket prices from each U.S. region traveling to Florida from 1993 to 2006. 0 100 200 300 400 500 600 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 91

91 Figure 3 35 Florida CSA: a verage round trip airline ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006. Source: Bureau of Transportation Statistics. Since 1993 both one way and round trip airline ticket s from each U.S. region reported higher prices ( USD 353 and USD 366 respectively) during the first quarter of the calendar year compared to the other three qu arters On the other hand, the third quarter reports the lowest prices on one way ( USD 319) and round trip airline ticket s ( USD 330) to Florida. Comparing across U.S. regions the Northeast region paid less for one way and round trip airline ticket s while the West region paid the most in all quarters. Also note that airline passengers from the West and South region s paid more (16% and 2% respectively) for a one way airline ticket than for a round trip airline ticket to Florida compared to the other two U.S. regions Figure 3 3 6 illustrates the seasonality of average one way (OW) and round trip (RT) airline ticket prices traveling to Florida from four U.S. regions during the period between 1993 and 2006. 0 50 100 150 200 250 300 350 400 450 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 92

92 Figure 3 36 Florida CSA : quarterly s easonal pattern of average one way ( OW) and round tr ip (RT) airline ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006 Source: Bureau of Transportation Statistics. Domestic Airline Ticket Prices by Destination CSA The national average price of a one way airline ticket to the South Florida CSA was the most expensive ( USD 378) among the five destination CSAs in Florida during the period between 1993 and 2006. The South Florida CSA was also the only destination CSA th at experienced an increase of 4 % in airline ticket prices during the same time period. The Orlando CSA reported the cheapest ticket price ( USD 312) during the 14 year period. O ne way airline ticket prices to the Tampa St. Petersburg CSA experienced the lar gest decrease (14%) from USD 350 in 1993 to USD 303 in 2006. The Jacksonville CSA ranked second with a 12 % decrease, followed by the Orlando CSA (10%), and the Fort Myers CSA (4%). The average price of a round trip airline ticket traveling from the U.S. t o the South Florida CSA was the most expensive ( USD 385) among the top five destination CSAs in Florida during 0 100 200 300 400 500 600 IQ OW IQ RT IIQ OW IIQ RT IIIQ OW IIIQ RT IVQ OW IVQ RT Average Fare in U.S. Dollars Quarter Northeast Midwest All Regions South West

PAGE 93

93 the period between 1993 and 2006. Both the South Florida CSA and the Fort Myers CSA experienced an increase of 4 % in their airline ticket prices during the same time period. The Orlando CSA and the Fort Myers CSA reported the cheapest ticket price ( USD 333) during the 14 year period. R ound trip airline ticket prices to the Jacksonville CSA experienced the largest decrease (10%) from USD 403 in 1993 to USD 363 in 2006. The Tampa St. Petersburg CSA ranked second with an 8 % decrease followed by the Orlando CSA (2 %). An individual analysis of a irline ticket prices for each of the five destination CSAs is presented next. It includes a discussion of one way and round trip airline ticket prices for domestic flights originating from four U.S. regions traveling to a specific destination CSA. South Florida CSA d omestic a irline t icket p rices One way airline ticket prices on flights to the South Florida CSA hav e increased 4 % from USD 339 in 1993 to USD 352 in 2006 Despite the overall increase, the average price in 2006 was lower than the average price over the 14 year period ( USD 378) The highest price was reported in 2000 when the one way airline ticket price rose to USD 430 Then, prices experienced a steady decline from 2001 to 2005. The downward trend halted in 2006 when prices increased again. Over the 14 year period the West region has experienced the largest increase in one way airline ticket prices (15% ) from USD 444 in 1993 to USD 511 in 2006. In contrast, the Midwest region has experienced the largest decrease (9%) from USD 261 in 1993 to USD 237 in 2006. Figure 3 3 7 shows average one way airline ticket prices from each U.S. region traveling to the Sou th Florida CSA from 1993 to 2006.

PAGE 94

94 Figure 3 37 South Florida CSA: a verage o ne way air line ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006 Source: Bureau of Transportation Statistics. Round trip airline ticket prices to the South Florida CSA have increased 4 % overall since 1993. Airline passengers traveling from the West region paid 11 % more for round trip airline ticket s in 2006 compared to 1993 Similarly, the Northeast region experienced a 10 % increase in round trip airline ticket prices to the South Florida CSA. In contrast, the Midwest and South regions have experienced round trip airline ticket price decreases (3% and 4% respectively) over the same time period As with one way airline ticket prices, airline passengers from the West regi on paid more for round trip airline ticket s to the South Florida CSA than any other region All four U.S. regions recorded the highest price in 2000. Figure 3 3 8 shows average round trip airline ticket prices f rom each U.S. region traveling to the South Florida CSA from 1993 to 2006. 0 100 200 300 400 500 600 700 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 95

95 Figure 3 38 South Florida CSA: a verage round trip air ticket prices for domestic fligh ts from four U.S. r egion s between 1993 and 2006 Source: Bureau of Transportation Statistics. Since 1993 both one way and round trip airline ticket s from each U.S. region reported higher prices ( USD 401 and USD 403 respectively) during the first quarter of the calendar year compared to the other three quarters On the other hand, the third quarter reports the lowest prices on one way ( USD 362) and round trip airline ticket s ( USD 368) to the South Florida CSA. Comparing across U.S. regions, the Northeast region paid less for its one way and round trip airline ticket s whi le the West region paid the most in all quarters. Also note that airline passengers from the West and South region s paid more (16% and 4% respectively) for a one way airline ticket than for a round trip airline ticket to the South Florida CSA compared to t he other two U.S. regions Figure 3 3 9 illustrates the seasonality of average one way (OW) and round trip (RT) airline ticket prices traveling to the South Florida CSA from four U.S. regions during the period between 1993 and 2006. 0 100 200 300 400 500 600 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 96

96 Figure 3 39 S outh Florida CSA: quarterly s easonal pattern of average one way ( OW ) and round trip ( RT ) airline ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006 Source: Bureau of Transportation Statistics. Orlando CSA d omestic a irline t icket p rices One way airline ticket prices on flights to the Orlando CSA decreased 10 % from USD 330 in 1993 to USD 296 in 2006 The highest price was reported in 2000 when the one way airline ticket price rose to USD 347 Then, price decre ases followed until the year 2005 when prices increased again. Over the 14 year period the Midwest region experienced the largest decrease in one way airline ticket prices (16%) from USD 260 in 1993 to USD 218 in 2006. The Northeast region ranks second wit h a 14 % decrease in one way airline ticket prices, followed by the South (9%) and West (6%) regions Still, the West region paid more for one way airline ticket s than any other region. Figure 3 40 illustrates average one way airline ticket prices from each U.S. region traveling to the Orlando CSA during the period between 1993 and 2006. 0 100 200 300 400 500 600 IQ OW IQ RT IIQ OW IIQ RT IIIQ OW IIIQ RT IVQ OW IVQ RT Average Fare in U.S. Dollars Quarter Northeast Midwest All Regions South West

PAGE 97

97 Figure 3 40 Orlando CSA: a verage one way airline ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006 Source: Bure au of Transportation Statistics. Similarly to one way airline ticket prices, round trip airline ticket prices to the Orlando CSA have decreased 2 % overall since 1993. Airline passengers traveling from the South region paid 6 % less for round trip airline ti cket s in 2006 compared to 1993, accounting for the largest decrease among U.S. regions Similarly, the West region experienced a 5 % decrease in round trip airline ticket prices to the Orlando CSA. An opposite situation was experienced by the other two U.S. regions. The Midwest and Northeast regions experienced price increases in their round trip airline ticket s (2% and 1% respectively) over the same time period Airline pas sengers from the West region paid more for round trip airline ticket s to the Orlando CSA than any other region All four U.S. regions recorded the highest price in 2000. Figure 3 4 1 illustrates average round trip airline ticket prices from each U.S. region traveling to the Orlando CSA from 1993 to 2006. 0 100 200 300 400 500 600 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 98

98 Figure 3 41 Orlando CSA: a verage round trip air ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006 Source: Bureau of Transportation Statistics. One way and round trip airline ticket s from any U.S. region reported higher prices ( USD 368 and USD 353 respectively) during the first quarter of the calendar year compared to the other three quarters On the other hand, the third quarter reports the lowest prices on one way ( USD 331) and round trip airline ticket s ( USD 314) to the Orlando C SA. Comparing across U.S. regions the Northeast region paid less for one way and round trip airline ticket prices while the West region paid the most in all quarters. Also note that airline pas sengers from the West region paid more (19%) for a one way air line ticket than for a round trip airline ticket to the Orlando CSA compared to the other three U.S. regions Figure 3 42 illustrates the seasonality of average one way (OW) and round trip (RT) airline ticket prices traveling to the Orlando CSA from four U .S. regions during the period between 1993 and 2006. 0 50 100 150 200 250 300 350 400 450 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 99

99 Figure 3 42 Orlando CSA: quarterly s easonal pattern of average one way (OW) and round trip (RT) airline ticket prices for domestic flights from four U.S. r egion s between 19 93 and 2006 Source: Bureau of Transportation Statistics. Tampa St. Petersburg CSA domestic airline ticket prices One way airline ticket prices on flights to the Tampa St. Petersburg CSA decreased 14 % from USD 350 in 1993 to USD 303 in 2006 The highest pr ice was reported in 2000 when the average one way airline ticket price totaled USD 364 Price decreases followed until the year 2005 when prices increased again but did not reach the levels of the year 2000. Over the 14 year period the Midwest region exper ienced the largest decrease in one way airline ticket prices (23%). The Northeast region ranks second with a 19 % decrease in one way airline ticket prices, foll owed by the South (11%) and West (6%) regions The West region paid more for one way airline tic ket s than any other region. Figure 3 4 3 illustrates average one way airline ticket prices from each U.S. region traveling to the Tampa St. Petersburg CSA during the period between 1993 and 2006. 0 100 200 300 400 500 IQ OW IQ RT IIQ OW IIQ RT IIIQ OW IIIQ RT IVQ OW IVQ RT Average Fare in U.S. Dollars Quarter Northeast Midwest All Regions South West

PAGE 100

100 Figure 3 43 Tampa St. Petersbu rg CSA: a verage one way airline ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006 Source: Bureau of Transportation Statistics. Similarly to one way airline ticket prices, round trip airline ticket prices to the Tampa St. Pete rsburg CSA have experienced a decrease (8%) since 1993. Airline passengers traveling from the Northeast region paid 15 % less for round trip airline ticket s in 2006 compared to 1993, accounting for the largest decrease among U.S. regions Similarly, the Wes t and South regions experienced a 7 % decrease in their round trip airline ticket prices to the Tampa St. Petersburg CSA, followed by the Midwest region where price decreased 5 % over the same time period Airline passengers from the West region paid more fo r r ound trip airline ticket s to the Tampa St. Petersburg CSA than any other region Figure 3 4 4 shows average round trip airline ticket prices from each U.S. region traveling to the Tampa St. Petersburg CSA from 1993 to 2006. 0 100 200 300 400 500 600 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 101

101 Figure 3 44 Tampa St. Petersburg CSA: a verage round trip air ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006 Source: Bureau of Transportation Statistics. One way and round trip airline ticket s from any U.S. region reported higher prices ( USD 336 and USD 351 respectively) during the first quarter of the calendar year compared to the other three quarters On the other hand, the third quarter reports the lowest prices on one way ( USD 312) and round trip airline ticket s ( USD 32 4) to the Tampa St. Petersburg CSA. Comparing across U.S. regions, the Northeast regi on paid less for its one way and round trip airline ticket s while the West region p aid the most in all quarters. Also note that airline pas sengers from the West region pai d more (21%) for a one way airline ticket than for a round trip airline ticket to the Tampa St. Petersburg CSA compared to the other three U.S. regions. Figure 3 4 5 shows the seasonality of average one way (OW) and round trip (RT) airline ticket prices tra veling to the Tampa St. Petersburg CSA from four U.S. regions during the period between 1993 and 2006. 0 50 100 150 200 250 300 350 400 450 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 102

102 Figure 3 45 Tampa St. Petersburg CSA: quarterly s easonal pattern of average one way (OW) and round trip (RT) airline ticke t prices for domestic flights from four U.S. r egion s between 1993 and 2006 Source: Bureau of Transportation Statistics. Jacksonville CSA domestic airline ticket prices One way airline ticket prices on flights to the Jacksonville CSA decreased 12 % during t he period between 1993 and 2006 The highest price was reported in 1998 when the average one way airline ticket price totaled USD 371 Over the 14 year period the Midwest region experienced the largest decrease in one way airline ticket prices (20%). The N ortheast region ranks second with a 16 % decrease in one way airline ticket prices, followed by the West region (10%) and South (5%) regions. On average the West region paid more for one way airline ticket s than any other region during the 14 year period. T he Northeast region reported the lowest one way airline ticket price during t he same time period. Figure 3 46 illustrates average one way airline ticket prices from each U.S. region traveling to the Jacksonville CSA during the period between 1993 and 2006. 0 100 200 300 400 500 IQ OW IQ RT IIQ OW IIQ RT IIIQ OW IIIQ RT IVQ OW IVQ RT Average Fare in U.S. Dollars Quarter Northeast Midwest All Regions South West

PAGE 103

103 Figure 3 46 Jacksonville CSA: a verage one way airline ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006. Source: Bureau of Transportation Statistics. Similarly to one way airline ticket prices, round trip airline ticket prices to the Jacksonville CSA have experienced a decrease (10%) since 1993. The highest round trip prices were reported in 1993 when the average ticket price averaged USD 403 for a round trip flight to the Jacksonville CSA. Airli ne passengers traveling from the Midwest region paid 13 % less for round trip airline ticket s in 2006 compared to 1993, accounting for the largest decrease among U.S. regions Similarly, the Northeast region experienced an 11 % decrease in i t s round trip air line ticket prices to the Jacksonville CSA, followed by the West region with a 10 % decrease and the South region with a 6 % decrease over the same time period Airline passengers from th e West region paid more for round trip airline ticket s to the Jacksonvi lle CSA than any other U.S. region as illustrated in Figure 3 4 7 0 100 200 300 400 500 600 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 104

104 Figure 3 47 Jacksonville CSA: a verage round trip air ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006. Source: Bureau of Transp ortation Statistics. On flights to the Jacksonville CSA, one way and round trip airline ticket s from any U.S. region reported higher prices ( USD 352 and USD 369 respectively) during the first quarter of the calendar year compared to the other three quarter s On the other hand, the third quarter reports the lowest prices on one way ( USD 329) and round trip airline ticket s ( USD 341) to the Jacksonville CSA. Comparing across U.S. regions, the Northeast region pa id less for its one way and round trip airline ti cket s while the West region pa id the most in all quarters. Also note that airline passengers from the West region pa id more (21%) for a one way airline ticket than for a round trip airline ticket to the Jacksonville CSA compared to the other three U.S. reg ions Figure 3 4 8 shows the seasonality of average one way (OW) and round trip (RT) airline ticket prices traveling to the Jacksonville CSA from four U.S. regions during the period between 1993 and 2006. 0 100 200 300 400 500 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 105

105 Figure 3 48 Jacksonvi lle CSA: quarterly s easonal pattern of average one way (OW) and round trip (RT) airline ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006. Source: Bureau of Transportation Statistics. Fort Myers CSA domestic airline ticket pri ces One way airline ticket prices on flights to the Fort Myers CSA decreased 4 % during the period between 1993 and 2006 The highest price was reported in 2000 when the average one way airline ticket price totaled USD 380 Over the 14 year period the North east and Midwest regions experienced a decrease in one way airline ticket prices (18% and 11%, respectively) In contrast, the West region reported a 4 % increase in one way airline ticket prices, followed by the South region with a 1 % increase. On average, the West region paid more for one way airline ticket s than any other region during the 14 year period, while the Northeast region reported the lowest one way airline ticket price during the same time period. Figure 3 4 9 illustrates average one way airline ticket prices from each U.S. region traveling to the Fort Myers CSA during the period between 1993 and 2006. 0 100 200 300 400 500 600 IQ OW IQ RT IIQ OW IIQ RT IIIQ OW IIIQ RT IVQ OW IVQ RT Average Fare in U.S. Dollars Quarter Northeast Midwest All Regions South West

PAGE 106

106 Figure 3 49 Fort Myers CSA: a verage one way airline ticket prices for domestic flights from four U.S. r egion s betwe en 1993 and 2006 Source: Bureau of Transportation Statistics. Unlike one way airline ticket prices, round trip airline ticket prices to the Fort Myers CSA have increased 4% since 1993. The highest round trip prices were reported in 2000 when the average t icket price averaged USD 347 for a round trip flight to the Fort Myers CSA. Airline passengers traveling from the West region paid 8 % more for round trip airline ticket s in 2006 compared to 1993, accounting for the largest increase among U.S. regions Simi larly, the Midwest region experienced a 7 % increase in round trip airline ticket prices to the Fort Myers CSA. The Northeast region was the only U.S. region that experience d a decrease in round trip airline ticket prices (1%), while round trip airline tick et price s from the South region in 2006 were the same as the price s reported in 1993 Airline passengers from the West region paid more for round trip airline ticket s to the Fort Myers CSA than any other region Figure 3 50 illustrates average round trip a irline ticket prices from each U.S. region traveling to the Fort Myers CSA from 1993 to 2006. 0 100 200 300 400 500 600 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 107

107 Figure 3 50 Fort Myers CSA: a verage round trip air ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006. Source: Bureau of Transportation Statistics. Unlike the other four destination CSA s the Fort Myers CSA reported higher prices in one way and round trip airline ticket s from any U.S. region during the third quarter of the calendar year compared to the oth er three quarters On the other hand, the first quarter reported the lowest prices on one way ( USD 297) and round trip airline ticket s ( USD 303) to the Fort Myers CSA. Comparing across U.S. regions, the Northeast region pa id less for one way and round trip airline ticket s while the West region pa id the most in all quarters. Also note that airline passengers from the West region pa id on average 19 % more for a one way airline ticket than for a round trip airline ticket to the Fort Myers CSA, followed by the S outh region with 7% Figure 3 5 1 shows the seasonality of average one way (OW) and round trip (RT) airline ticket prices traveling to the Fort Myers CSA from four U.S. regions during the period between 1993 and 2006. 0 50 100 150 200 250 300 350 400 450 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Northeast Midwest South West

PAGE 108

108 Figure 3 51 Fort Myers CSA: quarterly s easonal pattern of average one way (OW) and round trip (RT) airline ticket prices for domestic flights from four U.S. r egion s between 1993 and 2006. Source: Bureau of Transportation Statistics. International Airline Round Tr ip Airline Ticket Prices by World Region Annual average prices of round trip airline ticket s for international flights to Florida have increased 4 % overall since 1995. But airline passengers traveling from Latin America paid 15 % less for round trip airline ticket in 2006 ( USD 342) compared to 1995 ( USD 402) Also, round trip airline ticket prices from Canada have experienced a decrease of 4 % from USD 360 in 1995 to USD 346 in 2006. In contrast, Europe reported a round trip airline ticket price increase of 1 5 % during the same time period Figure 3 5 2 illustrates average round tri p airline ticket prices from three world regions from 1995 to 2006. During the period between 1995 and 2006, round trip airline ticket price s from Canada reported an annual average pr ice of USD 354. February recorded the highest price equivalent to 6 % above the annual average price, while August and December registered the lowest round trip 0 100 200 300 400 500 600 IQ OW IQ RT IIQ OW IIQ RT IIIQ OW IIIQ RT IVQ OW IVQ RT Average Fare in U.S. Dollars Quarter Northeast Midwest All Regions South West

PAGE 109

109 airline ticket price ( USD 338 or 5% below the annual average price). Airline passengers fr om Lat in America paid more for round trip airline ticket s during December ( USD 384 ), equivalent to 4 % above the annual average price ( USD 370). September and October recorded the lowest price from Latin America Europe registered higher prices in June ( USD 971), a 7 % increase fro m the annual average price. Europe paid the lowest price of the year in March ( USD 850), equivalent to a 6 % d ecrease from the annual average Figure 3 5 3 illustrates the seasonality of the relative change of round trip airline ticket pric es traveling to Florida from three world regions from 199 5 to 2006. Figure 3 52 Florida CSA: a verage international round trip airline ticket prices from three world region s between 1995 and 2006. Source: Bureau of Transportat ion Statistics. 0 200 400 600 800 1,000 1,200 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Canada Latin America Europe

PAGE 110

110 Figure 3 53 Florida CSA: monthly s easonal pattern of relative change from the average round trip airline ticket prices from three world region s between 1 995 and 2006. Source: Bureau of Transportation Statistic s. International Airline Round Trip Airline Ticket Prices by Destination CSA The top three destination CSAs (South Florida CSA, Orlando CSA, and Tampa St. Petersburg CSA) in terms of number of international airline passenger s were selected to analyze inter national airline round trip airline ticket prices from 1995 to 2006. Round trip airline ticket s from an international origin traveling to the South Florida CSA were the most expensive ( USD 518) among all three destination CSAs during the 11 year period. Al so, the South Florida CSA experienced an increase of 5 % in airline ticket prices from USD 569 in 1995 to USD 598 in 2006. The Tampa St. Petersburg CSA reported the cheapest international airline ticket ( USD 486). R ound trip airline ticket prices to the Orl ando CSA reported the largest decrease (5%) from USD 582 in 1995 to USD 555 in 2006. The Tampa St. Petersburg CSA ranked second with a 4 % decrease. Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Canada 1% 6% 3% 1% 1% 1% 4% 5% 1% 1% 1% 5% Latin America 2% 2% 3% 0% 1% 0% 1% 2% 4% 4% 1% 4% Europe 1% 0% 6% 5% 0% 7% 6% 1% 4% 0% 2% 5% 8% 6% 4% 2% 0% 2% 4% 6% 8% Relative Change from Average Fare Canada Latin America Europe

PAGE 111

111 An individual analysis of international airline ticket prices for each of the three destination CSAs is pres ented next. It includes a discussion of round trip airline ticket prices on international flights originating from three world regions (Canada, Latin America, and Europe) traveling to a specific destination CSA. South Florida CSA international airline tick et prices Annual average prices of round trip airline ticket s for international flights to the South Florida CSA have increased 5 % overall since 1995. But air line passengers traveling from Latin America paid 14 % less for round trip airline ticket s in 2006 ( USD 341) compared to 1995 ( USD 396) Also, round trip airline ticket prices from Canada have experienced a 6 % decrease from USD 372 in 1995 to USD 348 in 2006. In contrast, Europe reported a round trip airline ticket price increase of 18 % during the same time period Figure 3 5 4 illustrates average round trip airline ticket prices to the South Florida CSA from three world regions from 1995 to 2006. Figure 3 54 South Florida CSA: a verage international round trip airline ticket prices from three world region s between 1995 and 2006 Source: Bureau of Transportation Statistics. 0 200 400 600 800 1,000 1,200 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Canada Latin America Europe

PAGE 112

112 During the period between 1995 and 2006, the round trip airline ticket price from Canada reported an annual average price of USD 366. February recorded the highest price equivalent o f a 4 % increase from the annual average price, while August registered the lowest round trip airline ticket price ( USD 355 or 5 % below the annual average price). Airline passengers from L atin America paid more for round trip airl ine ticket s in March ( USD 380), equivalent to 4 % above the annual average price. September recorded the lo west price for tickets from Latin America. Europe registered higher prices during June ( USD 1,012), a 7 % increase from the annual average price. Europ e paid the lowest price of the year in March ( USD 878). Figure 3 5 5 shows the seasonality of the relative change of round trip airline ticket prices traveling to the South Florida CSA from three world regions between 1995 and 2006. Figure 3 55 S outh Florida CSA: monthly s easonal pattern of relative change from the average round trip airline ticket p rices from three world region s between 1995 and 2006. Source: Bureau of Transportation Statistics. Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Canada 2% 4% 3% 1% 1% 0% 4% 5% 1% 3% 2% 4% Latin America 2% 1% 4% 0% 1% 0% 1% 2% 3% 3% 1% 3% Europe 0% 1% 7% 5% 0% 7% 7% 1% 5% 1% 2% 6% 8% 6% 4% 2% 0% 2% 4% 6% 8% 10% Relative Change from Average Fare Canada Latin America Europe

PAGE 113

113 Orlando CSA international airli ne ticket prices Annual average prices of round trip airline ticket s for international flights to the Orlando CSA decreased 5 % overall during the period between 1995 and 2006. Airline passengers traveling from Latin America paid 43 % less for round trip air line ticket s in 2006 ( USD 294) compared to 1995 ( USD 514) Also, round trip airline ticket prices from Canada have experienced a decrease of 1 % during the same time period. In contrast, Europe reported a round trip airline ticket price increase of 17 % from USD 869 to USD 1,015 in 2006 A verage round trip airline ticket pr ices to the Orlando CSA from three world regions from 1995 to 2006 are presented in Figure 3 5 6 Figure 3 56 Orlando CSA: a verage international round trip air line ticket prices from three world region s between 1995 and 2006 Source: Bureau of Transportation Statistics. Between 1995 and 2006, the round trip airline ticket price from Europe reported an annual average price of USD 867. June recorded the highest pr ice equivalent to a 7 % increase from the annual average price, while March registered the lowest average round trip airline ticket price 0 200 400 600 800 1,000 1,200 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Canada Latin America Europe

PAGE 114

114 ( USD 809 or 7% below the annual average price). Airline passengers from La tin America paid more for round trip airline ticket s in February ( USD 491), equivalent to 15 % above than the annual average price. September recorded the lowest average pri ce from Latin America Canada registered higher prices during February ( USD 365), an 8 % increase from the annual average price of USD 337. Canada paid the lowest price of the year in August ( USD 317). Monthly seasonal pattern s of the relative change of round trip airline ticket prices trave ling to the Orlando CSA from three world regions between 1995 a nd 2006 are shown in Figure 3 5 7 Figure 3 57 Orlando CSA: monthly s easonal pattern of relative change from the average round trip airline ticket prices from three world region s between 1995 and 2006. Source: Bureau of Transportation Statistics. Tampa St. Petersburg CSA international airline ticket prices Annual average prices of round trip airline ticket s for international flights to the Tampa St. Petersburg CSA decreased 3% overall during the period between 1995 and 200 6 Airline passengers traveling from Latin America paid 8 % less for round trip airline ticket s in 200 5 ( USD Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Canada 1% 8% 5% 1% 2% 1% 3% 6% 0% 1% 3% 4% Latin America 2% 15% 7% 1% 4% 4% 1% 7% 11% 10% 1% 2% Europe 1% 0% 7% 5% 0% 7% 6% 1% 4% 1% 2% 5% 15% 10% 5% 0% 5% 10% 15% 20% Relative Change from Average Fare Canada Latin America Europe

PAGE 115

115 344) compared to 1995 ( USD 373) Also, round trip airline ticket prices from Canada have experienced a decrease of 4 % between 1995 and 2006. In contrast, Europe reported a round trip a irline ticket price increase of 12 % from USD 932 to USD 1 041 in 2006 A verage round trip airline ticket prices to the Tampa St. Petersburg CSA from three world regions from 1995 to 2006 are shown in Figure 3 5 8 Note that average round trip price s for Lat in America were not available in 2006. Figure 3 58 Tampa St. Petersburg CSA: a verage international round trip airline ticket prices from three world region s between 1995 and 2006 Source: Bureau of Transportation Statistics. B etween 1995 and 2006, round trip airline ticket s from Europe reported an annual average price of USD 926 June recorded the highest price equivalent to a 7 % increase from the annual average price, while March registered the lowest average round trip airlin e ticket price ( USD 864 or 7% below the annual average price). Airline passengers from La tin America paid more for round trip airline ticket s in July ( USD 514), equivalent to 54 % above the annual average price ( USD 333). September recorded the lowest avera ge price from Latin America. Canada registered 0 200 400 600 800 1,000 1,200 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Fare in US Dollars Year Canada Latin America Europe

PAGE 116

116 higher prices during April ( USD 352), a 7 % increase from the annual average price of USD 329. Canada paid the lowest price of the year in July ( USD 301). Monthly seasonal patterns of the relative change of rou nd trip airline ticket prices trave ling to the Tampa St. Petersburg CSA from three world regions between 199 5 and 2006 is presented in Figure 3 59 Seasonal values for Latin America were not available for January and February. Figure 3 59 Tampa St. Petersburg CSA: monthly s easonal pattern o f the relative change from the average round trip airline ticket price s from three world region s between 1995 and 2006. Source: Bureau of Transportation Statistics. Freight and Mail Transpor ted v ia Commercial Passenger Airlines to Florida Domestic freight transportation 2 to the Florida CSA accounted for 3 % of the total freight traffic transported via commercial passenger airlines in the United States. Domestic freight to the Florida CSA via c ommercial passenger airlines has increased nearly 486 % from 178.5 million 2 Includes freight transported by commercial passenger airlines only. Freight transported by cargo airlines (e.g., FEDEX ) is not included. Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Canada 0% 6% 4% 7% 6% 1% 8% 5% 0% 3% 5% 3% Latin America 0 0 11% 2% 9% 0% 54% 14% 15% 8% 15% 10% Europe 0% 1% 7% 6% 0% 7% 6% 1% 4% 1% 2% 6% 20% 10% 0% 10% 20% 30% 40% 50% 60% Relative Change from Average Fare Canada Latin America Europe

PAGE 117

117 pounds in 1990 to 1.045 billion pounds in 2006. Approximately 63 % of the more than 6.44 billion pounds of freight were transported from the South region to the Florida CSA during thi s time period. The Midwest region ranked second with a 15 % share, followed by the Northeast (13%) and West (9%) regions The South and Midwest regions experienced the largest increases in total domestic freight transported to the Florida CSA (782% and 647% respectively). Meanwhile, the Northeast region almost double d its freight sent to the Florida CSA while the West region experienced a 60 % increase during the same time period Figure 3 60 shows total domestic freight transported to the Florida CSA betwee n 1990 and 2006. Similar to domestic airline passengers, domestic freight from three of the four U.S. regions was transported more frequently during March than any other month between 1990 and 2006 The Northeast, Midwest, and South region s transported ap proximately 10 % of total freight during that month. The West region recorded its highest level during December with 10 % of its total annual freight transported to the Florida CSA July registered the lowest level of domestic freight transported from the M idwest, Northeast, and West regions while December registered the lowest level for freight originated from the South region. Figure 3 6 1 presents the seasonal pattern of total freight transported by commercial passenger airlines to the Florida CSA per mon th, by U.S. region from 1990 to 2006.

PAGE 118

118 Figure 3 60 Florida CSA: t otal domestic freight transported from four U.S. r egion s using commercial passenger airlines between 1990 and 2006 Source: Bureau of Transportation Statistics. Figure 3 61 Florida CSA: monthly s easonal pattern of total freight transported from four U.S. r egion s using commercial passenger airlines between 1990 and 2006 Source: Bureau of Transportation Statistics. 0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Freight in Thousands (Lbs) Year Other West Northeast Midwest South Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec West 8% 8% 9% 8% 8% 7% 7% 8% 8% 9% 9% 10% Midwest 8% 8% 10% 9% 9% 7% 7% 8% 8% 9% 9% 9% Northeast 9% 9% 9% 9% 9% 7% 7% 8% 7% 9% 8% 9% South 8% 8% 10% 9% 9% 8% 8% 8% 8% 8% 8% 7% 6% 7% 7% 8% 8% 9% 9% 10% 10% Percentage of Total Pounds West Midwest Northeast South

PAGE 119

119 The South Florida C SA received nearly 4.0 billion pounds of domestic freight transported to the Florida CSA equivalent to 58 % of the total share. The Orlando CSA ranked second with 23 % followed by the Tampa St. Petersburg CSA with 10 % of the total domestic freight transport ed to the Florida CSA The top five destination CSAs experienced large increases in domestic freight traffic of which Jacksonville recorded the largest increase ( 814 % ) during the 17 year period. T otal domestic freight transported to each of the six destina tion CSAs in Florida is presented in Figure 3 6 2 Figure 3 62 Total d omestic freight transported to six destination CSAs in Florida using commercial passenger airlines between 1990 and 2006. Source: Bureau of Transportation S tatistics. Unlike domestic freight transportation, domestic mail transportation 3 to the Florida CSA via commercial passenger airlines has decreased approximately 58 % during the period between 1990 and 2006 Approximately 42 % of the more than 2.45 billion p ounds of mail sent to the 3 Inclu des mail transported by commercial passenger airlines only. Mail transported by cargo airlines (e.g., FEDEX ) is not included. 0 100,000 200,000 300,000 400,000 500,000 600,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Freight in Thousands (Lbs) Year Other Fort Myers Jacksonville Tampa St. Petersburg Orlando South Florida

PAGE 120

120 Florida CSA originated from the South region during this time period. The Northeast region ranked second with a 25 % share, followed by the Midwest (22%) and West (10%) regions The South and Midwest regions experienced the largest decreases in total domestic mail transported to the Florida CSA (74% and 66% respectively). Mail sent from the Northeast region decreased 43 % from 43.4 million pounds in 1990 to 24.6 million pounds in 2006, while the West region experienced a 14 % decrease during th e same time period Figure 3 63 presents total domestic mail transported via commercial passenger airlines to the Florida CSA between 1990 and 2006. Figure 3 63 Florida CSA: t otal do mestic mail transported from fou r U.S. r egion s using commercial passenger airlines between 1990 and 2006. Source: Bureau of Transportation Statistics. Domestic mail via commercial passenger airlines from the four major U.S. regions was transported more frequently during December than any other month between 1990 and 2006 The Northeast, Midwest, and South region s transported approximately 11 % of total mail during that month. The West region recorded its highest level during December with 11.5 % of its total 0 20,000 40,000 60,000 80,000 100,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Mail in Thousands of Pounds Year Other West Midwest Northeast South

PAGE 121

121 annual mail transported to the F lorida CSA September registered the lowest level of domestic mail transported from four major U.S. regions. Figure 3 6 4 illustrates the monthly seasonal pattern of domestic mail transported to the Florida CSA from 1990 to 2006. Figure 3 64 Florida CSA: monthly s easonal pattern of total mail transported from four U.S. r egion s using commercial passenger airlines between 1990 and 2006 Source: Bureau of Transportation Statistics. The South Florida CSA received nearly 1.11 billio n pounds of domestic mail which represented 45 % of total mail transported to the Florida CSA The Tampa St. Petersburg CSA ranked second with 22 % followed by the Orlando CSA (18%), Jacksonville CSA (13%), and Fort Myers CSA (1%). Four of the top five dest ination CSAs experienced decreases in domestic mail traffic of which Jacksonville recorded the largest drop ( 89 % ) during the 17 year period. In contrast, mail transported to the Fort Myers CSA increased 15 % du ring the same time period. T otal domestic m ail transported to each of the six destination CSAs in Florida is shown in Figure 3 6 5 Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec West 9% 8% 9% 8% 8% 8% 7% 8% 7% 8% 8% 12% Midwest 10% 9% 9% 9% 8% 7% 7% 7% 7% 8% 8% 10% Northeast 10% 9% 10% 9% 8% 7% 7% 7% 7% 8% 8% 11% South 10% 8% 9% 8% 8% 8% 8% 8% 7% 8% 8% 11% 6% 7% 8% 9% 10% 11% 12% Percentage of Total Pounds West Midwest Northeast South

PAGE 122

122 Figure 3 65 Total domestic mail transported to six destination CSAs in Florida using commercial passenger airlines between 1990 and 2006 Sou rce: Bureau of Transportation Statistics. International freight transportation to the Florida CSA using commercial passenger airlines has increased approximately 206 % during the period between 1990 and 2006 Nearly 88 % of the more than 23.72 billion pounds of freight were transported from Latin America to the Florida CSA during this time period. Europe ranked second with a 10 % than half a percent. Europe experienced an increase in total freight transported to the Florida CSA f rom 52.88 million pounds in 1990 to 164.95 million pounds in 2006. Latin America also grew 196 % during the 17 year period, as presented in Figure 3 6 6 International freight from Canada and Latin America was transported more frequently during February than any other month between the year 1990 and 2006. Europe recorded its highest levels during August when it transported 13 % of its total annual freight to the Florida CSA Europe registered the lowest level of international freight transported (7.2%) in Janu ary, as shown in Figure 3 6 7 0 20,000 40,000 60,000 80,000 100,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Freight in Thousands (Lbs) Year Other Fort Myers Jacksonville Orlando Tampa St. Petersburg South Florida

PAGE 123

123 Figure 3 66 Florida CSA: t otal international freight transported from three w orld r egion s using commercial passenger airlines between 1990 and 2006. Source: Bureau of Transportation Statistics. Figure 3 67 Florida CSA: monthly s easonal pattern of total freight transported from three world regions using commercial passenger airlines between 1990 and 2006. Source: Bureau of Transportation Statistics. Between 1990 and 2 006 the South Florida CSA received approximately 23.0 billion pounds of international freight which represents 97 % of total international freight transported to 0 500 1000 1500 2000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Freight in Thousands (Lbs) Year Canada Other Europe Latin America Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Canada 9% 11% 11% 9% 8% 7% 6% 6% 5% 8% 11% 11% Europe 7% 8% 8% 7% 8% 8% 8% 13% 8% 9% 9% 8% Latin America 9% 9% 9% 9% 8% 7% 8% 8% 8% 9% 8% 9% 4% 5% 6% 7% 8% 9% 10% 11% 12% 13% 14% Percentage of Total Freight Canada Europe Latin America

PAGE 124

124 the Florida CSA The Orlando CSA ranked second with less than 3 % The other three destination CSAs accounted for less than a quarter percent during the same time period. The South Florida CSA reported an increase of 203 % in international freight traffic T otal international freight transported to each of the six destination CSAs in Florida is shown in Figure 3 6 8 Figure 3 6 8 Total international freight transported to six destination CSA s in F lorida using commercial passenger airlines between 1990 and 2006. Source: Bureau of Transportation Statistics. Unlike internatio nal freight transportation, international mail transported to the Florida CSA using commercial passenger airlines has decreased approximately 65 % during the period between 1990 and 2006 In 1990 more than 16 million pounds of mail were transported to the F lorida CSA but by 2006 the total had decreased to only 5.53 million pounds. Nearly 54 % of the more than 109 million pounds of mail w as transported from Latin America to the Florida CSA during this time period. Europe ranked second with a 35 % share, followe d by Canada with a share of 11 % Both Latin America and Europe experienced a decrease of approximately 75 % in total mail transported to the Florida CSA In contrast, Canada has reported large gains in mail 0 500 1,000 1,500 2,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Freight in Millions (Lbs) Year Other Fort Myers Jacksonville Orlando Tampa St. Petersburg South Florida

PAGE 125

125 transported to the Florida CSA from 0.1 million po unds in 1990 to more tha n 1.4 million pounds in 2006. T otal international mail transported to the Florida CSA from four world regions between 1990 and 2006 is illustrat ed in Figure 3 69 Figure 3 69 Florida CSA: t otal interna tional mail transported from four world regions using commercial passenger airlines between 1990 and 2006. Source: Bureau of Transportation Statistics. International mail from Canada and Europe was transported more frequently during December than any other month between the year 1990 and 2006 Latin America recorded its highest levels during the month of May when it transported approximately 12 % of total annual mail to the Florida CSA Europe registered the lowest level of international mail transported (7. 2%) in September. Canada recorded its lowest levels during July (6.3%), while Latin America did so in October (6.2%) Figure 3 70 illustrates the monthly seasonal pattern of domestic mail transported to the Florida CSA from 1990 to 2006. 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Mail in Million of Pounds Year Canada Other Europe Latin America

PAGE 126

126 Figure 3 70 Florida CSA: monthly s easonal pattern of total mail transported from three world regions u sing commercial passenger airlines between 1990 and 2006. Source: Bureau of Transportation Statistics. Between 1990 and 2006 the South Flori da CSA received approximately 106.2 million pounds of international mail which represents 97 % of total international mail transported to the Florida CSA The Orlando CSA ranked second with 2 % The other three destination CSAs accounted for less than a perc ent during the same time period. All destination CSAs reported decreases in total international mail transported since 1990. The South Florida CSA reported a decrease of 64 % from 15.7 million pounds in 1990 to 5 .6 million pounds in 2006. T otal internationa l mail transported to each of the six destination CSAs in Flo rida is presented in Figure 3 71 Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Canada 10.7% 10.7% 8.9% 7.5% 6.8% 6.4% 6.3% 6.5% 6.3% 7.0% 10.0% 13.0% Europe 7.9% 7.4% 8.5% 7.9% 8.5% 8.5% 8.3% 7.3% 7.3% 7.6% 9.0% 12.0% Latin America 8.9% 8.9% 9.1% 8.8% 12.2% 9.5% 8.4% 6.8% 6.4% 6.2% 6.6% 8.4% 4% 5% 6% 7% 8% 9% 10% 11% 12% 13% 14% Percentage of Total Mail Canada Europe Latin America

PAGE 127

127 Figure 3 71 Total international mail transported to six destination CSAs in Florida using commercial passenger airlines between 19 90 and 2006. Source: Bureau of Transportation Statistics. Economic, Social, and Weather Indicators Additionally, the following statistics were included in this study: annual gross domestic product, per capita personal disposable income and population coll ected from every state by the U.S. Bureau of Economic Analysis; brand advertising expenditures from selected private firms and generic advertising expenditures from Florida government entities collected by TNS Media Intelli gence; monthly statistics on aver age monthly prices of kerosene type jet fuel collected b y the U.S. Department of Energy; Law Enforcement. Gross Domestic Product, Personal Disposable Income and Population A nnual g ross dome stic product (GDP), per capita personal disposable income and population estimates for each state and the District of Columbia were retrieved from the online 0 20,000 40,000 60,000 80,000 100,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Freight in Thousands (Lbs) Year Other Fort Myers Jacksonville Orlando Tampa St. Petersburg South Florida

PAGE 128

128 database provided by Bureau of Economic Analysis through its Regional Economic Analysis Division. R etrieved data covered the period between 1990 and 2006. S tates were then aggregated according to the U.S. Census geographic regions scheme as shown previously in Table 3 1. Annual GDP, P DI PC and population totals calculated for each U.S. region a re disc ussed next. In 2006 GDP for the four U.S. regions was reported at USD 12.58 trillion, which represents a 126 % increase from 1990 when GDP totaled USD 5.57 trillion. The Northeast region accounted for the largest share among the four U.S. regions, with nea rly 30 % of total GDP. The Midwest region followed with 26.68 % The South (21.90%) and West (21.88%) regions ranked third and fourth respectively. Among all U.S. regions, the South region experienced the largest increase (150%) in GDP during the 17 year pe riod. The West region ranked second with a 142 % increase, followed by the Midwest (116%) and Northeast (107%) region s T otal GDP by U.S. region from 1990 to 2006 is presented in Figure 3 7 2 During the period between 1990 and 2006, per capita personal disp osable income ( PC PDI) averaged USD 23,195 for the four U.S. regions. Also, the PDI PC, reported at USD 31,568 in 2006, increased 90 % during the same time period. Among the four U.S. regions, the Northeast region is the only region with a PDI PC above the national average. Its PDI PC amounted to USD 25,746, equivalent to 11 % above the national average. In regards to PDI PC growth, the South region experienced the largest increase (95%) during the 17 year period. The Midwest region ranked second with a 91 % i ncrease in its PDI PC, followed by the We st (88%) and Northeast (86%) regions A verage PDI PC for each U.S. region from 1990 to 2006 is presented in Figure 3 7 3 T otal population for four U.S. regions was estimated at 282.29 million in 2006, which represen ts a 17 % increase since 1990 when the population was estimated at 240.94 million. The

PAGE 129

129 Midwest region accounted for the largest population share among the four U.S. regions, with nearly 28 % The Northeast region followed with 27 % The South (23%) and West ( 22%) region s ranked third and fourth respectively. Among all U.S. regions, the West region experienced the largest increase (31%) in population from approximately 49.03 million in 1990 to 64.02 million in 2006. The South region ranked second with a 25 % in crease, followed by the Midw est (11%) and Northeast (8%) regions Figure 3 7 4 presents population estimates from each U.S. region from 1990 to 2006. Figure 3 72 Annual gross domestic product from four U.S. r egions between 199 0 and 2006 Source: Bureau of Economic Analysis Regional Economic Analysis Division. 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Billions of U.S. Dollars Year South West Midwest Northeast

PAGE 130

130 Figure 3 73 Annual per capita personal disposable income from four U.S. regio n s between 1990 and 2006. Source: U.S. Department of Commerce Bureau of Economic Analysis. Figure 3 74 Annual p opulation estimates from four U.S. r egion s between 1990 and 2006. Source: U.S. Department of Commerce Bureau of Economic Analysis. 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 U.S. Dollars Year South West Midwest Northeast 0 10 20 30 40 50 60 70 80 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Population in Millions Year West South Northeast Midwest

PAGE 131

131 Brand and Generic Advertising Expenditures Description, s election, and a ggregation of a dvertising expenditures d ata Data used to analyze brand and generic advertising expenditures were collected by ( Taylor Nelson Sofres ) TNS Media Intelligence and are available online through subscription. B ra n d ad vertising expenditure data used in this research include the perio d between 1995 and 2006, while generic adv ertising expenditure data cover the period between 2002 and 2006. Tourism promotion accounts of four major tourist companies were selected to analyz e brand advertising efforts conducted to attract visitors to Florida. Busch Entertainment Corporation Walt Disney Company Carnival Cruise Lines and NBC Universal have tourist attractions in Florida, as well as in other locations in the United States and abroad. These companies have several accounts related to the tourism promotion efforts conducted in the United States. For the purpose of this research, these accounts were divided into three major categories: Florida, Combined, and Non Florida. The Florida category denotes all accounts specifically stating that their budget was committed to advertise its Florida based attr actions only. For Florida category of Walt Disney Company includes advertising expenditures related to attractio ns in Florida such as Walt Disney World Disney C ruises and Disney Hotels accounts stating their budget was used to advertise attractions in Florida and elsewhere. category of Walt Disney Co mpany includes those accounts related to general advertising expenditures for attraction s in Florida and California. Finally, all other accounts that advertise attractions outside the state of Florida were included in the Non Florida category. For examp le, the Non Florida category

PAGE 132

132 of Walt Disney Company includes only those accounts related to advertising expenditures for attraction s located outside of Florida such as Disneyland in California As for generic advertising expenditures, tourism promotio n accounts from Florida cities, Florida counties and the state of Florida were selected from the year 2002 to 2006 The category includes all tourism promotion efforts conducted by cities (e.g., Orlando, Miami, Tampa) in Florida to promote their re spective city tourism promotion efforts conducted by counties (e.g., Dade, Hillsborough, Alachua) in Florida to promote their respe ctive count y. F State category includes all accounts in which promotional ef forts were conducted to promote the state of Florida as a whole. Brand a dvertising e xpenditures More than USD 2.63 billion were spent by four major tourist companies with businesses in Florida during the period between 1995 and 2006. On average these four companies spent approximately USD 220 million per year during the 12 year period. Tourism promotion efforts to advertise their attractions in Florida accounted for 44 % and the combined effort accounted for 53 % of total advertising expenditures during the 12 year period. Walt Disney Company spent more dollars ( USD 1.29 billion) in advertising than the other three companies, equivalent to 49 % of total brand advertising expenditures. Carnival Cruise Lines ranked second (25%), followed by Busch Entertainment Group (18%), and NBC Universal (9%). Walt Disney Company spent nearly USD 994.94 million to promote attractions in Florida, which represented approximately 77 % of total advertising expenditures during the 12 year period. NBC Universal was a distant se cond spending USD 128.7 million in efforts to promote attractions in Florida. Still, it represented more than one half of its advertising expenditure budget. Note that some companies advertise their Florida and non Florida attractions in the same advertisi ng spot. For example, Busch Entertainment Group usually combined the

PAGE 133

133 three water parks in the same advertising spot. In fact, 91 % of the advertising expenditures of Busch Entertainment Group fall under the Note that all accounts rela ted to tourism promotion s from Carnival Cruise Lines Comb T otal brand advertising expenditures spent by these four companies are presented in Figure 3 7 5 Figure 3 75 Total brand advertising e xpenditures from four tourism related companies in Florida between 199 5 and 2006. Source: TNS Media Intelligence. During the period between 1995 and 2006 Busch Entertainment Group NBC Universal and Walt Disney Company made most of their advertising ef forts to promote their attractions in Florida during the first five months of the calendar year. NBC Universal spent two thirds of its total advertising budget to promote Florida attractions between January and May, while Busch Entertainment Group and Wa lt Disney Company spent approximately 62 % of their budget during those five months Advertising expenditures decreased later in the calendar year. For example, Busch Entertainment Group allotted 1 % of its advertising budget to November, while NBC Univers al assigned 3 % July, August, and September registered the lowest levels of Bush Entertainment Group NBC Universal Inc. Walt Disney Company Carnival Cruise Lines Florida 29,415,200 128,684,900 994,937,900 Combined 422,623,600 88,127,000 234,317,100 649,163,700 Non Florida 13,802,800 9,746,100 68,341,500 200 400 600 800 1,000 1,200 Millions of U.S. Dollars Florida Combined Non Florida

PAGE 134

134 advertising expenditures by Walt Disney Co mpany Figure 3 76 illustrates monthly seasonal pattern s of total brand adverting expenditures of Busch E ntertainment Group NBC Univer sal and Walt Disney Company fro m 1995 to 2006 Figure 3 76 Florida attractions: monthly s easonal pattern of total brand adverting expenditures from three tourism related companies between 1995 and 2006. Source: TNS Media I ntelligence. During the period between 1995 and 2006 Busch Entertainment Group NBC Universal and Carnival Cruise Lines made most of their advertising efforts to promote their combined attractions in Florida and elsewhere during the first six months of the calendar year. Carnival Cruise Lines spent 65 % of its advertising budget to promote cruise lines between January and June and advertised most heavily in January (16%) and February (12%). NBC Universal and Busch Entertainment Group showed similar p atterns in terms of advertising expenditures throughout the calendar year between 1995 and 2006. NBC Universal spent 65 % of its total advertising budget to jointly promote Florida and Non Florida attractions between January and June. It spent more heavily in April (19%) and March (16%). Busch Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Busch E. Group 15% 14% 15% 12% 7% 8% 12% 6% 5% 4% 1% 2% NBC Universal 9% 15% 14% 16% 14% 6% 5% 5% 5% 5% 3% 4% Walt Disney Co. 14% 13% 12% 11% 11% 4% 3% 3% 3% 8% 9% 8% 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% Brand: Florida Category Percentage of Total Expenditutes Busch E. Group NBC Universal Walt Disney Co.

PAGE 135

135 Entertainment Group spent more advertising dollars in June (17%) than any other month, followed by May (15%) and April (13%). In total, it spent almost 70 % of its budget during the first semester. Advertising expendi tures decreased later in the calendar year for both companies. Busch Entertainment Group allotted 3 % of its advertising budget to the each of the last four months of the year, while NBC Universal assigned no more than 7 % Despite spending more than 55 % of its budget during the first semester, Walt Disney Company also spent a large amount of dollars to jointly advertis e attractions in Florida and California between September and November (33%). Walt Disney Company also spent heavily during January (17% of its advertising budget), more than the NBC Universal and Busch Entertainment Group combined. Advertising expenditures were reduced during December (3%), July (4%), and June (5%) as shown in Figure 3 77. Figure 3 77 Flor ida and non Florida attractions: monthly s easonal pattern of total brand adverting expenditures from three tourism related companies between 1995 and 2006. Source: TNS Media Intelligence. 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% 20% Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Brand: Combined Category Percentage of Total Expenditutes Month Busch E. Group NBC Universal Walt Disney Co. Carnival Cruises

PAGE 136

136 Generic a dvertising e xpenditures Cities, counties, and the state of Florida spent more than USD 339 million in promotional efforts during the period between 2002 and 2006. On average these three entities combined spent approximately USD 67.84 million per year during the 5 year period. T ourism promotion efforts conducted by the state to advertise state attractions accounted for 63 % of total generic advertising expenditures. City governments ranked second with 15 % followed by counties with 15 % of total generic advertising expenditures during the 5 year period. Since 2002 gen eric advertising expenditures from the state increased 134 % while the cities and advertising decreased 23 % and advertising expenditures from counties decreased 6 % during the 5 year period. T otal generic advertising expenditures by government entities in Florida from 2002 to 2006 are presented in Figure 3 78. Figure 3 78 Total generic advertising expenditures spent by government entities from Flor ida between 2 002 and 2006. Source: TNS Media Intelligence. Advertising efforts conducted by city, county and state government in Florida during the first six months of the calendar year accounted for 67 % of total advertising expenditures. City governments throughout Florida spent most heavily in May, 18 % of their advertising budget, to 0.0 10.0 20.0 30.0 40.0 50.0 60.0 2002 2003 2004 2005 2006 Millions of U.S. Dollars Year County City State

PAGE 137

137 promote their city attractions during May. April ranked second with 12 % followed by March (11%). Advertising ex penditures were reduced during December (3%) and September (4 %). County governments in Florida spent more heavily in April (22%), March (19%), and May (16%) to advertise their county attractions. In total, they spent almost 70 % of their budget during the first semester. Advertising expenditures decreased later in th e calendar year, when county governments assigned less than 4 % of their advertising budget to each of the last three months of the year. Finally, the state government spent more heavily during May (14% of its advertising budget). The s tate government assig ned similar amounts of advertising dollars (approximately 6% to each month) during the second semester of the calendar year. Figure 3 7 9 presents monthly seasonal pattern s of total generic advertisi ng expenditures for each of the three government entities between 2002 and 2006. Figure 3 79 Monthly s easonal pattern of total generic adverting expenditures spent by city, county, and state government to promote Florida between 2002 and 2006. Source: TNS Media Intelligence. Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec County 3% 3% 19% 22% 16% 7% 6% 6% 10% 4% 3% 2% City 8% 10% 11% 12% 18% 10% 7% 5% 4% 6% 6% 3% State 9% 8% 7% 10% 14% 12% 6% 6% 6% 6% 8% 6% 0% 5% 10% 15% 20% 25% Generic Advertising Percentage of Total Expenditutes County City State

PAGE 138

138 Foreign Exchange Rate: Euro to U.S. Dollar Foreign exchange rate data w ere retrieved from the Federal Reserve through its Statistical Release published on its W eb site. Monthly rates from January 1990 to December 2006 were obtained for the Euro, official currency of the European Union. This foreign exchange rate, Euro to U.S. dollar, was used to describe economic conditions in Europe and its relationship to airline passenger travel to Florida. An exchange rate level over 1.00 denotes a weak U.S. dollar and airline tickets to Florida are more affordable to Europeans. The U.S. dollar has weakened 2 % and 2003 when it was stronger than the Euro, but it has grown weaker since 2004. Figure 3 80 presents annual air passenger traffic from Europe to Florida and the annual exchange rate average (Euro to USD ) from 1990 to 2007. Note that an exchange rate level over 1.00 denotes a weak U S dollar. Figure 3 80 A nnual air passenger traffic traveling from Europe to Florida and annual exchange rate (Euro to USD ) between 1990 and 2006. Source: Bureau of Transportation Statistics and Federal Reserve. 0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50 0 500 1,000 1,500 2,000 2,500 3,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Exchange Rate ( to USD) Number of Passengers (in 1,0000) Year Europe Exchange rate

PAGE 139

139 Historic Jet Fuel Prices Data on monthly kerosene type jet fuel prices for three different stations New York Harbor, Gulf Coast, and Los Angeles in the United States ; and two international stations Rotterdam and Singapore were retrieved from U.S. Department of Energy Energy Information Administration and are a vailable online. Retrieved d ata covered the period between 1990 and 2006. Kerosene type jet fuel prices have increased 106 % during the period between 1990 and 2006. Over the 17 year period jet fuel prices have followed a similar ly increasing pattern in the three U.S. stations and the two international stations. K erosene type j et fuel prices in the New York Harbor and the Gulf Coast stations have increased 114 % while prices in Los Angeles recorded a 93 % increase. Similarly, kerosene type jet fuel prices in the two international stations have increased. Figure 3 8 1 shows the annual average price of kerosene type jet fuel in five stations worldwide from 1990 to 2006. Figure 3 81 Historic average kerosene type jet fuel prices from five worldwide locations between 1990 and 2006. Source: Department of Energy. 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 New York Harbor 91 64 59 56 53 52 66 59 43 52 91 74 71 87 119 171 195 U.S. Gulf Coast 90 61 57 53 49 49 61 56 40 50 85 72 69 83 115 171 192 Los Angeles 106 63 60 60 54 57 66 63 45 58 94 77 73 89 128 174 203 Rotterdam 97 67 59 55 50 51 64 58 41 52 88 74 70 86 120 169 194 Singapore 102 67 61 58 52 54 67 59 39 51 82 68 67 78 113 161 192 0.00 50.00 100.00 150.00 200.00 250.00 Price in USD (Cents per gallon) New York Harbor U.S. Gulf Coast Los Angeles Rotterdam Singapore

PAGE 140

140 Between 1990 and 2006, kerosene type jet fuel prices reported an annual average price of USD 0.83 per gallon. October recorded the highest price equivalent to an 11 % increase fr om the annual average price for all five locations, while January registered the lowest jet fuel price ( USD 0.83 or 7% below the annual average price). Monthly seasonal patterns of the relative change of kerosene type jet fuel prices from five locations w o rldwide between 1990 and 2006 are presented in Figure 3 8 2 Figure 3 82 Monthly s easonal pattern of historic average kerosene type jet fuel prices from five worldwide locations between 1990 and 2006. Source: Department of En ergy. Hurricanes and Wildfires A ffecting Florida Hurricane data w ere retrieved from the National Oceanic and Atmospheric Administration (NOAA) and the National Weather Service Archive of Hurricane Seasons which are available online. Only tropical storms th at a ffected Florida were selected. Tropical s torms were cate gorized from 1 to 7 to denote strength in wind speed. C ategories 1 5 were assigned to each Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec New York Harbor 6% 6% 6% 4% 3% 6% 1% 3% 8% 10% 6% 3% U.S. Gulf Coast 7% 8% 7% 4% 3% 5% 0% 4% 9% 13% 6% 2% Los Angeles 8% 7% 6% 2% 3% 3% 1% 5% 7% 9% 7% 2% Rotterdam 8% 7% 6% 4% 4% 5% 0% 3% 8% 11% 8% 4% Singapore 6% 6% 7% 4% 4% 6% 1% 3% 7% 11% 8% 5% 10% 5% 0% 5% 10% 15% Relative Change from Average New York Harbor U.S. Gulf Coast Los Angeles Rotterdam Singapore

PAGE 141

141 storm following the Saffir Simpson Hurricane Scale, category 6 for a tropical storm, and category 7 for a tropical depression. Forty nine storms affected Florida during the period between 1990 and 2006, which represents an annual average of less than three storms during the 17 year period. More than one half of the storms (57%) were categorized as tropical s torms and less than 30 % of the storms have been categorized as hurricanes. The year 2005 was the busiest in terms of number of storms affecting Florida. Five tropical storms, two category 3 hurricanes and one category 1 hurricane affected the peninsula th is year Figure 3 83 presents total number of storms that affected Florida from 1990 to 20 06 Figure 3 83 Number of tropical s torms (by category) affecting Florida between 1990 and 2006 Source: National Oceanic and Atmospher ic Administration. Data on the number of wildfires including total of acres burned across Florida were used in this study as reported by the Florida Department of Agriculture and Consumer Services Division of Forestry. Daily data retrieved from January 19 90 to December 2006, were then aggregated into monthly data. Also, wildfires were grouped in to a specific category according to the size of 0 1 2 3 4 5 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Number of Storms Year Cat 1 Cat 2 Cat 3 Cat 4 Cat 5 Trop Dep Trop Storm

PAGE 142

142 the wildfire (number of acres burned) and is described as follows : c ategory 1 includes all wildfires that burn betwe en 0.10 and 0.29 acres; category 2 represents wildfires where 0.30 to 9.99 acres were burned; category 3 represents 10.00 to 99.99 acres burned; category 4 represents 100.00 to 299.99 acres burned; category 5 represents 300.00 to 999.99 acres burned; categ ory 6 represents 1 000.00 to 4 999.99 acres burned; and category 7 represents 5 000.00 or more acres burned. More than 73,000 wildfires have affected Florida burning nearly 2.95 million acres during the period between 1990 and 2006. The number of wildfire s decreased 26 % from 6,526 total wildfires in 1990 to 4,802 total wildfires in 2006. But the number of acres burned has not decreased as much as the number of wildfires. Despite the decrease in the number of wildfires, the acres burned have only decreased 9% Approximately 60 % of the wildfires fell in the category 2, followed by category 1 (23%), and category 3 (18%). In other words, more than 97 % of all wildfires were no larger than 100 acres. More acres were burned in 1998 than any other year during the 1 7 year period. Nearly 5,000 wildfires were responsible for burning more than 0.5 million acres in 1998. Note that the year 2005 registered 2,263 total wildfires (only 3% of total ) that burned less than 1% of total acres burned during the 17 year period. C oincidentally, it was the busiest year in terms of storms affecting Florida as described previously. Figure 3 8 4 presents total number of wildfires by category affec ting Florida from 1990 to 2006. Most of the wildfires (86%) occurred during the first six m onths of the calendar year. Also, the first six months registered approximately 72 % of the total acres burned. May recorded the largest number of wildfires with 10,739 that burned more than 643,000 acres, followed by February (13%) with 373,500 acres and M arch (13%) with 264,600 acres. June was the most devastating in terms of number of acres

PAGE 143

143 burned. More than one in every four acres burned (nearly 824,000 acres) during the 17 year period were recorded during June. In contrast, October registered the fewest number of acres burned (less than 1%) during t he same time period. Figure 3 85 shows the percentage of wildfires and acreage burned in Florida from 1990 to 2006. Figure 3 84 N umber of wildfires ( by category ) affecting Flori da between 1990 and 2006. Source: Florida Department of Ag riculture and Consumer Services Division of Forestry. Figure 3 85 Percentage of wildfires and acreage burned in Florida ( by month ) between 1990 and 2006. Source: Flori da Department of Ag riculture and Consumer Services Division of Forestry. 0 1,000 2,000 3,000 4,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Number of Fires Year Trop Dep Trop Storm Cat 5 Cat 4 Cat 3 Cat 1 Cat 2 0% 10% 20% 30% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Percentage of Total Month Acres Burned Number of Wildfires

PAGE 144

144 Average Temperatures in Origin Regions and Destination CSAs Weather data including monthly maximum, minimum and average temperatures, precipitation, and snowfall from different cities in each U.S. region and Florida were retrieved from the National Climatic Data Center, a NOAA division. Weather data from the cities of New York (Northeast region), Chicago (Midwest region), Atlanta (South region), and Los Angeles (West region) were obtai ned and used as a general representation of the temperatures in each U.S. region. Also, weather data from Miami (South Florida CSA), Orlando (Orlando CSA), Tampa (Tampa St. Petersburg CSA), Jacksonville (Jacksonville CSA), and Fort Myers ( F ort Myers CSA) w ere collected. D ata from these five destination CSAs composed a Florida average. Florida recorded an annual average temperature of 73 degrees Fahrenheit during the period between 1990 and 2006. The period from May to September recorded the hottest average temperatures that range d from 78 to 81 degrees Fahrenheit. The hottest months on record were July and August ( average of 83 degrees Fahrenheit) The coldest period was between December and February ( average range between 62 and 64 degrees Fahrenheit ) with January ( average of 6 0 degrees Fahrenheit ) being reported as the coldest month Across U.S. regions, the Midwest region reported the coldest temperatures ( average of 29 degrees Fahrenheit in January) during winter, followed by the Northeast region ( average of 35 degrees Fahrenheit in January). The West region was four to six degrees cooler than Florida during winter, and reported the smallest fluctuations in temperatures when compared to the temperatures recorded in Florida. Also, all U.S. regions but the West region reported similar temperatures to Florida during the summer months of July and August. The Northeast region recorded an average temperature of 78 degrees Fahrenheit during these two months, only three degrees cooler than Florida. Also, the Midw est region recorded average temperatures in the rage of 74 to 76 degrees Fahrenheit, while the South region reported average temperatures in the 80s during these two months. In

PAGE 145

145 contrast, the West region was, on average, 13 degrees Fahrenheit cooler than Fl orida during these summer months. The difference between the average temperature in Florida and each of the four U.S. regions are shown in Figure 3 86 Figure 3 86 D ifference between average temperatures in Florida and each U .S. r egion between 1990 and 2006. Source: National Climatic Data Center. Precipitation in Florida Florida recorded an annual rainfall of 55.3 inches during the period between 1990 and 2006. The year 1997 was the wettest (64.4 inches), followed by 1994 (62. 9 inches) and 2005 (61.3 inches) during the 17 year period, while 1990 reported the lowest rainfall levels at 39.6 inches. The South Florida CSA recorded the highest precipitation levels during the same period (65.8 inches), followed by the Fort Myers CSA (57.9), Jacksonville CSA (54.0 inches), Orlando CSA (51.8), and Tampa St. Petersburg CSA (46.9 inches). The period between June and September recorded the highest precipitation levels during the calendar year. Average rainfall in these months ranged from 7 .6 to 8.8 inches. The driest month was November with an average rainfall of 2.2 inches. The wettest month in each of the top five destination CSAs was June with an average rainfall range of 7.1 to 10.6 inches. Figure 3 8 7 presents average rainfall by month in each of the six destination CSAs between 1990 and 2006. 0 5 10 15 20 25 30 35 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Difference in Temperature ( F) : Florida vs. U.S. Region Month Northeast Midwest South West

PAGE 146

146 Figure 3 87 Monthly a verage r ainfall in each of the five destination CSAs in Florida between 1990 and 2006. Source: National Climatic Data Center. Crime Rates in Flo rida Crime data were collected from the Florida Law Enforcement Agency Uniform Crime Reports and are available online. The crime index includes the following categories: violent crimes of murder, forcible sexual offenses, robbery and aggravated assault ; an d the property crimes of burglary, larceny theft and motor vehicle theft that were reported to Florida law enforcement. The crime index is defined as the number of crimes reported divided by the population and multiplied by 100,000. crime index has decreased by 45 % All crime categories included in the crime index have decreased during the 17 year period, especially the ones most frequently committed against tourists: burglary of holiday homes, vehicle theft, and robbery. Burglary ha s experienced the largest decrease (38%) from 275,104 total burglaries in 1990 to 170,133 in 2006. In addition, robbery has decrease d 37 % from 54,015 robberies in 1990 to 34,123 in 2006. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Tampa St. Petersburg 2.3 2.9 2.4 2.4 1.6 7.1 7.2 8.0 6.5 2.4 1.3 2.9 Orlando 2.1 2.3 3.3 2.7 3.3 8.1 7.6 7.3 5.7 3.8 1.9 2.8 Jacksonville 3.3 2.8 3.7 2.7 2.4 7.6 7.3 6.1 8.2 4.7 1.9 2.9 Ft. Myers 2.3 1.8 2.2 2.4 2.6 10.6 9.6 10.7 9.2 2.5 2.0 1.7 South Florida 1.7 2.1 2.7 3.1 5.5 10.4 6.3 9.3 10.9 8.1 3.7 2.3 CSA Average 2.3 2.4 2.8 2.7 3.1 8.8 7.6 8.3 8.1 4.3 2.2 2.5 0 2 4 6 8 10 12 Average Rainfall in Inches Tampa St. Petersburg Orlando Jacksonville Ft. Myers South Florida CSA Average

PAGE 147

147 Similarly, motor vehicle theft has experienced a 25 % decrease during the same time period. Figure 8 8 8 presents the crime index per 100,000 in Florida from 1990 to 2006. Figure 3 88 Annual crime index in Florida between 1990 and 2006. Source: Florida Law Enforcement Agency Uniform Crime Report s. Chapter Summary This chapter presented descriptive statistics for the domestic and international passenger, freight, and mail traffic traveling to Florida by means of airline transportation. All passenger, mail, and freight statistics were presented acc ording to the particular travel destination CSA Overall, the South region presented the highest levels of air passenger traffic while the West region exhibited the lowest levels of passengers traveling to Florida. The top three destination CSAs were Sout h Florida, Orlando, and Tampa St. Petersbur g, respectively. Regarding international passengers traveling to Florida, there were more passengers traveling from Latin America than any other region. Passengers from Europe and Canada followed. The South Florid a CSA received more passengers from Latin America than any other international origin, while the Orlando CSA received more from Europe and the Tampa St. Petersburg from Canada. 4,000 5,000 6,000 7,000 8,000 9,000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Crime Index Rate Year

PAGE 148

148 Airline ticket prices from the West region were more expensive than from any ot her region traveling to Florida. The Northeast region yielded the cheapest air line tickets Airline ticket prices from Europe were more expensive while airline tickets from La tin America were the cheapest among international origins. The chapter also incl uded e conomic indicators, as well as average temperatures and precipitation related to the origin U.S. regions. It also presented weather statistics on average temperatures, precipitation and hurricanes and wildfires affecting Florida. Oil prices, adverti sing expenditures, and crime rate statistics from Florida were also shown in this chapter. Advertising expenditure statistics are highly seasonal. Most of the generic and brand advertising expenditures occurred in t he first quarter of every year.

PAGE 149

149 CHAPTER 4 THEORETICAL FRAMEWOR K Motivation D emand for domestic air passenger traffic to Florida can be modeled in several ways Chapter 2 discussed several approaches currently used by econometricians Such approaches range from simple auto regressive models to m ore complicated specifications such as time varying parameter models. Most of these studies had used annual data to estimate demand and the scope had been limited to forecasting international demand for tourism. Nevertheless, studies by Phakdisoth and Kim (2007) and Proena and Soukiazis (2005) used a partial adjustmen t model framework to identify drivers of demand for tourism in Laos and Portugal, respectively. Conversely, this study developed a partial adjustment framework to model domestic demand as oppo sed to international, used monthly data instead of annual data, and more importantly, expand ed the dynamic 1 structure of the partial adjustment model from one to three lagged dependent varia bles In order to motivate the partial adjustment model framework, first let a model be constructed where the number of passengers is defined as a function of its past v alues as shown in Equation 4 1 Time series literature identifies a uto regressive integrated moving average (ARIMA) models as a popular approach to model and forecast this type of specification : = ( 4 1) Since the ARIMA model lacks a base of economic theory critical for policy implications, an extension can be made to add variables t hat can explain the behavior of N umber of 1 The term dynamic refers to lagged values of the dependent variable included as right hand side variables to model demand respon se to economic stimuli as explained by W.H. Greene in Econometric Analysis, Chapter 19.

PAGE 150

150 passengers is now defined in Equation 4 2 as a function of its past values and other explanatory variables : = ( 4 2) E xplanatory variables in attempt to account for the levels of while stil l recognizing that there remain patterns that cannot be directly explained. Hence, inclusion of is still expected to be of major importance and represents the dynamic component of the model As consumers, in general, do not immediately adjust to changes in their demand determinants, appropriate dynamic systems are essential, before plausible behavioral hypothesis can be tested (De Mello and Fortuna 2005). These authors also agreed that suitab le dynamic generalizations of demand systems are a rare feature in empirical studies. This study, by specifying a partial adjustment model, attempts to address this issue as a contribution to the research literature. Nerlove (1972) used adjustment costs ex ample to explain why consumers do not adapt instantaneously to changes in prices and how dynamics of behavior is a more suita ble approach to analyzing short term phenomena. The static or long run framework does not allow prices to change. It assumes that p rices are the same through time and hence, consumer behavior rests in a state of equilibrium. This assumption is very restrictive and may only be useful when applying a comparative statics analysis. I ntroduction of dynamics is necessary to the economic fr amework to impro ve the forecasting accuracy of models. Vanhove (2005) argues that lagged dependent variables are often used to take into account a time lag in the relationship between a dependent variable and an independent factor. I nclusion of lagged depe ndent variables in tourism demand functions allows for habit persistence and supply rigidities. S ources of these dynamics within the demand for

PAGE 151

151 airline transportation to Florida can be attributed to habits of consumption created th r ough time and by events in previous periods that affect consumer behavior by past purchases. In the case of air travel to Florida, habit can be a result of factors such as ownership of real estate or time shares in the state, institutional structure such calendar holidays that attract visitors, sense of security and safety, and a supply of unique products (e.g ., beaches, attractions). In addition, consumer s return to Florida to reduce search costs related to finding another place to travel. C onsumer s are also already familiarize d with the environment and know the services provided which creates a sense of comfort and security. Consumers face difficulties in order to respond to events immediate ly. Adjustment costs little or no flexibility to consumers when hurricanes, wildfires, and terrorist attacks occur. Rescheduling or cancellation of bookings will m ost likely carry a transaction cost. While some consumers will plan ahead and buy insurance to reduce their risk they still incur the cost of insurance. Such habits a nd events prevent consumers to adjust ing from one period to another and t his is where a dynamic s tructure is needed to explain demand behavior Lagged explanatory variables that take into account such behavior should be included in the model. By introducing a lagged dependent variable as an explanatory variable, the model attempts to capture any persistence effect or rigidity of tourist behavior. The underlying assumption is that the period which is required for full adjustment exceeds the interval of observation (Nerlove 1958). Other authors have used the same principles of the partial adjus tment model to include habit persistence in the estimation. Chamera and Deadman (1992) used a similar framework to

PAGE 152

152 explain how inertial factors prevent an immediate movement to a new desired consumption level. They also argue that this framework can be ext ended to aggregate consumption theories which also lead to estimating forms similar to the ones presented in the next section. The Partial Adjustment Model P artial adjustment model s include short run dynamics that static models fail to address. The partia l adjustment model has two components, a static component and a dynamic component. The static component, shown in Equation 4 3 states that desired amount of the dependent variable is determined by some explanatory variables : = 0 + 1 + ( 4 3 ) The dynamic component shown in Equation 4 4, can be explained as follows. The difference between the current level and the previous level 1 is a portion of the difference between the desired level and the previous level 1 1 = 1 ( 4 4 ) The coefficient is the speed of adjustment coefficient and can take values between 0 and 1. If 0 then 0 = 1 which impli es a slow speed of adjustment Consumers take more than one period to fully adjust to short run changes in their demand drivers. I f 1 then 1 = which impl ies a high speed of adjustment In other words, consumers adjust immed iately to any short run changes in their demand drivers. If consumers adjust fairly quickly to short run changes in their demand drivers, the adjustment coefficient becomes irrelevant and therefore, the dynamic specification expressed in Equation 4 4 is not necessary. The following equation is obtained when substituting E quation 4 3 into E quation 4 4 :

PAGE 153

153 1 = 0 + 1 + 1 = 0 + 1 + 1 + 1 = 0 + 1 + 1 + 1 = 0 + 1 + 1 1 + ( 4 5 ) The coefficient in Equation 4 3 can b e interpreted as the long run relationship between the dependent variable Y and the explanatory variable X. Note that this long run relationship cannot be directly estimated since the equilibrium quantity demanded cannot be observed An unobser vable quantity demanded may arise due to frequent changes in prices and income. However, the partial adjustment model specification provides a useful way to derive estimates of the long run relationship. Equation 4 5 shows that, under the partial adjustmen t model framework, the long run coefficient is multiplied by the speed of adjustment coefficient Since there are two coefficients associated with X, a transformation is needed to estimate Equation 4 5. The estimating equation is presented in Equation 4 6. Note that the new estimabl e coefficients are now interpreted as the short run relationship between the dependent variable Y and the explanatory variable X: = 0 + 1 + 2 1 + ( 4 6 ) where 0 = 0 1 = 1 and 2 = 1 There is no adjustment if 2 = 1 i e = 0 and the dependent variable at current pe riod is a function of its past value s which is a func tion of the related to that past period. Any changes in the explanatory variables in that past period are still affecting the dependent variable in the current period. Consumers have not adjusted to any changes in the determinants from the past period yet and hence, equilibrium has not been achieved. In this case, the dynamic specification is relevant in order to have a correct interpretation of the coefficients.

PAGE 154

154 The opposite extreme would be if 2 = 0 The speed of adjustment = 1 implies that 1 = Therefore, the dependent variable is solely determined by the explanatory variables at the current period The dynamic component of the model specified in Equation 4 5 is no longer relevant because consumers fully and i nstantaneously adjust to any changes in the determinants and consumers are able to achieve equilibrium within the same time period. In other words, short run elastici ties are equivalent to long run elasticities and equilibrium is achieved i mmediately Dome stic Air Passenger Traffic Partial Adjustment Model T he domestic air passenger traffic partial adjustment model (DAP PAM) identif ies factors that influence demand for air passenger traffic traveling to Florida. As described by Pollak (1970), past consumpt ion patterns are an important determinant of present consumption patterns. and thus repeated visitation occurs (habits). The DAP PAM will determine if a habit form ation pattern exists among passengers traveling to Florida. Assume that there is a desired domestic air passenger traffic demand and it is a function of some explanatory variables These explanatory variables, presented in Equation 4 7, include airline ticket prices, personal disposable income, average temperature, and average precipitation. Wildf ires, advertising expenditures, and dummies for storms, monthly seasonality and terrorism ( i.e., 9 11 terrorist attacks) are also included in = X = , HCAT LOW 2 3 4 5 6 7 2 3 4 5 6 7 8 9 10 11 12 (4 7) Then, the desired demand for air passenger traffic can be given a non linear specification where its explicit form can be expressed as

PAGE 155

155 = 0 + 1 2 3 4 5 6 7 8 9 8 + j TER j 7 = 2 14 + 12 = 2 ( 4 8 ) The adjustment process is assumed as: = 2 4 12 = = 2 4 12 where 0 < < 1 and = 2 4 12 = 1 ( 4 9 ) Taking the log linear form o f Equation 4 9, the r esult is = 2 4 12 = = 2 4 12 = = 2 4 12 + = 2 4 12 = = 2 4 12 + = 2 4 12 = + 1 = 2 4 12 where 0 1 and = 2 4 12 = 1 ( 4 10 ) The first component in Equation 4 10 represents the static component of the DAP PAM and the second component refers to the dynamic component of the model. The stat ic component was constructed using economic theory and included variables illustrated in Equation 4 7.

PAGE 156

156 V ariables included in the dynamic component resulted from a spectral analysis performed to the domestic passenger data Note that a restriction on the we ights has been imposed where the sum of the weights must add up to one The restriction was imposed in order to identify the elasticity of adjustment coefficient otherwise no inference can be made about the level of response realized in the current period versus past periods. A det ailed explanation of all variables included in the model is presented in the following sections. Construction of Empirical Domestic Air Passenger Model It is hypothesized that the elasticity of adjustment coefficient varies across destination CSA and origin region (CSA ORG) pair. Therefore, separate demand equations have been specified for each CSA ORG pair. Each equation include s a dependent variable, passenger traffic per 100,000 persons, on the left hand side and a constant term and twenty nine variables on the right hand side. The right hand side includes 26 variables representing the static component of the model and three va riables representing the dynamic component of the model. The static component includes a CPI adjusted air ticket price specific to the CSA ORG pair, income, temperature index, and rainfall index specific to the origin regio n, a fire index and three storm d ummy variables specific to all Florida, advertising expenditures, six 9 11 terrorist attack dummy variables, and eleven monthly dummy variables. The dynamic component includes dependent variable lagged two, four, and twelve months as determined by the spec tral analysis performed on the domestic air passenger data. The spectral analysis will be discussed later in the c hapter. The dependent variable, DAP, represents number of passengers traveling from a particular origin region (ORG) to a specific destination (CSA) adjusted by population (100,000 habitants) o f the ORG. P opulation estimates reported annually by the U.S. Census were used to estimate

PAGE 157

157 monthly population values. Refer to Chapter 3 for an explanation of the methodology used in the selection and aggr egation of airline passenger traffic for each CSA ORG pair. One major limitation of the variable is that it does not provide information about the passenger. The variable DAP fails to identify the passenger as a person traveling for leisure or business or itinerary Recall that every itinerary has two legs: first leg and returning leg. Still, the variable could give valuable insights on travel patterns. According to Visit Florida more than one half of domestic visitors travel to Florida via commercial flights. Since the objective of the study is to identify general patterns of air travel to Florida, the variable was used to model such patterns. Description of Variables in the St atic Component of the DA P PAM The variable FARE represents air ticket price s of each CSA ORG pair adjusted by the national Consumer Price Index for All Urban Consumers for air fares (CPI AFARE) published on a monthly basis by the U.S. Bureau of Labor Statis tics. Airline ticket prices to each Florida destination were deflated by the national average prices. The CPI AFARE is a reflection of airline ticket prices from all domestic origins to all domestic destinations. Therefore, FARE reflects deflated airline t icket prices to Florida relative to airline ticket prices to all destinations nationwide. If deflated airline ticket prices to Florida were exceptionally high, consumers may go somewhere else because prices to Florida are more expensive. Using the CPI ALL to deflate ticket prices would give similar results since the correlation between both indexes is 0.93. Refer to Chapter 3 for an explanation of the methodology used in the selection and a ggregation of a irline t icket p rice s for each CSA ORG pair. The INC v ariable represents the average per capita personal disposable income for each region. Note that the average per capita personal disposable income f rom Florida was excluded from the South region. Similarly to the

PAGE 158

158 population variable, the annual average pers onal disposable income reported by the B ureau of Economic Analysis (B EA ) was used to estimate monthly values. The next three variables included in the model are related to weather either in the origin or at the destination. TEMP and PCP refer to indexes o f average monthly temperature (in degrees Fahrenheit) and average monthly rainfall (in inches) at the origin region. FIRE represents an index of wildfires affecting Florida and HCATLOW, HCATMED, and HCATHGH refer to dummy variables for tropical depressions and storms and hurricanes affecting Florida. TEMP and PCP reflect the index of the temperature and precipitation reported in each U.S. region on a month ly basis normalized by an estimated monthly average value. A value of 1.0 represents average conditions during that month whereas any value less or greater than 1.0 represents temperature and rainfall below or above the normal. The FIRE variable refers to a value of weighted wildfires normalized by an estimated average value of such weighted wildfires for each month. The weighted value of fire is a combination of the size and number of wildfires where the number and size of wildfires are weighted by the size of the small wildfires Therefore, the value will give a higher value to larger wildfires (greater i mpact) compared to the smaller wildfires HCATLOW includes the number of tropical depressions and storms that affected Florida. HCATMED includes number of hurricanes of category 1 and 2 and HCATHGH includes number of hurricanes of category 3, 4, and 5. The se categories follow the classification defined by the Saffir Simpson hurricane scale. The ADV variable represents total advertising expenditures by destination CSA. All destination CSAs include total generic advertising efforts performed at the city, cou nty, and state level. Note that these totals were not available for each destination CSA and therefore, the total

PAGE 159

159 was used in each destination CSA. The variable also includes, if available, brand advertising expenditures of those companies that do business in a specific destination CSA. For example the ADV for the South Florida CSA includes total generic advertising expenditures for Florida plus brand advertising expenditures of Carnival Cruise Lines In addition to total generic advertising expenditures, the ADV for the Orlando CSA is composed by brand advertising expenditures from t he Walt Disney Company Universal Studios and SeaWorld Similarly, advertising expenditures from Busch Gardens were included in the ADV variable for the Tampa St. Petersbu rg CSA. Since no specific brands were available for the CSAs of Jacksonville and Fort Myers, total generic advertising expenditures for Florida were used. The dummy variables TER1 TER7 represent the terrorist attacks of September 11, 2001 in New York City Washington, D.C. and Pennsylvania. TER1 takes a value of one for all months prior to September, 2001 and 0 otherwise. TER2 takes a value of one for the twelve months following the terrorist attack including September 2001 and zero otherwise. TER3 TER7 t akes a value of one on a twelve month sequence from September 2002 to August 2007. For example, TER3 takes a value of one for each month from September 2002 to August 2003, zero otherwise. TER4 takes a value of one for each month from September 2003 to Aug ust 2004, zero otherwise, and so on. TER1 was dropped from the estimation to avoi d the dummy variable trap ( i.e., perfect multicollinearity). Therefore, all inferences of TER2 TER7 are compared to the base, TER1 (pre 9 11 levels ). B y setting each TER leve l in a twelve month sequence, the effect of the variable has not been confounded with the adjustment associated to seasonality. Another possibility to construct the dummy variable was t he o ne where a dummy variable would take a value of zero for all pre 9 11 months and a value of one for all months after 9 11. Such before and after 9 11 approach

PAGE 160

160 would have hindered the ability to make inferences about the recovery process that could occur through time and therefore was ultimately not considered. Monthly se asonality dummies and lagged dependent variables are included in the right hand side to account for seasonal and consumption patterns in previous periods respectively Dummies for monthly seasonality reflect any event that occurs every year influenced by either weather or institutional structure. For example in March and April of every year. Holidays such as spring break, Fourth of July, Memorial Day Weekend, Thanksgiving Weekend, and so on, give structure to the supply of services available to consumer s A vailability of these services is limited, since the event only occurs at a certain time of the year. Therefore consumers will travel that time of the year when the se services are available. For example, a consumer interested in bird watching in Florida will most likely visit the state during the bird watching season. Also, car racing fans will be enticed to travel to Daytona Beach in February more than any other month for the annual Daytona 500. Clearly, su ch events are a source of se asonal patterns and the use of monthly seasonal dummies in the model will capture them. On the other hand the lagged dependent variables show how past events influence ence, capture consumer behavior M onthly dummy variables are represented by MTH1 MTH12 to account for seasonal effects. Similarly to TER1, MTH1 was dropped from the estimation in order to avoid the dummy variable trap Therefore, all inferences of MTH2 MTH 12 are compared to the base, MTH1 (Janu ary). T ransformations of the variables were performed and stored using TSP TM software. A copy of the TSP TM program is illustrated in Appendix B

PAGE 161

161 Identification of the Variables in the Dynamic Component of the DAP PAM I dentification of the dynamic component can be made by either running a correlogram or conducting a spectral analysis on domestic airline passenger data Since the correlogram presents some ambiguity surrounding the choice of a suitable dynamic model (Har vey and Todd 1983), a spectral analysis has been conducted instead. The spectral analysis helps identify any seasonal fluctuations of different lengths in the data. Figure 4 1 shows the power spectrum of the domestic air passenger traffic data for the Flo rida Northeast pair. P eriodogram values were plotted against the time domain (months) and are interpreted in terms of variance of the data at the respective month. The spectral analysis shows a high level of variance every two, four, and twelve months. All other 23 CSA ORG pairs exhibited similar behavior in the level of variance at months two, four, and twelve. The lack of information in identifying which months report the highest variance is a major limitation of the spectral analysis. One must use some theoretical expectations to explain the peaks illustrated by the spectral analysis. It can be hypothesized that individuals go on vacation once a year to explain the twelve months peak in the variance of passenger traffic. For example, the school holiday s eason (June through August) represents a key season for those touristic destinations trying to attract visitors to their states. Since most of the Florida attractions are family oriented such as Disney World theme parks, it can be assumed that Florida wil l experience an increase in visitors every year during the school holiday season.

PAGE 162

162 Figure 4 1 Power s pectrum of domestic air passenger traffic data for the Florida Northeast (33 1001) pair The four month peak can be attribut ed to the institutional structure of holidays. A major calendar holiday comes around roughly every four months a such as Spring Break (March), Fourth of July (July), and Thanksgiving Day (November). Florida is a major destination during spring break while the Fourth of July is in the middle of summer which attracts many to the Florida beaches. Thanksgiving holiday records one of the highest levels of air passenger traffic around the country also and Florida benefits from the long holiday weekend. It is more difficult to explain the two month peak since no clear reason to explain a high variance of the passenger traffic can be identified. Nevertheless, the lag was included in the model as the spectral analysis suggested. Given the results given by the spectr al analysis performed on the dependent variable and the theoretical reasoning discussed above, the dyna mic component of the model for domestic air passenger traffic includes the dependent variable lagged two, four, and twelve months. Table 4 1 presents a s ummary of the variables included the DAP PAM. 11.9 4.0 2.4 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 215.0 43.0 23.9 16.5 12.6 10.2 8.6 7.4 6.5 5.8 5.2 4.8 4.4 4.1 3.8 3.5 3.3 3.1 2.9 2.8 2.7 2.5 2.4 2.3 2.2 2.1 2.0 Periodogram Values Periodocity (in months)

PAGE 163

163 T able 4 1 Summary of the variables included in the DAP PAM. Variable Description PASSENGERS DAP D,O = Domestic air passenger traffic traveling from region (O) to destination (D) ad justed by population of origin (O); measured in passengers per 100,000. FARE FARE D,O = Average air ticket price to destination (D) from origin (O) adjusted by the CPI U Airfare component; measured in U.S. dollars. INCOME PCDI O = Per capita personal disposa ble income in region (O), measured in U.S. dollars. TEMPERATURE TEMP O =Index of average temperature in region (O); unity. PRECIPITATION PCP O = Index of total precipitation in region (O); unity. FIRE FIRE D = Index of wildfires in Florida; unity. STORMS HCA TLOW D =(Tropical storms & depressions ); unity. HCATMED D =(HCAT1+HCAT2); unity. HCATHGH D =(HCAT3+HCAT4+HCAT5); unity. ADVERTISING ADV D =Advertising expenditures incurred to promote destination (D); measured in thousands of U.S. dollars. 9 11 ATTACKS TER2 =[ (YY:MM>01:08 & YY:MM<02:09)=1] or (otherwise=0) TER3=[ (YY:MM>02:08 & YY:MM<03:09)=1] or (otherwise=0) TER4=[ (YY:MM>03:08 & YY:MM<04:09)=1] or (otherwise=0) TER5=[ (YY:MM>04:08 & YY:MM<05:09)=1] or (otherwise=0) TER6=[ (YY:MM>05:08 & YY:MM<06:0 9)=1] or (otherwise=0) TER7= (YY:MM>06:08=1) or (otherwise=0) SEASON MTHS2=(February=1) or (otherwise=0) MTHS3=(March=1) or (otherwise=0) MTHS4=(April=1) or (otherwise=0) MTHS5=(May=1) or (otherwise=0) MTHS6=(June=1) or (otherwise=0) MTHS7=(Ju ly=1) or (otherwise=0) MTHS8=(August=1) or (otherwise=0) MTHS9=(September=1) or (otherwise=0) MTHS10=(October=1) or (otherwise=0) MTHS11=(November=1) or (otherwise=0) MTHS12=(December=1) or (otherwise=0) PASSENGERS t 2 DAP t 2,D,O = Domestic air p assenger traffic traveling from region (O) to destination (D) adjusted by population of origin (O) lagged two months ; measured in passengers per 100,000. PASSENGERS t 4 DAP t 4,D,O = Domestic air passenger traffic traveling from region (O) to destination (D ) adjusted by population of origin (O) lagged four months; measured in passengers per 100,000. PASSENGERS t 12 DAP t 12,D,O = Domestic air passenger traffic traveling from region (O) to destination (D) adjusted by population of origin (O) lagged twelve mont hs; measured in passengers per 100,000.

PAGE 164

164 Using the log linear form of Equation 4 5 and including the result in Equation 4 7 and applying the restriction on the weights of the lagged dependent variables, an expression for the DAP PAM can be represented in l og linear as = 0 + 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 + 8 + 7 = 2 + 14 + 12 = 2 + 1 2 ln 2 ln 12 + 1 4 ln 4 ln 12 + 1 ln 12 + ( 4 11 ) Also note that there are 24 CSA ORG pairs re sulting from four origin region s with passenger tra ffic to six destination CSAs. The four pairs for the CSA=33 represent the addition of the top five destination CSAs and the other smaller destination CSAs not included in the individual analysis. I dentification of the destination CSA (D) and the origin re gion (O) is presented in Table 4 2. Refer to Chapter 3 for a detailed explanation regarding the selection and aggregation procedures performed to construct the five individual CSAs.

PAGE 165

165 Table 4 2 Identification of d estination CSA s and o rigin U.S. r egions Destination (D) Destination CSA Origin (O) Origin U.S. Region 1 South Florida 1001 Northeast 5 Orlando 1002 Midwest 8 Tampa St. Petersburg 1003 South 9 Jacksonville 1004 West 14 Fort Myers 33 Florida C oefficients in Equation 4 5 and Equation 4 1 1 can be interpreted as the long run relationship between the demand for air passenger traffic and X. C oefficients associated with the variables in logarithm form are interpreted as the average propensity to travel given a change in X N ote that this long run relationship cannot be directly estimated since the equilibrium quantity demanded cannot be observed because e xplanatory variables are continually changing. However, the partial adjustment model spe cification provides a useful way to derive estimates of the long run relationship. Equation 4 11 shows that the long run coefficients are multiplied by the speed of adjustment coefficient The speed of adjustment coefficient can now be inte rpreted as the elasticity of adjustment since the model has been specified in logarithm form. Since there are two coefficients associated with X, a transformation is needed to estimate Equation 4 1 1 The new estimable coefficients are ill ustrated in Equation 4 1 2 and interpreted as the short run relationship between the demand for air passenger traffic and X.

PAGE 166

166 = 0 + 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 + 8 + 7 = 2 + 14 + 12 = 2 + 27 ln 2 ln 12 + 28 ln 4 ln 12 + 29 ln 12 + where 29 = 1 0 1 = 2 4 12 = 1 and = = 1 26 ( 4 12 ) There is no partial adjustment if = 0 an d the demand for domestic air passenger traffic is determined exclusively by its past values. The opposite extreme would be if = 1 which implies that 1 = Therefore, demand for domestic air passenger traffic is so lely determined by explanatory variables and the dynamic component of the model specified in Equation 4 9 is no longer relevant. It is hypothesized that the elasticity of adjustment lies between 0 and 1. In other words, demand for air passenger tra ffic to Florida can be explained by conditions occurring in the current period and conditions that occurred in past periods. Estimation Possibilities Initially, the model in Equation 4 1 2 was estimated using ordinary least squares (OLS). Several tests were performed to identify any heteroskedasticity, serial correlation, and stationarity Since lagged dependent variables were included in the right hand side of the equation the augmented Dickey Fuller test was performed to determine if the data contained a

PAGE 167

167 unit root. The null hypothesis was rejected at a 95 % confidence level concluding that residuals from the OLS regression are stationary. The Durbin h test was conducted to determine if there was a correlation in the errors corresponding to successive time periods Given the results from this test, the null hypothesis of no serial correlation was rejected. The OLS estimation approach can no longer be considered a viable estimation procedure for Equation 4 1 2 because it yields inconsistent estimates. In order to correct for this problem a first order auto regressive process (AR1) for the error was included in Equation 4 1 2 = 1 + where ~ 0 2 ( 4 13 ) Therefore, by combining Equation 4 1 2 and Equation 4 13 an estimating equation that takes into account the AR(1) pro cess can be defined as = 0 + 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 + 8 + 7 = 2 + 14 + 12 = 2 + 27 ln 2 ln 12 + 28 ln 4 ln 12 + 29 ln 12 + 1 + where 29 = 1 0 1 = 2 4 12 = 1 ~ N 0 2 and = = 1 26 ( 4 14 )

PAGE 168

168 Another estimation possibility that had to be considered was the fact that the error s could be correlated not only over time, but also across cross section units. Initially, OLS has treated each of the 24 CSA ORG pairs in the DAP PAM as separate equations. The procedure fails to address any auto correlation and also the possibility that t he errors from some of the equations could be correlated. The AR (1) estimation procedure takes into account the auto correlation but not the cross section correlation Then, the technique of multivariate regression generally gives more efficient estimates than OLS and AR (1) regressions applied separately to each equation There are three ways to model this possible cross section correlation. One general possibility i s the fact that all 20 equation residuals are correlated with each other due to some set of unobservables common to all destinations and origins alike. The second possibility is that there is a set of unobservables unique to the origin region not captured by the specified model. The third and final possibility is that there are some unobservab les unique to the destination for which the variables specified in the model do not account Therefore, four variations of the model were specified and are summarized as follows: 1. AR1 estimation for each equation separately: assumes that there is no presen ce of dependent errors across equations. It only takes into account an AR (1) process in the error term. There is only one set of 24 equations. 2. SUR AR1 ALL estimation that includes all U.S. regions traveling to all destination CSAs; assumes that there is i nformation in all U.S. regions and destination CSAs not being explained by the model but that could be temporally dependent across all U.S. origins and destination CSA s, and hence, a source of correlation The model is also considering a correlation in the errors corresponding to successive time periods ( i.e., AR (1) process). There is only one set of twenty equations. The destination CSA representing Florida ( i.e., CSA= 33) was not included in each set since it represents the aggregation of all destination CSAs in Florida and was estimated separately 3. SUR AR1 ORG estimation for each U.S. region traveling to all destination CSAs: assumes that there is information unique to a particular region not explained in the model that could be temporally dependent acro ss destination CSAs The model is also considering a correlation in the errors corresponding to successive time periods ( i.e., AR (1) process). There are four sets of equations represent ing each U.S. origin and each set includes five equations. For example the set for the Northeast region includes all five

PAGE 169

169 destination CSAs related to the Northeast region : 1 1001, 5 1001, 8 1001, 9 1001, and 14 1001 The destination CSA representing Florida ( i.e., CSA= 33) was estimated separately 4. SUR AR1 CSA estimation fo r all U.S. regions coming to a particular destination CSA: assumes that there is information in a particular destination CSA not explained in the model that could be temporally dependent across regions. The model is also considering a correlation in the er rors corresponding t o successive time periods ( i.e., AR (1) process). There are six sets of equations represent ing each U.S. origin and e ach set includes four equations For example, the set for the South Florida CSA includes all four origin U.S. regions r elated to the South Florida CSA: 1 1001, 1 1002, 1 1003, and 1 1004. Chapter Summary Chapter 4 presented the theoretical framework of the partial adjustment model and also the empirical application to the demand for air passenger traveling to Florida. The model, which includes a static and a dynamic component, is an attractive approach since it yields a speed of adjustment coefficient that shows how fast passengers adjust to events that affect their decision to travel to Florida. An empirical model was intr oduced to show how the demand for air passenger traffic could be represented and estimated. Economic theory was used to build the static component and spectral analysis was performed to determine the dynamic component of the model. Preliminary diagnostics showed that the model is stationary (augmented Dickey Fuller test). In addition the Durbin h statistic suggested the errors have an auto regressive scheme of order 1. These diagnostics helped concluded that OLS estimates will be inconsistent and therefore other estimation procedures were needed.

PAGE 170

170 CHAPTER 5 RESULTS Separate equations for each CSA ORG pair were constructed. Each equation includes a dependent variable, passenger traffic per 100,000 persons, on the left hand side and a constant term and 2 8 variables on the right hand side. The right hand side includes a CPI adjusted air ticket price specific to the CSA ORG pair, income, temperature index, and rainfall index specific to the origin region, a fire index specific to the destination CSA, three hurricane dummy variables, advertising expenditures, six 9 11 terrorist attack dummy variables, eleven monthly dummy variables, and the dependent variable lagged four and twelve months. C omparison of the Estimation Alternatives The 1 coefficient estimates were analyzed to determine which of the three approaches yields more stable coefficients. Recall that the speed of adjustment parameter must lie between zero and one, equivalent to state that the 1 coefficient must lie between zero and one, to achieve stability. Results show that the SUR AR1 ALL estimation yield s the fewe st unstable coefficients among the four estimation alternatives. The estimate from the CSA ORG pair 14 1004 has a negative sign, but it is not signific ant. Three pairs exhibit unstable 1 coefficient estimates using the SUR AR1 CSA estimation, but none were significantly different from zero. The SUR AR1 ORG yielded similar results. T able 5 1 shows the 1 coefficient estimates and their corresponding t value for each CSA ORG pair u nder each estimation procedure. C oefficient estimates from the SUR AR1 ALL are shown in the follow ing section. R esults are presented in sub sections according to each destination CSA.

PAGE 171

171 Results for Demand for Passengers by CSA This section presents results of the estimated model developed in Chapter 4 for Florida and each destination CSA using the SUR AR1 ALL estimation approach. A copy of the TSP TM program is illustrated in Appendix C Note that the SUR AR1 ALL model was estimated allowing the auto correla tion coefficient to differ across CSA ORG pairs following the specification adopted by Maekawa and Hisamatsu (2002) and Greene (2003). Each sub section includes a discussion of the results and a table with the coefficient estimates and corresponding t values, R squared, a nd number of observations. E lasticity of adjustment coefficient estimates and other explanatory variables (e.g., air fare, income, dummies for storms and terror) from the static component will be discussed. Note that estimates for the a ir fare, income, advertising, temperature, precipitation, and fire variables can be interpreted as short run elasticities since the estimate is the multiplication of and Long run elasticities can be calculated by dividing the estimate by the elasticity of adjustment L ong run elasticities yield greater which is in line with demand theory. Passengers have a gr eater ability to respond to changes in the dem and drivers in the long run than in the short run. Due to asymmetry in information and relative ly inflexible budget allocations, it takes time before cha nges affect demand (Syriopoulus 1995). E stimates of the d ummies controlling for monthly seasonality are not discussed in this chapter. But results suggest that there are strong seasonal patterns in each CSA ORG pair and therefore the model required the inclusion of these dummy variables. Since inference of these coefficient estimates only allows a comparison against January, little insight can be obtained. Chapter 6 includes a section that discusses seasonal patterns of demand for air passengers traveling to Florida.

PAGE 172

172 Since there is a strong theoretical expectatio n of the sign of air fare (negative), income (positive), fire (negative), dummies for storms (negative), advertising (positive), and dummies for terror (negative), their calculated t values were compared against a critical t value of |1.658| which represent s a 95 % confidence on a one tailed t distribution If the t value (in absolute value) of the coefficient is greater than |1.658|, the null hypothesis would be rejected. C alculated t values of the constant term, t emperature, precipitation, and dummies for m onthly seasonality were compared against a critical t value of |1.96| which represents a 95 % confidence on a two tailed t distribution The validity of the final model SUR AR1 ALL, was checked by analyzing the residuals of each set of equations. A uto cor relation and partial correlation coefficients of the residuals with several time lags were ex amined. Results indicated that residuals were random (no signs of any other auto correlation scheme) One can let X change today and calculate how long the effect takes to be fully realized. The 29 ) slightly different. The 1 29 ) tells whether complete rigidity = 0 partial adjustment 0 < < 1 or complete adjustment = 1 occurs If complete r igidity is present, the model could be re estimated as a pure ARIMA and determine the significance of each coefficient associated to the lagged dependent variables. Given the results obtained from the model estimated, this step was not necessary because no CSA ORG pair showed complete rigidity. Therefore, the following discussion centers in short and long run changes in demand given a change in explanatory variables

PAGE 173

173 Florida Results Table 5 2 presents the coefficient estimates and their corresponding t v alues for the demand for air passengers from four U.S regions traveling to Florida using the SUR AR1 ALL approach E lasticity of adjustment estimates are less than one and significant at 95 % confidence in three U.S. regions traveling to the Flori da CSA. Th ese results suggest that current demand for air passengers traveling to The Florida CSA is driven by both current and past events. Analysis of coefficients in dynamic component The elasticity of adjustment estimate from the Northeast region suggests that for a one percentage increase in past demand, current demand for air passengers on average increases by 0.35 % Almost two thirds of the adjustment occurs in the current period, while events that occurred in past periods affect current demand by 0.35. Simil arly, the elasticity of adjustment estimate of the other three U.S. regions can be analyzed. Results suggest that similar to the Northeast region, demand from the West region adjusts somewhat immediately to events in the current period (0.64). The South re gion adjusts more rapidly to current events (0.79) than to past events. The remaining 0.21 is attributed to past events The coefficient for the Midwest region was significant given a one tailed test Analysis of coefficients in static component Within th e economic variables, air fare estimates have the correct sign in all U.S. regions but they are only significant in the Northeast and South region s For example, the airfare elasticity of demand is 0.249 for the Northeast region and can be interpreted as follows : a percentage increase in the price of airline tickets from the Northeast region to The Florida CSA represents a 0.249 % decrease in d emand for air passenger traffic Income estimates have a positive sign and are significant in all U.S. regions Th e short term income elasticities of demand are 1.011, 1.420, 1.008, and 1.253 for the Northeast region

PAGE 174

174 Midwest, South, and West region, respectively. Note that income has a similar effect (~1 % increase) o n the demand for air pa ssengers from every region. C oefficient estimates for advertising expenditures have the correct sign but they are all not significant. In review, only income estimates are significant across all U.S. regions while airfare is significant in two U.S. regions (Northeast and South). Te mperature and precipitation, which are weather variables related to the origin, are not significant, while some of the weather variables related to The Florida CSA are significant The HCATHGH, HCATMED, and FIRE estimates have the correct sign (negative) a nd are significant in most cases. However, HCATLOW behaves as expected (i.e., estimates have negative sign) in the West region only Estimates from the Northeast, South, and West regions suggest that tropical storms and depressions have no effect o n demand Hurricanes have a negative impact o n demand from all four U.S. regions, at a significant level. Only the estimate for the HCATMED from the West region is not significant. Wildf ires also have a negative impact on demand, but only the estimate from the Sou th region is significant. As expected, dummies for terror (TER2 TER7) have a negative effect o n demand from all U.S. regions. Although the effect is still negative, coefficient estimates suggest that the impact is reduced as time increases. Meanwhile, the effect of the 9 11 terrorist attacks o n demand from the Northeast region is negative but since September 2003 is no longer significant. The e ffect of the terrorism dumm ies o n demand for a ir p assengers traveling to The Florida CSA is presented in Figure 5 1. Note that the g ray colored cylinders indicate that the estimate is not significant with 95 % confidence. South Florida CSA Results C oefficient estimates and their corresponding t value s of the SUR AR1 ALL model for four U.S regions traveling to the South Florida CSA are presented in Table 5 3. The stability

PAGE 175

175 condition holds since elasticity of adjustment estimates are less than one In addition, all coefficients are significant in the four U.S. regions traveling to the South Florida CSA. These results sugg est that current demand to the South Florida CSA depends o n both current and past events. Analysis of coefficients in dynamic component The elasticity of adjustment estimate from the Northeast region suggests that for a one percentage change in past deman d, current demand for air passengers change s by 0.24. In other words, almost three fourths of the adjustment occurs in the current period, while the remaining portion of the adjustment depends on events that occurred in past periods Results suggest that a pproximately 81% of the demand from the West region is driven by current events and the remaining 19 % depends on events from past per iods. The Midwest and South region s yielded higher estimates for the elasticity of adjustment than the Northeast and Midwes t regions Both U.S. regions adjust more rapidly to current events (0.85 and 0.86, respectively) than to past events. Analysis of coefficients in static component All estimates of the short term elasticities of airfare income and advertisi ng at the curr ent period have correct sign and are significant in all U.S. regions For example, the airfare elasticity of demand is 0.194 for the Northeast region and can be interpreted as follows. A percentage increase in the price of airline tickets from the Northea st region to the South Florida CSA represents a 0.194 % decrease in demand for air passenger traffic Income has a positive effect and is significant in all four U.S. regions while advertising expenditure coefficient estimates are not significant in any re gion. The weather var iables related to Florida, HC ATHGH, HCATMED, and FIRE, have correct sign (negative) and are significant in most cases. However, HCATLOW behaves as expected in

PAGE 176

176 the West region only Nevertheless, estimates from all U.S. regions suggest that tropical storms and depressions have no effect o n demand Hurricanes have a negative impact o n demand from all four U.S. regions at a significant level. The HCATMED from the West region is the exception. Wildf ires a lso have a negative impact on dema nd from the Midwest, South, and West region s Demand from the Northeast region is not affected by wild fires in Florida. Most of the dummies for terror (TER2 TER7) have a negative effect o n demand from all U.S. regions But coefficient estimates suggest tha t the impact has been decreasing through time. Meanwhile, the effect of the 9 11 terrorist attacks o n demand from the Northeast and West region s is negative but since September 2003 is no longer significant. Figure 5 2 exhibits the e ffect of the terrorism dummies o n demand for a ir p assengers traveling to the South Florida CSA Note that the g ray colored cylinders indicate that the estimate is not significant with 95 % confidence. Orlando CSA Results Table 5 4 shows the coefficient estimates and their corresp onding t values, R squared, and number of observations of the SUR AR1 ALL model for the four U.S regions tra veling to the Orlando CSA. E lasticity of adjustment estimates are less than one but only significant in t he Northeast and Midwest region s. These res ults suggest that current demand s for air passengers from the South and West region s traveling to the Orlando CSA do not respond to events that occurred in past periods. In other words, their demand is in long run equilibrium because the response is immedi ate Analysis of coefficients in dynamic component The elasticity of adjustment estimate from the Northeast region sugg ests that for a one percentage increas e in past demand, current demand for air passengers on the average increases

PAGE 177

177 by 0.41. Fifty nine p ercent of the adjustment occurs in the current period, while the remaining portion of the adjustment dep ends on events that occurred in the past Results suggest that approximately 79 % of the demand from the Midwest region is driven by current events and the remaining 21 % depends on events from past periods The W est and South region s yielded higher estimates for the elasticity of adjustment than the Northeast and Midwest regions. Also, the estimate 29 from the West and South region s were statistically not different from zero In other words, b oth regions adjust instantly to current events. Analysis of coefficients in static component S hort term airfare elasticities of demand are not significant in any region traveling to the Orlando CSA, while incom e elasticity estimat es at the current period have the correct sign and are significant in all U.S. regions The income elasticity of demand is 2.364 for the West region which sugges ts that a percentage i ncrease in income from the West region represents a 2.4 % increase in demand for air passenger traffic to the Orlando CSA. All estimates of the advertising expenditures have the wrong sign and were significant in three of the four U.S. regions W eather var iables related to Florida, H CATHGH, HCATMED, and FIRE, have correct sign (negative) and are significant in most cases. Howev er, HCATLOW behaves as expected i n the West region only Hurricanes of higher categories ( i.e., HCATHGH) have a negative impact o n d emand from all U.S. regions while hurricanes of category 1 and 2 have a negative impact o n demand from the Midwest region only. Estimates for FIRE indicate that wildfires affecting Florida have an impact on demand from the South region only. All dummies f or terror (TER2 TER7) have a negative and significant effect o n demand from all U.S. regions But coefficient estimates suggest that the impact has been decreasing through time. The e ffect of the terrorism dummies o n demand for a ir p assengers traveling to the

PAGE 178

178 Orlando CSA is shown in Figure 5 3. Note that all estimates are significant with 95 % confidence. Tampa St. Petersburg CSA Results C oefficient estimates and their corresponding t values, R squared, and number of observations of the SUR AR1 ALL model for four U.S regions traveling to the Tampa St. Petersburg CSA are presented in Table 5 5. E lasticity of adjustment estimates are less than one but only one of them is significant. These results suggest that c urrent demand for air passengers from all U.S. reg ions but the Northeast region do not depend on events that occurred in past periods. In other words, there is no rigidity in the demand for air passengers from the Midwest, South, and West regions. Analysis of coefficients in dynamic component The elasti city of adjustment estimate is not significant in three U.S. regions indicating that demand to the Tampa St. Pete rsburg CSA is solely determined by current events C onsumers traveling from the Midwest, South, and West regions instantly adjust to any change s in the drivers of demand for air passengers to the Tampa St. Petersburg CSA Conversely, demand for air passengers from the Northeast region respond s to past events The elasticity of adjustment estimate from the Northeast region suggests that for a one percentage increase in past demand, current demand for air passengers on the average increases by 0.16 % In other words, e ighty four percent of the adjustment occurs in the current period, while the remaining portion of the adjustment depends on past event s Analysis of coefficients in static component E stimates of the short term income elasticities of dem and at the current period have correct sign and are significan t in each of the four U.S. regions S hort term airfare elasticities of demand are significa nt in two of the four U.S. regions For example, the estimate of the airfare elasticity

PAGE 179

179 of demand for the South region is 0.159 suggesting that a percentage increase in the average price of an airline ticket from the South region represents a 0.159 % decre ase in demand for air passenger traffic to the Tampa St. Petersburg CSA. Most of the estimates of the advertising expenditures have incorrect sign and one of them is significant. W eather variables related to Florida, HCATHGH, HCATMED, and FIRE, have the c orrect sign (negative) and are significant in some U.S. regions Howev er, HCATLOW behaves as expected in the West region only Hurricanes of higher category have a negative impact o n demand from all U.S. regions but only estimates from the South and West r egions were significant. The Midwest region yielded significant values for hurricane s of lower categories, while estimates of the tropical storm variables related to the Northeast region were not significant. Estimates for FIRE indicate that wildfires affe cting Florida do have a negative impact on demand from any region, but it is only significant in the South region. Similarly to the Orlando CSA, virtually all dummies for terror (TER2 TER7) have a nega tive and significant impact on demand from every region to the Tampa St. Petersburg CSA Figure 5 4 presents the e ffect of the terrorism dummies o n demand for a ir p assengers traveling to the Tampa St. Petersburg CSA. Jacksonville CSA Results Table 5 6 shows coefficient estimates and their corresponding t value s, R square, and number of observations of the S UR AR1 ALL model for four U.S regions trav eling to the Jacksonville CSA. E lasticity of adjustment estimates are less than one in all four U.S. regions R esults in the Midwest and South regions suggest that th e current demand for air passengers traveling to the Jacksonville CSA depend on both current and past events. Elasticity of adjustment e stimate s from the Northeast and West region s are not significant, indicating that demand from these U.S. regions do es no t depend on events that occurred in past periods.

PAGE 180

180 Analysis of coefficients in dynamic component The elasticity of adjustment estimate from the Midwest region is statistically different from zero indicating that demand to the Jacksonville CSA is determined by current and past events. The elasticity of adjustment estimate suggests that for a one percentage increase in past demand, current demand for air passengers on the average increases by 0.32 % In other words, nearly two thirds of the adjustment occurs i n the current period, while the remaining portion of the adjustment depends on events that occurred the past Results suggest that approximately 72 % of the demand from the South region is driven by current events and the remaining 28 % depe nds on events fr om past periods. Conversely, the estimate from the Northeast and West regions suggests that demand from these two U.S. regions is not driven by past events. Analysis of coefficients in static component E stimates of the short term elasticities of airfare income and advertising at the current period have correct sign but are significant in some U.S. regions only. E stimates of the short term income elasticities of demand are significant in three of the four U.S. regions : Northeast, Midwest, and South regio ns. Short term price elasticities of demand are significant in the Northeast and West regions only. For example, the estimate of the airfare elasticity of demand for the West region is 1.177 suggesting that a percentage increase in the price of airline ti cket from the West region represents a 1.18 % decrease in demand for air passenger traffic to the Jacksonville CSA. Estimates of the advertising expenditures at the current period (ADV) are not significant in any of the four U.S. regions Most of the estim ates from the weather variables related to Florida have either incorrect sign (positive) or are not significant. However, HCATMED from the Midwest region is negative and significant suggesting that hurricanes of category 1, 2 and 3 have a negative impact on

PAGE 181

181 demand to the Jacksonville CSA E stimates for FIRE and HCATHGH from the South region are both negative a nd significant suggesting that demand for air passengers from this U.S. region is adversely affected by hurricanes of higher categories and wildfires E stimates of dummies for terror (TER2 TER7) have a negative and significant impact on the demand from three of the four U.S. regions Although most of the estimates from the West region are negative, they are not significant. The effect of the terrorism dummies o n demand for a ir p assengers traveling to the Jacksonville CSA is presented in Figure 5 5 Note that all coefficients from the West region (gray colored cylinders ) indicate that estimates are not significant with 95 % confidence Fort Myers CSA Resu lts C oefficient estimates and their corresponding t values, R square, and number of observations of the SUR AR1 ALL model for the four U.S regions traveling to the Fort Myers CSA are exhibited in Table 5 7. E lasticity of adjustment estimates are less than one in three of the four U.S. regions The coefficient estimate from the West region is negative and hence, unstable. Nevertheless, the estimate is not statistically significant. The estimate from the Northeast region exhibits a stable estimate at a signif icant level. These results suggest that current demand for air passengers from the Northeast region traveling to the Fort Myers CSA respond s to events that occurred in past periods. Conversely, there is no habit persistence in the demand for air passengers from the Midwest, South, and West regions. Analysis of coefficients in dynamic component The elasticity of adjustment estimate is not significant in three U.S. regions and indicates th at demand to the Fort Myers CSA is solely determined by current events It also indicates that consumers traveling from the Midwest, South, and West regions instantly adjust to any changes in the drivers of demand for air passengers Conversely, demand for air passengers from the

PAGE 182

182 Northeast region respond s to past events, esp ecially to those events that occurred twelve months ago. The elasticity of adjustment estimate from the Northeast region suggests that for a one percentage increase in previous demand, current demand for air passengers increases by 0.27 % Seventy three per cent of the adjustment occurs in the current period, while the remaining portion of the adjustment depends on events that occurred in past periods Analysis of coefficients in static component Estimates of the short term airfare elasticities from the Nor theast and West region s have a negative impact o n demand. Estimates for the advertising expenditures are not significant in three U.S. regions while the estimate for the advertising expenditures from the Northeast region has the incorrect sign and it is s ignificant. E stimates of the short term elasticities of income are significant in three of four U.S. regions : Northeast, Midwest, and South regions. The estimate of the income elasticity of demand f r om the South region is 1.413 suggesting that a percentage increase in personal disposable income from the South region represents a 1.41 % increase in demand for air passengers to the Fort Myers CSA. Only a few estimates from the weather var iables related to Florida have correct sign (negative) and are significa nt. FIRE and HCATMED estimates from the Midwest region are negative and significant suggesting that wildfires and hurricanes of category 1, 2, and 3 have a negative impact o n demand to the Fort Myers CSA The HCATHGH estimates from the Northeast and South regions are also negative suggesting that hurricanes of cate gory 4 and 5 affect negatively demand for air passengers to the Fort Myers CSA Wildf ires also have a negative effect on d emand for air passengers from the South region. E stimates of the dummies f or terror (TER2 TER7) have a negative sign but only a few are significant. Estimates from the West region are positiv e and significant which contradicts theoretical expectations established earlier. The effect of the terrorism dummies o n demand for

PAGE 183

183 a ir p as sengers traveling to the Fort Myers CSA is presented in Figure 5 5 Note that the gray colored cylinders indicate that estimates are not significant with 95 % confidence Chapter Summary This chapter discussed the approach chosen to estimate the model prese nted in Equa tion 4 13 of Chapter 4. The SUR AR1 ALL estimation approach performed better than the other three 29 coefficient estimates. Results were presented and discussed for each destination CSA and for Florida (CSA=33). The discussion focused on statistical significance of various estimates such as the elasticity of adjustment estimates and the economic, weather, and terrorism variables. Es timates of the dummies controlling for monthly seasonality sugges ted that there are strong seasonal patterns in each CSA ORG pair and that the inclusion of these dummy variables was necessary Results also showed that the advertising expenditure estimates were not well behaved in almost all CSA ORG pairs. Since advertis ing expenditures are highly seasonal, the dummy variables for monthly seasonality may be capturing the effect of the advertising variable.

PAGE 184

184 Figure 5 1 Florida CSA: e ffect of terrorism dummies o n demand for airline p assengers from four U.S. regions Note: Gray colored cylinders indicate estimate not significant at 95 % confidence. Figure 5 2 South Florida CSA: effect of terrorism dummies on demand for airline passengers from four U.S. regions. No te: Gray colored cylinders indicate estimate not significant at 95% confidence. 0.400 0.300 0.200 0.100 0.000 0.100 Northeast Midwest South West TER2 TER2 TER2 TER2 TER4 TER4 TER4 TER4 TER6 TER6 TER6 Coefficient Estimate U.S. Region 0.400 0.300 0.200 0.100 0.000 0.100 Northeast Midwest South West TER2 TER2 TER2 TER2 TER4 TER4 TER4 TER4 TER6 TER6 TER6 Coefficient Estimate U.S. Region

PAGE 185

185 Figure 5 3 Orlando CSA: effect of terrorism dummies on demand for airline passengers from four U.S. regions. Figure 5 4 Tampa St. Petersburg CSA: effect of terrorism dummies on demand for airline passengers from four U.S. regions. Note: Gray colored cylinders indicate estimate not significant at 95% confidence. 0.400 0.300 0.200 0.100 0.000 0.100 Northeast Midwest South West TER2 TER2 TER2 TER2 TER4 TER4 TER4 TER4 TER6 TER6 TER6 TER6 Coefficient Estimate U.S. Region 0.600 0.500 0.400 0.300 0.200 0.100 0.000 0.100 Northeast Midwest South West TER2 TER2 TER2 TER2 TER4 TER4 TER4 TER4 TER 6 TER6 TER6 TER6 Coefficient Estimate U.S. Region

PAGE 186

186 Figure 5 5 Jacksonvill e CSA: effect of terrorism dummies on demand for airline passengers from four U.S. regions. Figure 5 6 Fort Myers CSA: effect of terrorism dummies on demand for airline passengers from four U.S. regions. 0.600 0.500 0.400 0.300 0.200 0.100 0.000 0.100 Northeast Midwest South West TER2 TER2 TER2 TER2 TER4 TER4 TER4 TER4 TER 6 TER6 TER6 TER6 Coefficient Estimate U.S. Region 2.000 1.000 0.000 1.000 2.000 3.000 4.000 Northeast Midwest South West TER2 TER2 TER2 TER2 TER4 TER4 TER4 TER4 TER 6 TER6 TER6 TER6 Coefficient Estimate U.S. Region

PAGE 187

187 Table 5 1 Coefficient estimates 29 and their corresponding t value using five different estimation approaches Estimation Approach OLS AR1 SUR AR1 ALL SUR AR1 CSA SUR AR1 ORG CSA ORG Pair Estimate T value Estimate T value Estimate T value Estimate T value Estimate T value 1 1001 0.134 1.037 0.114 1.069 0.237 3.667 0.297 3.564 0.087 1.070 1 1002 0.086 0.446 ^ 0.096 0.925 ^ 0.151 2.434 0.146 1.815 0.077 0.928 1 1003 0.104 0.739 ^ 0.066 0.660 ^ 0.137 2.490 0.203 2.851 0.065 0.902 1 1004 0.050 0.387 0.093 0.925 0.186 2.229 0.248 2.707 0.002 0.017 5 1001 0.194 1.299 0.188 1.782 0.409 5.558 0.377 4.270 0.372 4.106 5 1002 0.210 1.421 0.133 1.265 0.215 2.880 0.168 1.805 0.187 1.988 5 1003 0.102 0.678 ^ 0.070 0.707 0.052 0.750 0.078 0.873 0.027 0.027 ^ 5 100 4 0.262 1.938 0.188 1.917 0.133 1.510 0.397 3.895 0.127 1.164 ^ 8 1001 0.266 1.969 ^ 0.140 1.094 ^ 0.163 2.278 0.163 1.731 0.090 1.001 8 1002 0.111 0.679 0.038 0.342 ^ 0.071 0.948 0.053 0.536 0.017 0.197 8 1003 0.015 0.142 0.046 0.492 0.130 1.8 60 0.192 2.250 0.133 1.691 8 1004 0.124 0.819 0.161 1.534 0.025 0.206 0.017 0.122 ^ 0.039 0.272 9 1001 0.029 0.221 0.040 0.349 0.042 0.309 0.183 1.292 ^ 0.005 0.035 9 1002 0.034 0.239 ^ 0.274 2.475 0.319 3.559 0.331 3.117 0.497 4.870 9 1003 0 .068 0.533 0.151 1.440 0.278 3.525 0.246 2.634 0.288 3.118 9 1004 0.104 0.785 0.115 1.102 0.245 1.919 0.094 0.660 0.272 1.978 14 1001 0.197 1.651 0.287 2.552 0.268 2.454 0.142 1.015 0.212 1.756 14 1002 0.032 0.265 0.041 0.370 ^ 0.157 1.344 0.00 2 0.011 0.058 0.489 14 1003 0.138 1.036 0.197 1.933 0.172 1.811 0.218 1.804 0.168 1.694 14 1004 0.246 1.790 ^ 0.135 1.302 ^ 0.211 1.609 ^ 0.061 0.450 ^ 0.085 0.614 ^ Number of Unstable Estimates 6 6 1 3 3 indicates significance at 95% confid ence using a one tailed test. ^ indicates coefficient estimate is unstable (negative sign).

PAGE 188

188 Table 5 2 Florida CSA: c oefficient estimates and their corresponding t values for the demand for air passengers traveling from four U. S regions using the SUR AR1 ALL approach Northeast (1001) Midwest (1002) South (1003) West (1004) Estimate T value Estimate T value Estimate T value Estimate T value C 5.383 2.601 8.130 3.687 4.112 2.513 8.904 3.429 FARE 0.249 3.318 ^ 0.055 0.726 0.246 3.852 ^ 0.106 1.090 INC 1.011 4.465 ^ 1.420 5.773 ^ 1.008 5.483 ^ 1.253 4.324 ^ TEMP 0.025 0.846 0.013 0.742 0.058 1.805 0.080 0.927 PCP 0.002 0.855 0.000 0.034 0.004 1.676 0.002 1.816 FIRE 0.004 0.522 0.010 1.575 0.013 2.243 ^ 0.007 1.0 26 HCATLOW 0.005 0.707 0.013 1.788 ^ 0.007 1.068 0.005 0.661 HCATMED 0.024 1.924 ^ 0.046 3.732 ^ 0.027 2.533 ^ 0.011 0.901 HCATHGH 0.054 2.972 ^ 0.044 2.558 ^ 0.051 3.271 ^ 0.053 3.108 ^ ADV 0.008 0.484 0.001 0.037 0.014 1.033 0.007 0.428 TE R2 0.278 8.016 ^ 0.265 7.308 ^ 0.280 9.311 ^ 0.215 6.133 ^ TER3 0.152 3.523 ^ 0.325 6.725 ^ 0.173 4.531 ^ 0.167 3.664 ^ TER4 0.070 1.472 0.306 5.571 ^ 0.128 3.131 ^ 0.063 1.252 TER5 0.081 1.511 0.344 5.527 ^ 0.170 3.643 ^ 0.142 2.557 ^ TER6 0.074 1.305 0.359 5.364 ^ 0.174 3.570 ^ 0.074 1.258 TER7 0.089 1.413 0.324 4.511 ^ 0.171 3.181 ^ 0.078 1.218 MTH2 0.029 1.554 0.089 4.196 0.028 1.969 0.084 4.937 MTH3 0.128 4.984 0.276 9.276 0.166 8.262 0.054 2.629 MTH4 0.0 56 2.002 0.015 0.457 0.056 2.677 0.068 2.665 MTH5 0.090 3.203 0.204 6.056 0.021 0.982 0.121 4.216 MTH6 0.139 4.440 0.185 5.141 0.014 0.624 0.076 2.874 MTH7 0.063 1.661 0.181 3.912 0.010 0.332 0.085 2.463 MTH8 0.044 1.1 57 0.257 6.470 0.048 1.740 0.090 2.791 MTH9 0.327 8.917 0.474 12.27* 0.216 7.813 0.290 8.233 MTH10 0.097 3.836 0.155 5.299 0.027 1.203 0.158 5.630 MTH11 0.014 0.420 0.062 1.876 0.002 0.087 0.132 4.505 MTH12 0.019 0 .677 0.052 2.144 0.053 2.612 0.025 1.001 2 7 0.08 1.542 0.095 1.955 0.045 1.006 0.019 0.328 2 8 0.06 1.157 0.073 1.506 0.022 0.499 0.189 3.477 29 0.35 4.267 0.154 1.808 ^ 0.208 2.779 0.363 3.733 0.648 12.87* 0.713 16.19* 0.668 14.50* 0.684 1 1.74 0.651 0.846 0.792 0.637 W 2 0.224 0.614 0.215 0.053 W 4 0.163 0.476 0.104 0.521 W 12 0.613 0.090 0.682 0.426 R 2 0.963 0.963 0.948 0.970 N=146 indicates significance at 95% confidence using a two tailed te st. ^ indicates significance at 95% confidence using a one tailed test.

PAGE 189

189 Table 5 3 South Florida CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U.S regions using the SUR AR1 ALL approach Northeast (1001) Midwest (1002) South (1003) West (1004) Estimate T value Estimate T value Estimate T value Estimate T value C 3.574 2.224 4.960 2.811 4.475 3.129 7.344 2.412 FARE 0.194 3.032 ^ 0.108 2.403 ^ 0.148 3.281 ^ 0.103 1.116 INC 0.873 5.380 ^ 1.023 5.573 ^ 1.023 6.808 ^ 1.137 3.651 ^ TEMP 0.021 0.633 0.007 0.401 0.065 1.986 0.144 1.532 PCP 0.002 0.665 0.005 1.703 0.004 1.530 0.000 0.052 FIRE 0.007 0.847 0.018 2.430 ^ 0.017 2.583 ^ 0.014 1 .794 ^ HCATLOW 0.010 1.137 0.012 1.520 0.011 1.616 0.001 0.147 HCATMED 0.031 2.154 ^ 0.064 4.881 ^ 0.037 3.226 ^ 0.017 1.237 HCATHGH 0.058 2.709 ^ 0.060 3.162 ^ 0.052 3.046 ^ 0.052 2.598 ^ ADV 0.002 0.650 0.001 0.218 0.003 1.052 0.005 0.9 51 TER2 0.199 6.385 ^ 0.213 6.422 ^ 0.236 8.567 ^ 0.139 3.324 ^ TER3 0.134 3.619 ^ 0.290 6.810 ^ 0.208 6.082 ^ 0.174 3.052 ^ TER4 0.038 0.947 0.277 5.861 ^ 0.155 4.173 ^ 0.034 0.503 TER5 0.028 0.618 0.343 6.349 ^ 0.192 4.608 ^ 0.095 1.257 TER6 0.046 0.959 0.379 6.436 ^ 0.218 4.904 ^ 0.009 0.114 TER7 0.066 1.224 0.367 5.695 ^ 0.202 4.082 ^ 0.042 0.466 MTH2 0.002 0.109 0.038 2.053 0.001 0.090 0.126 7.047 MTH3 0.096 3.882 0.193 7.497 0.151 8.141 0.040 1.712 MTH4 0.0 04 0.147 0.102 3.329 0.033 1.553 0.056 1.942 MTH5 0.193 6.527 0.351 10.90* 0.029 1.365 0.154 4.894 MTH6 0.256 8.009 0.389 11.30* 0.052 2.307 0.093 3.020 MTH7 0.128 3.724 0.372 9.333 0.037 1.461 0.050 1.433 MTH8 0.1 04 3.143 0.396 11.39* 0.095 4.023 0.050 1.512 MTH9 0.461 13.85* 0.638 18.36 0.322 12.86* 0.389 11.14* MTH10 0.203 7.968 0.326 11.9 7 0.105 4.909 0.241 7.998 MTH11 0.053 1.737 0.125 4.287 0.002 0.086 0.109 3.705 M TH12 0.030 1.231 0.059 2.836 0.075 4.539 0.013 0.522 27 0.110 2.654 0.099 2.688 0.123 3.647 0.117 2.305 28 0.068 1.711 ^ 0.124 3.420 0.024 0.733 0.121 2.400 29 0.237 3.667 0.151 2.434 0.137 2.490 0.186 2.229 0.543 12.93* 0.639 16.69* 0. 614 17.05* 0.758 15.44* 0.763 0.849 0.863 0.814 W 2 0.463 0.657 0.897 0.626 W 4 0.286 0.820 0.175 0.650 W 12 0.251 0.477 0.073 0.276 R 2 0.942 0.963 0.927 0.949 N=146 indicates significance at 95% confidence using a tw o tailed test. ^ indicates significance at 95% confidence using a one tailed test.

PAGE 190

190 Table 5 4 Orlando CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U.S regio ns using the SUR AR1 ALL approach Northeast (1001) Midwest (1002) South (1003) West (1004) Estimate T value Estimate T value Estimate T value Estimate T value C 8.654 4.050 10.509 4.890 8.659 4.290 19.331 5.850 FARE 0.095 1.422 0.007 0.094 0. 064 1.119 0.059 0.606 INC 1.225 5.211 ^ 1.543 6.536 1.464 6.800 ^ 2.364 6.748 ^ TEMP 0.037 1.106 0.003 0.149 0.056 1.473 0.211 2.060 PCP 0.002 0.771 0.002 0.507 0.003 0.892 0.004 2.964 FIRE 0.002 0.207 0.004 0.577 0.013 1.897 ^ 0.002 0.220 H CATLOW 0.006 0.742 0.027 3.219 ^ 0.010 1.335 0.000 0.027 HCATMED 0.019 1.408 0.031 2.245 ^ 0.016 1.304 0.004 0.300 HCATHGH 0.042 2.097 ^ 0.048 2.450 ^ 0.040 2.162 ^ 0.049 2.431 ^ ADV 0.013 1.192 0.029 2.850 ^ 0.015 1.808 ^ 0.038 2.941 ^ TER2 0.269 8.010 ^ 0.232 6.448 ^ 0.278 8.078 ^ 0.292 7.053 ^ TER3 0.179 4.119 ^ 0.314 6.796 ^ 0.258 5.714 ^ 0.330 5.965 ^ TER4 0.099 2.148 ^ 0.270 5.131 ^ 0.206 4.098 ^ 0.252 4.027 ^ TER5 0.129 2.593 ^ 0.305 5.217 ^ 0.225 3.998 ^ 0.327 4.728 ^ TER6 0.124 2.287 ^ 0.298 4.723 ^ 0.211 3.475 ^ 0.309 4.045 ^ TER7 0.186 3.007 ^ 0.294 4.324 ^ 0.219 3.249 ^ 0.356 4.209 ^ MTH2 0.060 3.131 0.079 3.995 0.049 3.068 0.069 3.877 MTH3 0.181 7.074 0.225 9.260 0.206 9.835 0.105 4.850 MTH4 0.141 5.506 0.074 2.783 0.095 4.265 0.036 1.419 MTH5 0.027 1.077 0.013 0.463 0.080 3.305 0.069 2.408 MTH6 0.027 0.976 0.036 1.175 0.087 3.403 0.031 1.089 MTH7 0.057 1.682 0.027 0.733 0.086 2.933 0.066 1.999 MTH8 0.057 1.712 0.101 2.999 0.037 1.363 0.132 4.147 MTH9 0.245 7.057 0.299 8.417 0.238 8.202 0.321 8.984 MTH10 0.010 0.388 0.010 0.339 0.033 1.307 0.113 3.925 MTH11 0.057 1.906 0.025 0.883 0.023 0.954 0.147 5.468 MTH12 0.006 0.272 0.012 0.585 0.038 2.180 0.042 1.990 27 0.136 3.111 0.051 1.193 0.059 1.509 0.004 0.086 28 0.048 1.149 0.020 0.457 0.003 0.069 0.139 2.909 29 0.409 5.558 0.215 2.880 0.052 0.750 0.133 1.510 0.598 13.04* 0.650 15.96* 0.701 19.10* 0.700 14.78* 0.591 0.785 0.948 0.867 W 2 0.333 0.237 1.123 0.034 W 4 0.118 0.091 0.049 1.046 W 12 0.550 0.672 0.074 0.079 R 2 0.957 0.937 0.926 0.953 N=146 indicates significance at 95% confidence using a two tailed test. ^ indicate s significance at 95% confidence using a one tailed test.

PAGE 191

191 Table 5 5 Tampa St. Petersburg CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U.S regions using the SUR AR1 ALL approach Northeast (1001) Midwest (1002) South (1003) West (1004) Estimate T value Estimate T value Estimate T value Estimate T value C 14.351 7.455 12.970 6.204 9.928 6.149 32.606 6.732 FARE 0.054 0.983 0.143 2.113 ^ 0.159 2.989 ^ 0.107 0.815 INC 1.856 8.917 ^ 1.808 7.918 ^ 1.497 8.420 ^ 3.608 7.063 ^ TEMP 0.024 0.739 0.028 1.150 0.041 0.969 0.229 1.544 PCP 0.001 0.344 0.002 0.407 0.008 2.438 0.004 2.236 FIRE 0.008 0.969 0.008 0.875 0.013 1.979 ^ 0.014 1.365 HCATLOW 0.011 1.301 0.026 2.806 ^ 0.010 1.359 0.003 0.276 HCATMED 0.015 1.086 0.028 1.886 ^ 0.018 1.568 0.014 0.783 HCATHGH 0.031 1.510 0.033 1.485 0.032 1.843 ^ 0.045 1.762 ^ ADV 0.008 1.337 0.008 1.324 0.001 0.207 0.030 2.4 12 ^ TER2 0.202 6.576 ^ 0.235 6.380 ^ 0.262 8.671 ^ 0.157 3.293 ^ TER3 0.179 4.777 ^ 0.355 7.598 ^ 0.221 5.809 ^ 0.164 2.755 ^ TER4 0.065 1.629 0.327 6.231 ^ 0.192 4.708 ^ 0.101 1.581 TER5 0.071 1.594 0.356 6.182 ^ 0.210 4.595 ^ 0.136 1.925 ^ TER6 0.068 1.434 0.433 6.973 ^ 0.275 5.667 ^ 0.173 2.238 ^ TER7 0.119 2.178 ^ 0.409 5.895 ^ 0.285 5.215 ^ 0.214 2.372 ^ MTH2 0.090 4.935 0.149 7.110 0.047 3.185 0.023 0.980 MTH3 0.257 10.29* 0.384 13.25* 0.228 10.57* 0.200 6.455 MTH4 0.162 6.541 0.121 4.017 0.094 4.378 0.005 0.154 MTH5 0.000 0.010 0.082 2.638 0.039 1.719 0.075 1.868 MTH6 0.026 0.960 0.043 1.315 0.032 1.359 0.045 1.200 MTH7 0.092 2.882 0.022 0.545 0.032 1.184 0.025 0.570 MTH8 0.070 2.330 0.137 4.122 0.037 1.537 0.095 2.569 MTH9 0.256 8.756 0.378 11.26* 0.196 7.939 0.337 8.055 MTH10 0.005 0.187 0.077 2.707 0.007 0.323 0.152 4.238 MTH11 0.065 2.363 0.003 0.097 0.039 1.772 0.142 3.844 MTH12 0.040 1.922 0.115 5.426 0.078 4.881 0.007 0.290 27 0.146 3.469 0.095 2.247 0.088 2.118 0.006 0.078 28 0.014 0.355 0.009 0.220 0.012 0.302 0.116 1.817 ^ 29 0.163 2.278 0.071 0.948 0.130 1.860 ^ 0.025 0.206 0.544 12.81* 0.594 14.14 0.636 14.44* 0.637 10.8 3* 0.837 0.929 0.870 0.975 W 2 0.897 1.333 0.678 0.223 W 4 0.088 0.124 0.091 4.581 W 12 0.191 0.457 0.231 3.359 R 2 0.970 0.942 0.943 0.976 N=146 indicates significance at 95% confidence using a two tailed test. ^ ind icates significance at 95% confidence using a one tailed test.

PAGE 192

1 92 Table 5 6 Jacksonville CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U.S regions using the S UR AR1 ALL approach Northeast (1001) Midwest (1002) South (1003) West (1004) Estimate T value Estimate T value Estimate T value Estimate T value C 15.938 2.697 9.923 3.783 13.188 6.798 13.900 0.983 FARE 0.311 2.427 ^ 0.044 0.489 0.019 0.308 1.177 1.792 ^ INC 1.871 3.116 ^ 1.211 4.283 ^ 1.657 7.502 ^ 1.421 1.013 TEMP 0.132 2.196 0.034 1.058 0.059 1.302 0.099 0.123 PCP 0.000 0.059 0.009 1.662 0.001 0.415 0.008 0.818 FIRE 0.008 0.827 0.006 0.662 0.017 2.526 ^ 0.071 1.5 06 HCATLOW 0.023 2.201 ^ 0.033 3.202 ^ 0.016 2.177 ^ 0.015 0.294 HCATMED 0.002 0.095 0.038 2.230 ^ 0.014 1.179 0.072 0.901 HCATHGH 0.010 0.387 0.023 0.947 0.030 1.667 ^ 0.037 0.318 ADV 0.034 1.643 0.026 1.486 0.014 1.284 0.101 0.996 TER2 0.323 5.342 ^ 0.277 6.557 ^ 0.300 9.914 ^ 0.117 0.592 TER3 0.380 4.353 ^ 0.337 6.080 ^ 0.253 6.349 ^ 0.190 0.805 TER4 0.270 2.558 ^ 0.408 6.316 ^ 0.197 4.649 ^ 0.352 1.361 TER5 0.179 1.488 0.428 5.457 ^ 0.255 5.274 ^ 0.185 0.641 TER6 0. 090 0.657 0.346 4.482 ^ 0.293 5.620 ^ 0.173 0.561 TER7 0.015 0.095 0.324 4.086 ^ 0.321 5.610 ^ 0.225 0.663 MTH2 0.119 4.954 0.105 4.922 0.028 1.902 0.086 1.023 MTH3 0.319 8.375 0.378 11.37* 0.228 10.27* 0.159 1.485 MTH4 0.292 7.498 0.257 8.405 0.163 7.443 0.151 1.217 MTH5 0.227 6.037 0.179 5.786 0.149 6.495 0.079 0.570 MTH6 0.208 5.206 0.207 6.172 0.147 5.964 0.048 0.311 MTH7 0.234 4.723 0.238 5.882 0.165 5.713 0.109 0.646 MTH8 0.250 4.969 0.167 4.362 0.109 3.965 0.1 04 0.585 MTH9 0.017 0.381 0.049 1.360 0.064 2.400 0.439 2.392 MTH10 0.185 4.515 0.180 5.109 0.101 4.009 0.495 3.260 MTH11 0.194 5.350 0.190 5.502 0.101 4.283 0.368 2.806 MTH12 0.215 6.109 0.194 6.384 0.113 5.532 0.074 0.550 27 0.012 0.193 0.192 3.376 0.103 2.216 0.125 1.359 28 0.016 0.256 0.085 1.578 0.077 1.796 ^ 0.204 2.594 29 0.042 0.309 0.319 3.559 0.278 3.525 0.245 1.919 ^ 0.826 16.34* 0.555 9.740 0.596 12.97* 0.585 7.589 0.958 0.681 0.722 0.755 W 2 0.294 0.602 0.370 0.510 W 4 0.375 0.266 0.276 0.830 W 12 0.919 0.132 0.354 0.340 R 2 0.947 0.917 0.958 0.840 N=146 indicates significance at 95% confidence using a two tailed test. ^ indicates significance at 95% confidence using a one tailed test.

PAGE 193

193 T able 5 7 Fort Myers CSA: coefficient estimates and their corresponding t values for the demand for air passengers traveling from four U.S regions using the SUR AR1 ALL approach. Northeast (1001) Midwest (1002) South (1003) West (1004) Estimate T value Estimate T value Estimate T value Estimate T value C 13.315 3.948 12.005 5.016 10.596 4.083 38.241 1.432 FARE 0.402 3.120 ^ 0.139 1.279 0.0 39 0.489 2.508 2.265 ^ INC 1.692 4.827 ^ 1.604 5.668 ^ 1.413 5.145 ^ 3.911 1.478 TEMP 0.025 0.372 0.035 1.008 0.003 0.053 0.588 0.423 PCP 0.002 0.358 0.006 0.911 0.000 0.070 0.002 0.091 FIRE 0.007 0.625 0.020 2.009 ^ 0.014 1.711 ^ 0.105 1.229 HCATLOW 0.012 0.979 0.024 2.336 ^ 0.004 0.496 0.062 0.669 HCATMED 0.016 0.792 0.045 2.536 ^ 0.014 0.959 0.049 0.334 HCATHGH 0.076 2.620 ^ 0.012 0.479 0.041 1.857 ^ 0.047 0.225 ADV 0.044 2.054 ^ 0.012 0.665 0.018 1.364 0.208 1.177 TER 2 0.317 5.835 ^ 0.210 5.492 ^ 0.168 4.113 ^ 0.437 1.108 TER3 0.159 2.419 ^ 0.224 4.796 ^ 0.072 1.316 1.640 3.183 ^ TER4 0.034 0.461 0.182 3.568 ^ 0.037 0.590 2.662 4.481 ^ TER5 0.135 1.544 0.108 2.006 ^ 0.038 0.545 3.207 4.786 ^ TER6 0.103 1. 057 0.144 2.607 ^ 0.054 0.738 3.415 4.730 ^ TER7 0.095 0.880 0.102 1.653 0.068 0.828 3.714 4.652 ^ MTH2 0.050 1.527 0.145 3.507 0.106 4.218 0.016 0.100 MTH3 0.163 3.392 0.359 5.868 0.258 7.661 0.369 1.756 MTH4 0.022 0.396 0.019 0.268 0.057 1.442 0.161 0.623 MTH5 0.297 4.553 0.599 6.772 0.129 2.766 0.641 2.290 MTH6 0.436 6.164 0.682 7.437 0.206 4.079 0.941 2.987 MTH7 0.493 6.116 0.628 5.954 0.189 3.222 1.118 3.069 MTH8 0.508 6.634 0.720 7.610 0 .279 5.347 0.958 2.658 MTH9 0.768 10.48* 0.919 10.63* 0.430 8.656 1.731 4.849 MTH10 0.354 7.610 0.421 7.072 0.119 3.287 0.963 3.624 MTH11 0.248 4.726 0.144 2.270 0.059 1.525 0.248 1.032 MTH12 0.171 4.334 0.001 0.038 0.009 0.382 0.356 1.530 27 0.047 0.729 0.128 1.832 ^ 0.044 0.748 0.020 0.218 28 0.083 1.425 0.024 0.384 0.017 0.308 0.061 0.766 29 0.268 2.454 0.157 1.344 0.172 1.811 ^ 0.211 1.609 0.653 11.36* 0.495 7.146 0.698 13.88* 0.668 8.557 0.732 0.843 0.828 1.211 W 2 0.174 0.817 0.256 0.096 W 4 0.311 0.156 0.099 0.288 W 12 0.863 0.027 0.644 0.616 R 2 0.978 0.986 0.972 0.893 N=146 indicates significance at 95% confidence using a two tailed test. ^ indicates significance at 95% confidence using a one tailed test. indicates unstable coefficient.

PAGE 194

194 CHAPTER 6 SIMULATION ANALYSI S C oefficient estimates for each CSA ORG pair presented in Chapter 5 were used to conduct a simulation analysis on the demand for air line passengers traveling to Florida. The present chapter includes seven sections. The first section discusses the simulation analysis and identifies the explanatory variables used in the analysis Sections two to six present results of the simulation anal ysis for income, airline ticket prices, terror, hurricanes, and monthly seasonality, respectively. The seventh section discusses s imulations performed o n the fire, rainfall and temperature variables. R esults from the simulation analysis determine how sens itive the demand for air passengers is to changes in a specific variable while all others are set to their a ctual values Also, the analysis simulated demand for air passengers in the absence of the 9 11 terrorist attacks. Then, this simulated demand is c ompared to the one that accounts for the terrorist attack. The difference between the two shows the impact of the terrorist attack o n demand for air passengers to Florida. Finally, the simulations were conducted to identify seasonal patterns exhibited by t he demand for air travel to Florida These inferences are a useful tool to draw recommendations for policymakers in the travel industry across the state Introduction The simulation analysis was conducted to measure the magnitude and the speed of the respo nse of the dependent variable to specific changes on one explanatory variable, with all other variables set at their actual values All measures of the response are relative to the monthly average number of passengers per 100,000 over the period between 19 96 and 2006. The expected average number of passengers for each of the 24 CSA ORG pairs was calculated using c oefficient estimates presented in Chapter 5. It is the average number of

PAGE 195

195 passengers calculated with all continuous explanatory variables set to t he ir actual values during t he 11 year period the terror dummy variable s set in the presence of the 9 11 attack and the monthly seasonality dummy variables set in the presence of seasonal variation Total number of passengers traveling from each U.S. regi on can be calculated by multiplying the average number of passengers by the population factor for each region: Northeast 540 (e.g., 54,000,000 people) Midwest 528, South 859, and West 649. For example, the average number of passengers per 100,000 from the Florida (33) Northeast (1001) pair under average conditions is 1 955. T otal number of passengers for the entire 11 year period is 1 955 540 12 months 11 years or 139 million passengers. On average, approximately 12.6 million passengers traveled from the N ortheast region each year since 1996. This estimate is consistent with values presented in Figure 3 2 of Chapter 3. Note that the population factor is the average for the whole 11 year period Appendix E presents monthly population factors during the peri od between 1996 and 2006 for each U.S. region. Since the simulations for the terror variables are presented by year and the seasonality variables by month, these factors can be used to calculate total passengers traveling from a particular region for a giv en year or month. Eight sets of simulations for eight explanatory variables were conducted to calculate the overall average, average by month, and average by year. The set of simulations for each variable was set to change at different levels and were def ined as follows: Income: 5 % decline to 5% increase in per capita personal disposable income in increments of 1% Airline ticket p rices: 20 % decrease to 20 % increase in airfares in increments of 5% Terror: absence of 9 11 terrorist attack Hurricanes: a bsen ce of hurricanes affecting Florida

PAGE 196

196 Wildf ires: a bsence of wildfires affecting Florida Rainfall: 20 % decrease to 20 % increase in rainfall from a region, in increments of 5% Temperature: 20 % decrease to 20 % increase in temperatures from a region, in increment s of 5% Seasonality: average number of passengers by month The next five sections discuss results from each set of simulations p erformed to each CSA ORG pair. R esults of the simulations for each variable are presented in separate sections and each section discusses the effects of the changes by destination CSA. Note that the dependent variable represents number of passengers per 100,000 habitants in the origin U.S. region per month unless otherwise noted. Note that some of the explanatory variables were not statistically significant in some CSA ORG pairs according to the model estimation presented in Chapter 5. If the explanatory variable was not significant for a particular CSA ORG pair, simulation analysis has no relevance because the variable effect is no t statistically significant Inferences are only applicable to those CSA ORG pairs where the variable was statistically significant. The magnitude of the response is discussed in each section as well as the speed of adjustment. Note that even if a variable is statistically significant, inferences may only be valid significant different from one, there is no difference between short and long run adjustment implying that the dependent variable adjusts immediately to changes in structural variables Thus, inferences were made about the differences between the short run and long run adjustments only for those CSA as significantly different from one. A final set of simulations was performed to identify the presence of seasonal patterns in the demand for air transportation to Florida and the simulation results are presented following the

PAGE 197

197 section on hurricanes It is worthwhile to indicate that no simulations were performed using the advertising variable. Results presen ted in Chapter 5 indicate that coefficient estimates for this variable exhibit wrong sign and most of them were not significant. Simulations for Income The first set of simulations was conducted on the income variable which represents per capita personal disposable incom e for a particular U.S. region ( ORG ): Northeast, Midwest, South and West regions. The income variable was set to fluctuate between a 5% decrease to a 5% increase in increments of 1% R esults presented in Table 5 2 to 5 7 in Chapter 5 showed that the income variable was statistically significant in 22 of the 24 CSA ORG pairs. Therefore, simulation values are valid for all but two pairs. Al so note that income elasticities of demand for 22 CSA ORG pairs were greater than one suggesting that air transportation is not only a normal good but also a luxury good. That is, demand for air transporta tion increases more rapidly than income. The follow ing discussion centers on the magnitude (measured in elasticities) and response speed (comparing short run adjustment to full effect) across each of the six destination CSAs including Florida in relation to each of the four originating U.S. regions. Florid a CSA Income Simulation s Table 5 2 in Chapter 5 indicates that income was significant in each of the four U.S. regions traveling to the Florida CSA Therefore, the simulation results presented in this section have statistical validity. Nevertheless, the sp eed of adjustment coefficient from the Midwest region is not statist ically different from one. S hort run and long run simulated values from the Midwest region are statistically equal and therefore, the response to changes in income occurs instantaneously.

PAGE 198

198 Only inferences about the differences between short and long run adjustments from the other three U.S. regions are valid The South region yielded 2 106 passengers per month during the 11 year period covered under average conditions and the presen ce of th e 9 11 terrorist attack, followed by the Northeast (1 955 passengers ), Midwest (1 335 passengers ), and West (370 passengers) region s The West region has the largest long run income elasticity of demand a t 1.90, followed by the Midwest (1.67), Northeast ( 1 .51), and South ( 1.25) regions Figure 6 1 presents the relationship between different levels of per capita personal disposable income and domestic demand for air passengers from four U.S. regions traveling to the Florida CSA. Comp aring income elasticities of demand, demand from the West region is more sensitive to changes in income than any other U.S. region but its response is more rigid compared to the other U.S. regions. In other words, the magnitude of the response is greater but it takes more time to realize the full effect related to a change in income. For example, a 3% increase in income represents a 5.8 % increase in demand but only 65 % of this increase is realized in the short run. The opposite occurs for the demand from the South region. The inco me elasticity of demand from this U.S. region is the lowest, but its response in the short run is faster. A 3% increase in income represents a 3.8 % increase in demand and 80 % of this increase is realized in the short run. S hort and long run changes in dome stic demand for air pas s engers at different levels of per capita personal disposable income f rom four U.S. regions traveling to the Florida CSA are illustrated in Figure 6 2

PAGE 199

199 Figure 6 1 Florida CSA Income simulation s : r el ationship between different levels of personal disposable income and domestic demand for air passengers f rom four U.S. regions. As terisk (*) denotes short run adjustment is statistically different from the long run adjustment given a two tailed test at 95% confidence 1,650 1,700 1,750 1,800 1,850 1,900 1,950 2,000 2,050 2,100 2,150 28.9 29.2 29.5 29.8 30.1 30.4 30.7 31.1 31.4 31.7 32.0 Passengers per 100,000 Florida (33) Northeast (1001) Short run* Long run 1,100 1,150 1,200 1,250 1,300 1,350 1,400 1,450 1,500 25.4 25.6 25.9 26.2 26.4 26.7 27.0 27.2 27.5 27.8 28.0 Florida (33) Midwest (1002) Short run Long run 1,800 1,850 1,900 1,950 2,000 2,050 2,100 2,150 2,200 2,250 2,300 23.9 24.1 24.4 24.6 24.9 25.1 25.4 25.6 25.9 26.1 26.4 Passengers per 100,000 PC PDI (USD 000) Florida (33) South (1003) Short run* Long run 200 250 300 350 400 450 500 25.9 26.1 26.4 26.7 27.0 27.2 27.5 27.8 28.1 28.3 28.6 PC PDI (USD 000) Florida (33) West (1004) Short run* Long run

PAGE 200

200 Figure 6 2 Florida CSA Income simulation s : s hort and long run change s in domestic demand for air p assengers from four U.S. regions at diffe rent levels of per capita personal disposable income Asterisk (*) den otes short run adjustment is statistically different from the long run adjustm ent given a two tailed test at 95% confidence 200 150 100 50 0 50 100 150 200 28.9 29.2 29.5 29.8 30.1 30.4 30.7 31.1 31.4 31.7 32.0 Change in Passengers per 100,000 Florida (33) Northeast (1001) Short run* Long run 200 150 100 50 0 50 100 150 200 25.4 25.6 25.9 26.2 26.4 26.7 27.0 27.2 27.5 27.8 28.0 Florida (33) Midwest (1002) Short run Long run 200 150 100 50 0 50 100 150 200 23.9 24.1 24.4 24.6 24.9 25.1 25.4 25.6 25.9 26.1 26.4 Change in Passengers per 100,000 PC PDI (USD 000) Florida (33) South (1003) Short run* Long run 200 150 100 50 0 50 100 150 200 25.9 26.1 26.4 26.7 27.0 27.2 27.5 27.8 28.1 28.3 28.6 PC PDI (USD 000) Florida (33) West (1004) Short run* Long run

PAGE 201

201 South Florida CSA Income Simulation s D emand from the Northeast region amounted to 911 pa ssengers per month during the 11 year peri od covered under average conditions and the presence of the 9 11 terrorist attack s followed by the South ( 750 ), Midwest ( 392 ), and West ( 156 ) regions Comparing income elasticities of demand, the West region is more sensitive to changes in income than any other U.S. region in both the short and long run, while the Northeast region is the least sensitive. A percentage change in income leads to a 1.40 % change in demand from the West region compared to a 1.14 % change in demand from the Northeast region in the long run The long run income elasticity of demand from the Midwest and South regions was 1. 19 and 1. 20 respectively Figure 6 3 illustrates the relationship between different levels of per capita personal disposable income and number of passengers for e ach of the four U.S. regions traveling to the South Florida CSA. The speed of adjustment of all four U.S. regions was statistically different from one, implying that the full effect of a change in income is not completely realized in the current period. Pa rt of the demand adjusts immediately and the remainder is realized in subsequent periods. Within the differences between the short run and the long run adjustment to changes in income response from the Northeast region indicates that 77 % is realized in th e short run For example, a 3% increase in income yields a long run incr ease of 31 passengers per month (3.4% increase) of which 24 are accounted as the short run or immediate increase in demand and the remaining seven are related to the lagged effect from the increase in income. Faster responses were found in the demands from the other three U.S. regions The Midwest region shows a short run response of 85 % followed by the South ( 86% ) and West ( 81 % ) regions Figure 6 4 shows sh ort and long run changes in domestic demand for air passengers at different levels of per capita personal disposable income f rom four U.S. regions traveling to the South Florida CSA.

PAGE 202

202 Figure 6 3 South Florida CSA Income simulations : r elationship betw een different levels of personal disposable income and number of passengers for each of the four U.S. regions. Asterisk (*) denotes short run adjustment is statistically different from the long run adjustment given a two tailed test at 95% confidence 800 820 840 860 880 900 920 940 960 980 28.9 29.2 29.5 29.8 30.1 30.4 30.7 31.1 31.4 31.7 32.0 Passengers per 100,000 South Florida (1) Northeast (1001) Short run* Long run 340 350 360 370 380 390 400 410 420 25.4 25.6 25.9 26.2 26.4 26.7 27.0 27.2 27.5 27.8 28.0 South Florida (1) Midwest (1002) Short run* Long run 660 680 700 720 740 760 780 800 820 23.9 24.1 24.4 24.6 24.9 25.1 25.4 25.6 25.9 26.1 26.4 Passengers per 100,000 PC PDI (USD 000) South Florida (1) South (1003) Short run* Long run 130 135 140 145 150 155 160 165 170 25.9 26.1 26.4 26.7 27.0 27.2 27.5 27.8 28.1 28.3 28.6 PC PDI (USD 000) South Florida (1) West (1004) Short run* Long run

PAGE 203

203 Figure 6 4 South Florida CSA Income simulations : short and long run changes in the number of passengers at different levels of per capita personal disposable income for each of the four U.S. regions. Asterisk (*) denotes short run adjustment is statistically different from the long run adjustment given a two tailed test at 95% confidence 60 40 20 0 20 40 60 28.9 29.2 29.5 29.8 30.1 30.4 30.7 31.1 31.4 31.7 32.0 Change in Passengers per 100,000 South Florida (1) Northeast (1001) Short run* Long run 60 40 20 0 20 40 60 25.4 25.6 25.9 26.2 26.4 26.7 27.0 27.2 27.5 27.8 28.0 South Florida (1) Midwest (1002) Short run* Long run 60 40 20 0 20 40 60 23.9 24.1 24.4 24.6 24.9 25.1 25.4 25.6 25.9 26.1 26.4 Change in Passengers per 100,000 PC PDI (USD 000) South Florida (1) South (1003) Short run* Long run 60 40 20 0 20 40 60 25.9 26.1 26.4 26.7 27.0 27.2 27.5 27.8 28.1 28.3 28.6 PC PDI (USD 000) South Florida (1) West (1004) Short run* Long run

PAGE 204

204 Orlando CSA Income Simulation s D emand from the Northeast region totaled 592 passengers per month during the 11 year period covered under ave rage conditions and the presence of the 9 11 terrorist attack s followed by the South ( 516 ), Midwest ( 451 ), and West ( 134 ) regions In the long run, demand from the West region is more sensitive to changes in in come than any other U.S. region while the So uth region is the least sensitive. A percentage change in income leads to a 2.72 % change in demand from the West region compared to a 1.54 % change in demand from the South region in the long run. The long run income elasticity of demand from the Northeast and Midwest regions was 2.00 and 1.94, respectively. T he relationship between different levels of per capita personal disposable income and the number of passengers for each of the U.S. regions traveling to the Orlando CSA is shown in Figure 6 5. Within th e differences between the short run and the long run adjustment to changes in income only the elasticities from the Northeast and Midwest regions can be compared. The speed of adjustment coefficient from the South and West regions is not statistically dif ferent from one. Thus, the response to changes in income occurs instantaneously. Fifty nine percent of the full r esponse from the Northeast region is realized in the short run. For example, a 3% increase in income yields a long run increase of 36 passenge rs per month ( 6.1 % increase) of which 22 are accounted as the short run or immediate increase in demand and the remaining 14 are related to the lagged effect from the increase in income. A quicker response was observed in the demand from the Midwest regio n. Th is U.S. region yielded a short run response of 79 % Figure 6 6 illustrate s short and long run changes in domestic demand for air passengers at different levels of per capita personal disposable income f rom four U.S. regions traveling to the Orlando C S A.

PAGE 205

205 Figure 6 5 Orlando CSA Income simulations : r elationship between different levels of personal disposable income and number of passengers for each of the four U.S. regions Asterisk (*) denotes short run adjustment is s tatistically different from the long run adjustment given a two tailed test at 95% confidence 450 500 550 600 650 700 28.9 29.2 29.5 29.8 30.1 30.4 30.7 31.1 31.4 31.7 32.0 Passengers per 100,000 Orlando (5) Northeast (1001) Short run* Long run 350 370 390 410 430 450 470 490 510 25.4 25.6 25.9 26.2 26.4 26.7 27.0 27.2 27.5 27.8 28.0 Orlando (5) Midwest (1002) Short run* Long run 420 440 460 480 500 520 540 560 580 23.9 24.1 24.4 24.6 24.9 25.1 25.4 25.6 25.9 26.1 26.4 Passengers per 100,000 PC PDI (USD 000) Orlando (5) South (1003) Short run Long run 100 110 120 130 140 150