U.S. Import Demand for Beer, Wine, and Spirits by Country of Origin

MISSING IMAGE

Material Information

Title:
U.S. Import Demand for Beer, Wine, and Spirits by Country of Origin a Differential Approach
Physical Description:
1 online resource (65 p.)
Language:
english
Creator:
Gao, Bo
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Food and Resource Economics
Committee Chair:
SEALE,JAMES L,JR
Committee Co-Chair:
GAO,ZHIFENG

Subjects

Subjects / Keywords:
alcohol -- demand -- elasticity -- import
Food and Resource Economics -- Dissertations, Academic -- UF
Genre:
Food and Resource Economics thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
As few empirical studies that estimate import demand for the U.S. alcoholic beverages exist, this research focuses on providing the latest market trend and specific elasticities for the main commodities under the group of U.S. alcoholic beverages. A differential approach for a general demand model is used to estimate the U.S. import demand for beer, wine and spirits by country of origin from the major exporters to the U.S. alcoholic beverage market. This paper also calculates conditional expenditure elasticities, Slutsky (Cournot) own-price, and cross-price elasticities. The empirical analysis provides policy recommendations for both foreign and domestic alcoholic beverage industries.
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 Bo Gao.
Thesis:
Thesis (M.S.)--University of Florida, 2014.
Local:
Adviser: SEALE,JAMES L,JR.
Local:
Co-adviser: GAO,ZHIFENG.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2014
System ID:
UFE0046815:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 U.S. IMPORT DEMAND FOR BEER, WINE, AND SPIRITS BY COUNTRY OF ORIGIN: A DIFFERENTIAL APPROACH BY BO GAO A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2014

PAGE 2

2 2014 Bo Gao

PAGE 3

3 ACKNOWLEDGEMENTS I would like to express my thanks to the Food and Resource Economics Department for providing me the opportunity to study and research in the United States. Without its support, nothing would have come true. Most gratefully, I would like to express my sincere gratitude and appreciation to Dr. James Seale, Jr., cha irman of my supervisory committee. His continuous support, guidance and encouragement made my academic journey fantastic. His advice and comments were extremely valuable. His patience and positive attitude impressed me deeply and gave me the confidence to complete this thesis. Nothing could be possible without his advice and guidance to fulfill all the achievements. He is a mentor who I really respect and believe in. In addition, I would also like to express my great thanks to my other committee member, Dr. Gao, Zhifeng, for his guidance and support. He has offered great organization and practical instructions. His tremendous help brings this thesis to a successful conclusion. Besides, I also thank other faculty, staff, and fellow students in the Food and Re source E conomics Department. Their help and support truly made my life in UF colorful and meaningful. At last, I would like to give my deepest thanks to my parents and my friends. Thanks go out to my parents for their support, trust, and love. Thanks go o ut to my friends for their understanding and help.

PAGE 4

4 TABLE OF CONTENTS page ACKNOWLEDGEMENTS ................................ ................................ ............................... 3 LIST OF TABLES ................................ ................................ ................................ ............ 6 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ............................. 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 11 2 LITERATURE REVIEW ................................ ................................ .......................... 17 3 DATA ................................ ................................ ................................ ...................... 20 4 METHODOLOGY ................................ ................................ ................................ ... 24 Four Basic Demand Systems and the General Model ................................ ............ 24 Two Stage Budgeting ................................ ................................ ............................. 30 5 U.S. IMPORT DEMAND FOR BEER, WINE, AND SPIRITS ................................ .. 33 Testing Restrictions and Model Choice ................................ ................................ ... 33 Parameter Estimates ................................ ................................ .............................. 34 Conditional Expenditure Elasticities ................................ ................................ ........ 35 Two Types of Price Elasticities ................................ ................................ ............... 36 6 U.S. IMPORT DEMAND FOR BEER, WINE AND SPIRITS BY COUNTRY OF ORIGIN ................................ ................................ ................................ ................... 41 U.S. Import Demand for Beer by Country of Origin ................................ ................. 41 Testing Restrictions and Model Choice ................................ ............................ 41 Parameter Estimates ................................ ................................ ........................ 42 Conditio nal Expenditure Elasticities ................................ ................................ 42 Two Types of Price Elasticities ................................ ................................ ......... 43 U.S. Import Demand for Wine by Country of Origin ................................ ................ 44 Testing Restrictions and Model Choice ................................ ............................ 45 Parameter Estimates ................................ ................................ ........................ 45 Conditional Expenditure Elasticities ................................ ................................ 46 Two Types of Price Elasticities ................................ ................................ ......... 46 U.S. Import Demand for Spirits by Country of Origin ................................ .............. 48 Testing Restrictions and Model Choice ................................ ............................ 49

PAGE 5

5 Parameter Estimates ................................ ................................ ........................ 49 Conditional Expenditure Elasticities ................................ ................................ 50 Two Types of Price Elasticities ................................ ................................ ......... 50 7 CONCLUSIONS ................................ ................................ ................................ ..... 59 LIST OF REFERENCES ................................ ................................ ............................... 62 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 65

PAGE 6

6 LIST OF TABLES Table page 3 1 U.S. budget import shares prices, quantities and values for aggregated beverages, 1992 2012. ................................ ................................ ....................... 22 3 2 U.S. average import values, quantities, quantities per capita, and value shares for beer, wine and spirits by country of origin 2008 2012. ....................... 23 4 1 Income and price elasticity expression s for the Rotterdam, CBS, AIDS, and NBR models. ................................ ................................ ................................ ...... 32 5 1 Test results for log likelihood ratio tests for different restrictions in the Rotterdam, CBS, AIDS, NBR, and general model for aggregated beverages. ... 39 5 2 Conditional parameter estimates for U.S. import demand for aggregated beverages. ................................ ................................ ................................ .......... 39 5 3 Marginal shares, conditional expenditure, Slutsky and Cournot elasticities of U.S. import demand for aggregated beverages at sample mean. ...................... 40 6 1 Test results for log likelihood ratio tests for different restrictions in the Rotterdam, CBS, AIDS, NBR, and general model for beer. ................................ 53 6 2 Conditional parameter estimates for U.S. import demand for beer by country of origin. ................................ ................................ ................................ .............. 53 6 3 Marginal shares, conditional expenditure and Slutsky (compensated price elasticities) of U.S. import demand for beer by country of origin at sample mean. ................................ ................................ ................................ ................. 54 6 4 Cournot (uncompensated price elasticities) of U.S. import demand for beer by country of origin at sample mean. ................................ ................................ .. 54 6 5 Test results for log likelihood ratio tests for different restrictions in the Rotterdam, CBS, AIDS, NBR, and general model for wine. ............................... 54 6 6 Conditional parameter estimates for U.S. import demand for wine by c ountry of origin. ................................ ................................ ................................ .............. 55 6 7 Marginal shares, conditional expenditure and Slutsky (compensated price elasticities) of U.S. import dem and for wine by country of origin at sample mean. ................................ ................................ ................................ ................. 55 6 8 Cournot (uncompensated price elasticities) of U.S. import demand for win e by country of origin at sample mean. ................................ ................................ .. 56

PAGE 7

7 6 9 Test results for log likelihood ratio tests for different restrictions in the Rotterdam, CBS, AIDS, NBR, and general model for spirits. ............................. 56 6 10 Conditional parameter estimates for U.S. import demand for spir its by country of origin. ................................ ................................ ................................ 57 6 11 Marginal shares, conditional expenditure and Slutsky (compensated price elasticities) of U.S. i mport demand for spirits by country of origin at sample mean. ................................ ................................ ................................ ................. 57 6 12 Cournot (uncompensated price elasticities) of U.S. import demand for spirits by country of origin at sample mean. ................................ ................................ .. 58

PAGE 8

8 LIST OF FIGURES Figure page 1 1 U.S. import trends for beer, wine, and spirits by value, 1992 2012. ................... 14 1 2 U.S. wine import value share by country of origin, 2012. ................................ .... 15 1 3 U.S. beer import value share by country of origin, 2012. ................................ .... 15 1 4 U.S. spirits import value share by country of origin, 2012. ................................ .. 16

PAGE 9

9 LIST OF ABBREVIATIONS AIDS Almost Ideal Demand System CBS The Central Bureau HS Harmonized System LR Log likelihood Ratio NBR The National Bureau of Research ROB Rest of Beverages ROW Rest of World

PAGE 10

10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science U.S. IMPORT DEMAND FOR BEER, WINE, AND SPIRITS BY COUNTRY OF ORIGIN: A DIFFERENTIAL APPROACH By Bo Gao May 2014 Chair: James L. Seale, Jr. Major: Food and Resource Economics As few empirical studies that estimate import demand for the U.S. alcoholic beverages exist, this research focuses on providing the latest market trend and specific elasticities for the main commodities under the group of U.S. alcoholic beverages. A differ ential approach for a general demand model is used to estimate the U.S. import demand for beer, wine and spirits by country of origin from the major exporters to the U.S. alcoholic beverage market. This paper also calculates conditional expenditure elastic ities, Slutsky (Cournot) own price, and cross price elasticities. The empirical analysis provides policy recommendations for both foreign and domestic alcoholic beverage industries.

PAGE 11

11 CHAPTER 1 INTRODUCTION (WHO 2011). 2011, the total import value of alcoholic beverages reach 15.39 billion dollars and make up to 20.8% of the world imports' share (UN Comtrade, 2012). The U.S. has a huge net demand for alc oholic beverages. Net imports are about 12.0 billion dollars in 2011. The large population and an increasing preference for imported alcoholic beverages cont ribute to the deficit on the commodities of alcohol. Beer, wine, and spirits are the major groups of alcoholic beverages. Distilled spirits ($5.2 billion), wine ($4.2 billion), and beer ($3.5 billion) lead the alcoholic beverages import in 2010. Figure 1 1 shows the trends of imported beer, wine, and spirits into the U.S. over the last two decades. U.S. beer imports increase over 329% by value; for wine imports, the growth is 366%; for spirits, imports are up 256%. The total value of the alcohol import mark et is over 15 billion dollars. Beer, wine, and spirits are 14.84% of total U.S. agricultural imports in 2012(USDA FATUS, 2012). From the perspective of the consumers, alcohol beverages plays an important role in daily life. According to the survey data fro m the National Center for Health Statistics (NCHS, 2012), 51.5% of adults (18 years of age and over) are current regular drinkers (at least 12 drinks in the past year). From a poll of Gallup (2013), 36% of U.S. regular drinkers consume beer most often, 35% for wine, and 23% for spirits. Rapid growth in U.S. wine imports has occurred in the last two decades. Total U.S. wine imports increase 366% in terms of value and 350% in terms of quantity. In 2012, the U.S. imports 1.17 billion liters of wines at a value of five billion dollars. France,

PAGE 12

12 Italy, Australia, and Chile are the major wine exporters to the U.S. These countries accounted for more than 74% of total U.S. wine imports by value (U.S. Department of Agriculture, 2013). The U.S. beer imports have also risen impressively from 1992 to 2012. Total U.S. beer imports rise 329% by value and 233% by volume. In 2012, the U.S. imports 3.25 billion liters of beer at a value of 3.7 billion dollars Mexico and Netherlands are the major beer exporters to the U.S. Th ese two countries account for more than 73% of total U.S. beer imports by value ( U.S. Department of Agriculture 2013). Tremendous development in U.S. spirits imports has occurred from 1992 to 2012. Total U.S. spirits imports grow 256% in terms of expendit ure and 76% in terms of quantity. In 2012, the U.S. imports 685 million liters of spirits at a value of 6.5 billion dollars. Mexico, United Kingdom, France and Sweden are the major exporters to the U.S. These countries account for more than 72% of total U. S. imports of spirits by value ( U.S. Department of Agriculture 2013). Despite its huge value and importance in daily consumption, few studies or research are found in the literature that focuses on U.S. alcohol trade, and, more specifically, U.S. import d emand for beer and spirits. As a result, little is known about demand relationships among these imported products. However, these relationships are extremely important for those exporters and our domestic alcohol industry. In this thesis, four functional approaches are utilized (i.e., Rotterdam, CBS, AIDS, and NBR) to estimate U.S. import demand elasticities for alcoholic beverages and beer, wine, and spirits by country of origin. In addition, the author estimates a general model that nests these four mod els to choose the best of the four competing models in terms

PAGE 13

13 of data fitting of aggregated beverages and beer, wine, and spirits by country of origin. The usual restrictions implied by demand theory are imposed and tested with log likelihood ratio (LR) tes ts. Conditional expenditure and price parameters are attained for imported alcoholic beverages. Conditional expenditure elasiticities, own price and cross price elasticities (both Slutsky and Cournot) are reported and discussed in the empirical results. T his thesis is organized as follow: Chapter 2 provides insight into the literature dealing with several demand models theoretically; prior studies of import demand analysis for alcoholic beve rages; and previous work analyzing U.S. import demand for alcoholi c beverages. Chapter 3 provides the data sources and descriptive statistics for U.S. import aggregated beverages and U.S. imported beer, wine, and spirits by country of origin. Chapter 4 presents the detailed description of four functional approaches (i.e. Rotterdam, CBS, AIDS, and NBR) and the general demand systems. The multistage budgeting method is discussed as well. Chapter 5 shows U.S. import demand for beer, wine, and spirits from the perspective of testing restrictions, model choice, parameter esti mates, conditional expenditure elasticities and two types of price elasticities (Slutsky and Cournot). Chapter 6 shows U.S. import demand for beer, wine and spirits by country of origin. This chapter also presents the empirical results and analysis similar to the chapter 5. Chapter 7 provides the general conclusions and recommendations through the empirical results.

PAGE 14

14 Figure 1 1. U.S. import trends for beer, wine, and spirits by value, 1992 2012. Source: United States of Department of Agriculture, 2013. 0 1000000 2000000 3000000 4000000 5000000 6000000 7000000 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 Dollars (Thousands) Beer Wine Spirits

PAGE 15

15 Figure 1 2 U.S. wine import value share by country of origin, 2012. Source: United States of Department of Agriculture, 2013. Figure 1 3 U.S. beer import value s hare by country of origin, 2012. Source: United States of Department of Agriculture, 2013. France 29% Itlay 27% Australia 11% Chile 7% Rest of World 26% Mexico 49% Netherlands 24% Rest of World 27%

PAGE 16

16 Figure 1 4 U.S. spirits import value share b y country of origin, 2012. Source: United States of Department of Agriculture, 2013. France 29% United Kingdom 25% Mexico 12% Sweden 6% Rest of World 28%

PAGE 17

17 CHAPTER 2 LITERATURE REVIEW Few studies estimate the import demand for alcoholic beverages. Andrikopoulos, Brox, and Carvalho (1997) estimate the demand for domestic and imported alcoholic beverages in Ontario, Canada with the Almost Ideal Demand System (A IDS) (Deaton and Muellbauer 1980). They use a dataset from 1958 to 1987 for domestic and imported produced spirits, wine and beer. The results show a strong tendency that the dynamic version of the AIDS better fits the data than its static version, but the re is little evidence of strong substitution effects between imported and domestic goods. Likewise, Carew, Florkowski, and He (2005) analyze demand for the imported and domestic table wine market in British Columbia, Canada with the AIDS model. They use re tail monthly scanner data of table wine that includes domestic and imported table wine in the province of British Columbia from March 1990 to April 2000. The findings uncover that roduced wines. It is shown that the expenditure elasticities for British Columbia produced, European, and rest of world (ROW) white wines are greater than those for red wines. The high expenditure elasticities with British Columbia white wines may represen t that these wines are of better quality. Lee, Kennedy, and Hilbun (2008) analyze an import demand system of the South Korean wine market from the source differentiated AIDS model. With the assumption of partial aggregation and block substitutability, empi rical findings indicate that South Korean consumers have a huge preference for French wines with high quality. Also, as a result of the free trade agreement between Chile and South Korea, Chilean wines have increased their market share in South Korea.

PAGE 18

18 Stu dies for U.S. import demand elasticities by country of origin for alcoholic beverages are limited, especially for the beer and spirits. Seale, Marchant and Basso (2003) analyze the demand for imports versus domestic production for the U.S. red wine market. Results for conditional expenditure elasticities show that the U.S. red wine industry increases earning relative to imports when U.S. consumers' total budgets on red wine increase. Own price and cross price elasticities indicate that an increase in the price of U.S. red wine leads to a decline in quantity demanded for U.S. red wine about six times larger than for Italian and F rench red wines and over 20 times larger than other import partners, thus it is bad for U.S. domestic industry. The findings suggest U.S. red wine producers may add to their revenue by decreasing prices; however, French and Italian industries may improve t otal revenues by increasing prices. There are no studies found that focus on the import demand for beer and spirits in the U.S. previously estimated (Wang et al. 199 6). Although there are few studies on import demand of alcoholic beverages in the U.S., there are many demand studies on other commodities, both domestic and import demand. In particular, consumer allocation models have been used in research that focus on the demand of commodities both in terms of domestic and import demand. Barten (1993) made a systematic comparison of four versions of differential demand systems. The usual demand systems include the Rotterdam system (Theil 1965), the Almost Ideal Demand s ystem (AIDS) (Deaton and Muellbauer 1980), the Central Bureau of Statistics (CBS) system (Keller and van Driel 1985) and the National Bureau of Research (NBR) system (Neves 1987). Rotterdam and AIDS are most popular in the

PAGE 19

19 agricultural economics research ( e.g., Eales and Unnevehr 1988; Lee, Seale, and Jierwiriyapant 1990; Sparks, Seale, and Buxton 1992). In general, different systems have various implications and effects of the specific data. For example, marginal expenditure shares and Slutsky terms are as sumed constants in the Rotterdam model, while they are assumed functions of budget shares in the AIDS. Besides, CBS and NBR models can be considered as income response variants of Rotterdam and the AIDS, respectively (Lee, Brown, and Seale 1994). As the d emand systems are more frequently applied in research, the issue of how to choose the model that best fits and represents the data become more important. A synthetic model of Barten (1993) combines the features of the four differential models (i.e. Rotterd am, AIDS, CBS and NBR) and allows nested hypothesis testing among the models to choose which of the four models best fit the data. Specifically, the generalized demand model, a synthetic differential demand model, nests these four demand systems with the a ddition of two more parameters. A likelihood ratio test is performed for each of the four models in comparison to the general model. Lee, Brown, and Seale (1994) develop a more general demand formulation of Barten (1993) general model and apply it to Taiwa nese consumer behavior and demands for 12 commodity groups from 1970 to 1989.

PAGE 20

20 CHAPTER 3 DATA The alcoholic beverage dataset is obtained from the Global Agricultural Trade System, supported by the Foreign Agricultural Service, U.S. Department of Agric ulture (2013). Specifically, the data sources of beer, wine and spirits are from the U.S. Census Bureau Trade data (2013). The exporters of these commodities to the U.S. chosen for study are the top exporters to the U.S. The import expenditure data contain the volume and price of beer, wine, and spirits. Unit value prices are computed by dividing the import values by the import quantities. The time period ranges from 1992 to 2012 for beer, wine, and spirits. The analyses use HS 4 classification standard to sort the groups population annual data (1992 2012) are adopted from the U.S. Census Bureau (2013) in order to conduct the following analysis on a per capita basis. Table 3 1 shows the basic descriptive statistics (budget share, unit price, quantity, and value) of the four beverage groups for U.S. imported aggregated beverages in 1992, at the sample mean, and 2012. The largest average expenditure share is spirits (0.351) and the smallest is ROB 1 (0.146). As shown in Table 3 1, the group budget shares are r elatively steady over the last two decades for beer and wine. The expenditure share of ROB increases the most from 0.100 to 0.230 and that of spirits drops most from 0.436 to 0.328. However, spirits still keeps the largest import 1 alcohol water and fermented beverages, etc.

PAGE 21

21 share of the four beverage groups in 2012. In terms of unit price level, all prices increase from 1992 to 2012, but at different rates. The prices of beer and wine just experience a slight increase, while the prices for spirits and ROB almost double. In 2012, the highest import pri ce of the four beverage groups is spirits, which is $9.49 per liter. Quantity and total value are also reported in Table 3 1. Table 3 2 shows the U.S. import statistics for beer, wine, and spirits by country of origin (Total value, quantity, and value shar e) from 2008 to 2012. The countries under study for beer imports are Mexico and Netherlands, the top two exporters of beer to the U.S. market at the sample mean from 2008 to 2012. For the wine estimation, Italy, France, Australia and Chile are chosen for a nalysis. They are the top four largest exporters of wine to the U.S. market at the sample mean from 2008 to 2012. For the spirits estimation, the paper estimate import demand from France, United Kingdom, Mexico and Sweden into the U.S. market. They are als o the top four biggest exporters of spirits into the U.S. market from 2008 to 2012. Due to the lack of import data for values and volumes on the other commodities, it is assumed that alcoholic beverage consumption is weakly separable from other commod ities (Barten 1977). Thus, the parameter estimates shown in the following chapters are conditional demand parameters.

PAGE 22

22 Table 3 1 U.S. budget import shares price s, quantities and values for aggregated beverages, 1992 2012. Year Beer Wine Spirits ROB a Budget share 1992 0.206 0.259 0.436 0.100 Average 0.235 0.268 0.351 0.146 2012 0.187 0.255 0.328 0.230 Price (nominal dollars) 1992 0.883 4.206 4.708 0.416 Average 0.982 4.753 6.957 0.550 2012 1.140 4.333 9.490 0.777 Quantity (1000 liters) 1992 976,151 258,016 388,114 1,011,539 Average 2,494,910 615,934 540,973 2,957,780 2012 3,251,923 1,167,454 685,392 5,853,150 Value (1000 dollars) 1992 861,859 1,085,217 1,827,372 420,530 Average 2,514,990 2,971,040 3,760,080 1,763,400 2012 3,705,746 5,058,590 6,504,081 4,547,860 Source: Global Agricultural Trade System, U.S. Department of Agriculture, 2013. a ROB represents rest of beverages which includes unsweetened water, non alcohol water and fermented beverages, etc.

PAGE 23

23 Table 3 2 U.S. average import value s quantit ies, quantities per capita, and value share s for beer, wine and spirits by country of origin 2008 2012. Country Value (1000 dollars) Quantity (1000 liters) Quantity, per capita (liter) Value share (percentage) Beer Mexico 1,647,396 1,639,889 5.30 46.16 % Netherlands 945,589 662,725 2.14 26.50 % ROW 975,613 907,377 2.94 27.34 % Wine Italy 1,340,814 257,009 0.83 29.51 % France 1,217,391 102,618 0.33 26.79 % Australia 608,855 207,657 0.67 13.40 % Chile 285,036 122,410 0.40 6.27 % ROW 1 091 936 286,472 0.93 24.03 % Spirits France 1,616,796 86 823 0.28 28.01 % United Kingdom 1,388,028 155,496 0.50 24.05 % Mexico 693,602 108,117 0.35 12.02 % Sweden 414,200 142,614 0.46 7.18 % ROW 1,659,025 341,397 1.10 28.74 % Source: Global Agricultural Trade System, U.S. Department of Agriculture, 2013.

PAGE 24

24 CHAPTER 4 METHODOLOGY Four B asic D emand S ystems and the G eneral M odel The Rotterdam model and the AIDS model are popular in current studies. However, there is n o special rule or theory for model selection. The scholars often choose these two models or their variants according to their personal needs or inference (Lee, Brown, and Seale 1994). In this paper, we apply the general demand system (Barten 1993) to test which single demand system b est fit s the U.S. import data of alcoholic beverages. The Rotterdam model, according to Barten (1964) and Theil (1965), is specified as (4 1) where represents the aver age budget share for commodity ; and are the price and quantity of good respectively; and represent and respectively; and is the Divisi a volume index for the change in real inco me and may be written as The demand parameters and are given by , (4 2) where is total budget or the expenditure, and is the (i, j) th element of the Slutsky substitution matrix. The parameter is the marginal budget or e xpenditure share for commo dity and is a compensated price term. The constraints of demand theory can be directly applied to the Rotterdam parameters. We have

PAGE 25

25 Adding up Homogeneity Symmetry (4 3) In the differential demand equation of the Rot terda m model, demand parameters re is no rule that indicates s and s shoul d be held constant. The following equation is another parameterization based on ( 4 4) Because the sum of budget share s , is unity, the above equation implies that and multiplies by b and differentiates with respect to b. (4 5) W ith the help of this equation, the ith marginal share differs from the corresponding budget share by Thus, the budget or expenditure share is not constant with income, and neither is the associated marginal share. The income elast icity based on Equation 4 5 is (4 6) T he above formula illuminates that a commodity with negative is a necessity. Conversely, a good with positive is a luxury. Th e good is unitary elastic when is equal to zero and prices hold constant. By replacing in Equation 4 1 with Equati on 4 5 and rearranging terms, we derive (4 7)

PAGE 26

26 T his model is referred to be the CBS model, following Keller and van Driel (1985). In this equation, and are constant coef ficients (Keller and van Driel 1985; Theil and Clements 1987). The AIDS model, which may be regarded as an expans ion or development of Equation 4 4 by including the price effects, can be written as (4 8) where is a price index, can be defined by The adding up restrictions for the above equation are Homogeneity can be satisfied if and symmetry can be satisfied if We can also rearrange the terms and transform the basic equation of the AIDS model to differential form, based o n replacing the Divisia price index for in Equation 4 8 (Deaton and Muellbauer 1980). As suggested by Bar ten (1993), we use the following relations

PAGE 27

27 (4 9) where j, the value is zero. Equation 4 8 can be rewritten as (Lee, Brown and Seale 1994), (4 10) Besides the Rotterdam, CBS and AIDS model, we can derive the NBR (Neves 1987) by replacing with in the AIDS model. Thus, the NBR model has the Rotterdam income coefficients but the AIDS price coefficients. More specifically, the NBR model can be shown as: (4 11) We can also rewrite the above equat ion as (Lee, Brown and Seale 1994), (4 12) Thus, we obtain the basic expressions of four demand models. In addition, the income and price elasticities are deriv ed and shown in Table 4 1. In the following chapters, these expressions are utilized to calculate and report the income and price elasticities for beer, wine, and spirits. Through the comparison, they have the same left hand side variable and right hand side variables and with different parameterizations. Four models can be considered as four unique ways to parameterize a general dif regarded to be constants in the Rotterdam and CBS but to be variables in the AIDS and NBR; the marginal budget shares are shown to be constant in the Rotterdam and NBR but to be variables in the A IDS and CBS. The four models analyzed above are single equations for each system, which are not nested. However, a general or synthetic model can be developed which nests all

PAGE 28

28 four basic models ( Barten 1994) The general demand model of an importing countr y may be written in conditional form as (4 13) where is the conditional budg et share of imported commodity i from country c in period t, is the budget share of good i from exporting country c in time period t, is the group budget share for group K in time period t, is the price of imported good j from exporting country c in time period t, is the quantity of good i imported from exporting country c in time period t, and equals to and respectively represents Divisia volume index for the change in real income, is t he cond itional marginal share of good i from country c where is the uncond itional marginal share of good i from country c, and is the marginal share of group K, where b is the budget or total outlay. In the general demand model, and ; and are two additional parameters to be estimated. The demand model will become the conditional Rotterdam model when both and are restric ted to zero, the conditional CBS when = 1 and = 0, the conditional AIDS when =1 and = 1, and the conditional NBR when =0 and = 1. The demand restrictions on the general demand model are

PAGE 29

29 Adding up Ho mogeneity Symme try (4 14) Note that, for an application to discrete data, the specifications are approximated by replacing by by and by Additionally, and are treated as constant parameters. The likelihood ratio test (LRT) for model selection can be sh own as: (4 15) where is the vector of parameter estimates of each of the four restricted demand models; is the vector of parameter estimates of the general model; and log L(.) is the log value of the likelihood function (Amemiya 1985). For example, under the null hypothesis that the restricted model best describes the data, test statistic LRT has an 2 (q) distribution, in which q is the number of restrictions imposed. The goodness of fit has been tested by the method invoked by Bewley (1986). As the single equation R 2 statistics are not suitable for measuring the goodness of fit for demand systems, we apply the method and expressions proposed by Bewley, Young, and Coleman (1987), (4 16) where LR is twice the difference between the log likelihood of the naive model of four demand systems and the log likelihood of the same dependent variables regressed on the term only; T is the number of observations and n is the number of equations in the system.

PAGE 30

30 Two Stage Budgeting In this paper, we utilize the multistage budgeti ng approach, as many studies do (Barten 1977). This method is easily accommodated by the differential approach to utility maximization and is u seful when one wants to estimate the demand for disaggregated (imported) goods (Seale, Sparks, and Buxton 1992). First, Countries allocate total income between domestic and imported goods. Then, total import expenditure is allocated among all imported good s including alcoholic beverages. Preferences among these goods are assum ed to be blockwise dependent (Theil 1978) or w eakly separ able (Barten 1977). Based on this structure i mported alcoholic beverages are separable from domestically produced beverages. F inally, we allocate the expenditure on beer, wine and spirits among the different supplying countries. We need to specify the multi stages for estimation of group demand. In this paper, two basic stages are necessary to estimate the import demand of U.S. b eer, wine and spirits. To explain the group demand analysis, we use the Rotterdam model in the following parts so we can display our details for the group demand classification easily. The basic equation for the Rotterdam model can be shown as (4 17) when we utilize multistage budgeting, we need to estimate the demand for aggregate gro ups of goods. In the above formula, represents the average budget share for commodity i; and are the price and quantity of good i, respectively; and represent and respectively; and is the Divisia volume index for the change in real income and may be written as In the first stage, the demand for aggregate groups of goods is

PAGE 31

31 (4 18) where is the Divisia volume index for alcohol g represents the group containing beer, wine, spirits and ROB (refers to the rest of beverages), and is the avera ge budget share for commodity group g. With this function, we can e stimate three equations for beer, wine, and spirits in a four equation system. We drop the ROB to avoid the singularity issue (Barten 1969) In the second stage, we estimate the import demand for beer, wine and spirits by country of origin, respectively. The functions for the individual commodity import demand can be presented as (4 19) w h ere is the average budge share for source country i for commodity in group g, i represents the ith source import country of each of the g (i.e., beer, wine, and spirits) commodities and is the Divisia volu me index for the change in group g real expenditure With this function, we can conduct several equations analysis for beer, wine and spirits by country of origin, respectively. In conclusion, the four basic demand model s and the general demand system are utilized in the following chapters for empirical analysis. LR tests and g oodness of fit test s are applied to the following estimations. With the help of demand elasticity expressions in T able 4 1, marginal shares, conditional expenditure elasticities, own price and cross price elasticities are obtained for U.S. import demand of beer, wine, and spirits by country of origin.

PAGE 32

32 Table 4 1 Income and price elasticity expressi ons for the Rotterdam, CBS, AIDS, and NBR model s Model Income elasticity Price parameter Slutsky price elasticity Co u rnot price elasticity Rotterdam CBS AIDS NBR

PAGE 33

33 CHAPTER 5 U.S. IMPORT DEMAND FOR BEER, WINE, AND SPIRITS U.S. import demand for beer, wine and spirits is estimated based on four differential demand system (i.e., Rotterdam, CBS, AIDS, and NBR) and the general demand system. F our equation demand system s are estimated from 1992 to 2012 for aggregated beverages ( beer, wine, spirits, and ROB) using the five mentioned demand systems In this part, the AIDS model fits the data best and demonstrated from testing restrictions and LR test s Conditional parameter estimates, conditional expenditure elasticities and two t ypes of price elasticities are presented in this chapter. Testing Restriction s and Model Choice Firstly, in order to test whether the dataset satisfy homogeneity and symmetry restriction s the restrictions of homogeneity and symmetry are tested using LR tests. Secondly, LR test s for model selection are applied to choose the best model among the four demand models. The log likelihood values and ratio statistics of the four demand models and the general demand system are reported in Table 5 1 for aggregate d beverages. The results in the first three rows are the log likelihood values; the results in the last th ree rows are the log likelihood ratio test statistics. The test is where is the log likelihoo d value of the restricted model, and is that of the 2 (q) distribution. The degree of freedom is the difference between the number of parameters in the unrestricted and restricted model. On the basis of the results for aggregated beverag es, the null hypothesis of homogeneity is not rejected for any model at the 0.05 significance level. Plus, the nul l hypothesis of symmetry is not rejected for any model at the 0.05 significance level. The

PAGE 34

34 model selection results show that only the AIDS mod el is not rejected at 0.01 significance level. Thus, the results for the AIDS model are displayed and explained further in next section. F or the aggregated beverages using the AIDS model with homogeneity and symmetry, the goodness of fit measure is = 0.67 which illuminates that the whole system explains 67% of the variation in allocation Parameter Estimates The AIDS parameters are reported in Table 5 2 while marginal shares, conditional expenditure, own price elasticities and cross price elasti citi es are shown in Table 5 3. The own price parameters are reported along the diagonal o f columns (2) (5) of Table 5 2. The own price parameters of the four group commodities from beverages are negative and all are less than one absolutely as we expect, rangi ng from 0.089 (beer) to 0.266 (spirits). Of the four parameters, they are all statistically significant at The cross price parameters are reported as the non diagonal numbers o f columns (2) (5) of Table 5 2. These parameters show whether th e commodity from other commodities are complements or substitutes, refer to the sign of the parameters. If the parameter is positive, it is a substitute; if negative, it is a complement. All are positive except that of the beer ROB pairing. Among the condi tional cross price parameters, two are statistically the same as zero. The results show that beer and ROB are complements, and all other pairing are substitutes. Marginal shares are equivalent to the shares of any additional dollar spent on the group on a whole that will be allocated on each commodity. In the AIDS, the marginal shares equal for aggregated beverages. All marginal shares parameters are

PAGE 35

35 calculated at sample mean and are reported in column (2) of Table 5 3. No te that the marginal shares here are functions of budget shares, and they are all significantly additional dollar being allocated to aggregated beverages; 20 cen ts will be allocated to beer while the rest go ing to ROB. Among these commodities, the beer industry gain s the least if an additional dollar were spent on aggregated beverages. Conditional Expenditure Elasticities The conditional expenditure elasticities of the fo ur alcoholic commodities are repor ted in column (3) of Table 5 3. These elasticities disclose the percentage change in the quantities demanded from each of commodities when U.S. import expenditure increase s by 1% on these beverages. The elasticities can be judged by comparing the value with one. If a n elasticity is higher than one, it is conditionally elastic, which indicates the budget share for this commodity will raise if U.S. expenditure on beverages inc reases; if it is lower than one, it is conditionally inelastic which means the budget share for this commodities will decrease if U.S. expenditure on beverage increases. All of these expenditure elasticities are positive and significantly different from ze among these four beverages. Wine (0.93) has the next highest value of expenditure elasticities among these commodities, but it is inelastic. Beer and spirits are less elastic than wine, 0.83 and 0.70, respectively. These results indicate that if the U.S. increase s expenditure on beverages by 1%, the budget share for spirits, wine and beer would decrease slightly, while the quantities demanded of these beverages would increase by 0.9 3%, 0.83%, and 0.70% for wine, beer, and spirits, respectively. The budget share for ROB would raise and its quantity demanded would increase by 2.13%.

PAGE 36

36 Two Types of Price Elasticities Slutsky and Cournot elasticitie s are reported in columns (4) (7 ) of Tabl e 5 3. Slutsky own price elasticities are compensated and account for the substitution effect. The Cournot own price elasticities are uncompensated, and they have both s ubstitution and income effects. Accordingly, Cournot own price elasticities are larger absolutely than corresponding Slutsky ones. The results imply all conditional own price elasticities are negative. For the own price elasticity results, no matter Slutsky and Cournot, they For spi rits, its Slutsky own price is inelastic but its Cournot own price elasticity is unitary The Slutsky (Cournot) own price elasticity is 0.76 ( 1.00 ). For wine and beer, both their Slutsky and Cournot own price elasticities are inelastic. The Slutsky (Cour not) own price ela sticity for wine is 0.67 ( 0.92) and for beer is 0.38 ( 0.57 ). Compared with the difference between the conditional Slutsky and Cournot own price elasti cities of each commodity, wine has the bigg est change. It indicates wine has the str ongest expenditure sensitivity among these commodities in the U.S. import market. These results also point out different effects of price fluctuations on U.S. beverage demand. If the price of imported beer goes up by 1%, the uncompensated quantity demanded of imported beers will go down only 0.38%. However, imported spirits are much more influenced by this kind of own price fluctuations. More specifically, the uncompensated quantity demanded of imported spirits into the U.S. market would decrease about 0.76 %. Conditional Slutsky cross price elasticities measure the responsiveness of the demand for a good to a change in the price of a substitute or complement good. Positive (negative) Slutsky cross

PAGE 37

37 pric e from the ith source country will lead to the quantity demanded of this good to increase (decrease) fr om jth source country. Thus, their goods would be substitutes (complements). In the 12 conditional Slutsky cross price elasticities, 10 are positive with two approximately equal to zero The ROB spirits (0.80) pairing possesses the largest positive cross price elasticity. It indicates that a 1% increase in the price of other beverages will increase the quantity demanded (compensated) for spirits by 0.80%. Among the 12 conditional Cournot cross price elasticities, seven are negative and others are positive. Spirits ROB (0.2 3) ha s the largest positive cross price elasticity. ROB beer has the biggest negative cro ss price elasticity. For example a 1% decrease in the price of ROB will increase by 0.51% the uncompensated quantity demanded for imported beer. For aggregated beverages, U.S import demand for beer, wine, and spirits are inelastic, as the absolute value of their Slutsky (Cournot) own price elasticiti es are less than 1. Spirits and wine are substitutes for all the imported beverages. A compensated price change in spirits and wine has a small positive influence on the quantity demanded for all the other imported beverages. Another interesting fact is th at, when we take the income effect plus the substitution effect into account, an increase in the beer price will decrease the demand for all the other beverages. For example, if there is a 1% increase in the price of beers, the demand for ROB will decrease by 0.51 %. The decrease also happens for the demand for wine and spirits. Additionally, the largest change between the value of the Slutsky and Cournot cross price elasticity is for ROB spirits. The Slutsky value is 0.80 and the Cournot is 0.05. This indic ates if the price of ROB increases, the income effects will affect the quantity demanded of spirits in that the

PAGE 38

38 quantities demand for spirits will increase only 0.05% as compared to 0.80% when only the substitution effect is taken into account To sum up all pairings are substitutes except beer and ROB which have no price effect on each other If the U.S. increase s an additional dollar on import ed aggregated beverages, both beer and wine would gain 25 cents. Besides, ROB has the highest expenditure elasticity, which means that ROB would have the largest percentage change in the quantities demanded if U.S. import expenditure on these beverages incre ase s by 1%. At last, ROB spirits possesses the larg est positive Slutsky elasticity, and ROB beer has the biggest negative Cournot elasticity.

PAGE 39

39 Table 5 1 Test results for log likelihood ratio tests for different restri ctions in the Rotterdam, CBS AIDS, NBR, and g eneral model for aggregated beverages. General Rotterdam CBS AIDS NBR Log likelihood Value Unrestricted 191.28 184.57 185.41 186.50 185.44 Homogeneity 189.103 181.551 182.725 183.97 182.63 Homogeneity and Symmetry 187.98 179.05 180.23 181.83 180.45 Test Statistics Homogeneity 4.35(3) 6.04(3) 5.38(3) 5.06(3) 5.63(3) Homogeneity and Symmetry 2.25(6) 4.99(6) 5.00(6) 4.27(6) 4.34(6) Model selection 15.10(2) 12.76(2) 10.27(2) 12.96(2) a Table values 2 are 7.81 and 12.59 for 3 and 6 degrees of freedom, respectively, at 2 is 10.6 for 2 degrees of freedom at 0.01 level. b N umbers in parentheses are degrees of freedom for tests. c M odels with homogeneity and symmetry constraints imposed. Table 5 2 Conditional parameter estimates for U.S. import demand for aggregated beverages Aggregated Beverage, 1992 2012, AIDS model Price parameters Income parameters Beverage Spirits Wine Beer ROB (1) (2) (3) (4) (5) (6) Spirits 0.266 *** 0.093 *** 0.056 *** 0.117 *** 0.106 *** (0.013) (0.014) (0.012) (0.017) (0.035) Wine 0.179 *** 0.034 0.052* 0.020 (0.028) (0.023) (0.027) (0.041) Beer 0.089 ** 0.001 0.040 (0.040) (0.028) (0.030) ROB 0.168 *** 0.165 *** (0.037) (0.052) a Numbers in parenthesis are standard errors. b *** indicates this number is statistically different from zero at 0.01 level. c ** indicates the number is statistically different from zero at 0.05 level. d indicates the number is statistically different from zero at 0.1 level.

PAGE 40

40 Table 5 3 Marginal shares, conditional expenditure, Slutsky and Cournot elasticities of U.S. import demand for aggregated beverages at sample mean. Aggregated Beverage, 1992 2012, AIDS model Marginal shares Expenditure elasticities Price elasticities Beverage Spirits Wine Beer ROB (1) (2) (3) (4 ) (5 ) (6 ) (7 ) Slutsky Spirits 0.25 *** 0.70 *** 0.76 *** 0.27 *** 0.16 *** 0.33 *** Wine 0.25 *** 0.93 *** 0.35 *** 0.67 *** 0.13 0.19 Beer 0.20 *** 0.83 *** 0.24 *** 0.15 0.38 ** 0.00 ROB 0.31 *** 2.13 *** 0.80 *** 0.35 0.00 1.15 *** Cournot Spirits 1.00 *** 0.08 0.01 0.23 *** Wine 0.02 0.92 *** 0.09 0.06 Beer 0.05 0.08 0.57 *** 0.12 ROB 0.05 0.22 0.51 *** 1.46 *** a *** indicates this number is statistically different from zero at 0.01 level. b ** indicates the number is statistically different from zero at 0.05 level. c indicates the number is statistically different from zero at 0.1 level.

PAGE 41

41 CHAPTER 6 U.S. IMPORT DEMAND FOR BEER, WINE AND SPIRITS BY COUNTRY OF ORIGIN The U.S. import demand for beer, wine and spirits by country of origin are estimated using the yea rly time series data with the general demand system from 1992 to 2012. Conditional expenditure, own price elasticities, and cross price elasticities are calculated and reported on a per capita basis in various demand systems. The log likelihood and ratio v alues of four basic demand models and general demand systems are reported in tables as well. U.S. Import Demand for Beer by Country of Origin U.S. import demand for beer by country of origin is estimated base d on the general demand system. T hree equation demand system s are estimated from 199 2 to 2012 for beer imported into the U.S. from Mexico, Netherlands, and ROW For beer by country of origin, the best is demonstrated by testing restrictions and LR test s Accordingly, c onditional parameters estimates, c onditional expenditure elasticities and two types of price elasticities are presented in this section based on the NBR model Testing Restriction s and Model C hoice The log likel ihood and ratio values of the four basic demand m odels and the general demand system are reported in Table 6 1 for imported beer by country of origin The null hypothesis of homogeneity is not rejected for any model at the 0.05 significance level and t he nul l hypothesis of symmetry is not rejected for any model at the 0.05 s ignificance level. T he model selection results show that the NBR fits the imported beer by country of origin data b est Thus, the results for the NBR model are presented and discussed further in this section.

PAGE 42

42 Parameter E stimates NBR parameters and margin al shares are reported in Table 6 2 while conditional expenditure, Slutsky own price elasticities and cross price elasticities are shown in Table 6 3. Cournot own price elasticities and cross price elasticities are presented in Table 6 4. The own price par ameters are reported along the diagonal of columns (2) ( 4 ) of Table 6 2. All own price parameters of the three countries for imported beer are negative a nd less than one absolutely as expect ed ranging from 0.0 78 ( Netherlands ) to 0.2 50 (Mexico). The cross price parameters are reported as the non diagonal numbers of columns (2) ( 4 ) of Table 6 2. Among the conditional cross price parameters, one is negative and the other two are positive; only one is significantly different from zero 1) The result s show that only the Netherlands Mexico pairing is complements. All marginal shares parameters are reported in column ( 5 ) of Table 6 2 In the NBR model, the marginal shares equal for beer and are constant They are all significantly different from ze t benefit with 4 8 cents from an additional dollar being allocated to imported beer on a whole 30 cents and 22 cents f rom an additional dollar would be spent on beer from Net herlands and ROW, respectively. Conditional E xpenditure E lasticities The conditional expenditure elasticities of beer by country of origin are reported in column ( 2 ) of Table 6 3. As we explained in the former chapter, if the elasticity is higher than one, it is conditionally elastic, which indica tes the budget share for this source country will increase if U.S. import expenditure on this commodity increases; if it

PAGE 43

43 is lower than one, it should be conditionally inelastic, which means the budget share for this source country will decrease if U.S. imp ort expenditure on this co mmodity increases; if it is equal to one, it is unitary and suggest s that the budget share will not change if U.S. expendit ure on this commodity changes. All of these expenditure elasticities are positive and significantly different from The c ountry with an elas tic conditional expenditure elasticity is Mexico (1.3 3) for beer. Countries with inelast ic conditional expenditure elasticities are Netherlands (0.9 3) and ROW (0.69). These results indicate that if U.S increase s expenditure on imported beer by 1%, the budget share for beer of Mexico would incre ase while those of the others would decrease. Additionally, quantities demanded of imported b eer would increase by 1.33%, 0.93%, and 0.69% for Mexico, Netherlands, and ROW, respectively. Thus, Mexico has the most benefit to gain from an increase in import expenditures for beer. Two T ypes of P rice E lasticities Slutsky and Cournot elasticities are reported in columns ( 3 ) ( 5 ) of Table 6 3 and columns (2) ( 4 ) of Table 6 4. Conditional Slutsky (Cournot) own price elasticity for Mexico is 0. 68 ( 1.53 ). Netherlands and RO W ar e less own price inelastic with conditional Slutsky (Cournot) own price elasticity of 0. 24 ( 0. 86 ) and of 0. 52 ( 1.05 ), respectively. Compared with the difference between the conditional Slutsky and Cournot own price elasticities of each source country, M exico h as the biggest differenc e This indicates that Mexican beer is most e xpenditure sensitivity among imports into the U.S. import market. These results also point out different effects of price fluctuations on U.S. beer import demand. If the price of Dutch beer goes up by 1%, the un compensated quantity demanded for Dutch beers would go down only 0. 86 %. However, Mexico and

PAGE 44

44 ROW imported beer are much more influenced by this kind of own price fluctuations. More specifically, the uncompensated quantity demand of Mexico and ROW imported b eer into the U.S. market would decrease about 1.53% and 1.05 % respectively In terms of condit ional cross price elasticities, the ROW Mexico ( 0.54 ) pairing possesses the largest posi tive cross price elasticity, which indicates that a 1% increase in the pr ice of beer from the Mexico will increase the quantity (compensated) demanded for ROW beer by 0.54% in the U.S. import market for Netherlands ROW is negative, but statistically the same as zero sending some evidence that they are complements for the U.S beer market. To sum up, imported beer from the Neth er lands and Mex ico may be complements but all other pairings are substitutes If U.S. increase s an additional dollar on imported beer, Mexico would gain the most be nefit with 4 8 cents. Besides, Mexico has the highest expenditure elasticity, which means that Mexico has the largest percentage change in the quantities demanded when U.S. import expenditure on beer increase s by 1%. At last, ROW Mexico possesses the larges t positive Slutsky elasticity and Mexico has the biggest negative Cournot own price elasticity ( 1.53) U.S. Import Demand for Wine by Country of Origin U.S. import demand for wine by country of origin is estimated based on the general demand system. F ive equation demand system s are estimated from 1992 to 2012 for imported wine into the U.S. market from Italy, France, Australi a, Chile, and ROW In this section LR tests indicate that the AIDS model fits the data best. Conditional parameters estimates, conditional expenditure elasticities and two types of conditional price elasticities are presented in this section.

PAGE 45

45 Testing R estriction s and M odel C hoice The log likelihood values and ratio values of the four basic demand m odels and general demand system are reported in Table 6 5 for imported wine. The null hypothesis of homogeneity is not rejected for any model at the 0.05 significance level When the symmetry restriction is added, the null hypothesis of symmetry is not rejected for any model at the 0.05 significance level. The model selection indicates that Rotterdam and NBR a re rejected by the general demand system and that the AIDS is better than the CBS. Also, the estimates of a nd are 1.59 and 0.91 with standard errors 0.48 and 0.55, respectively. Therefore, the results for the AIDS model are pre sented and discussed in this section. For the wine estimation the goodness of fit measure is = 0.45 indicating that the whole system explains 45 % of the variation in allocation. Parameter E stimates The AIDS parameters for imported wine are reported in Table 6 6 while marginal shares, conditional expenditure, Slutsky own price elasticities and cross price elastici ties are shown in Table 6 7. Cournot own price and cross price elasticities are presented in Table 6 8. The own price parameters are reported alon g the diagonal of columns (2) (6 ) of Table 6 6. The own price parameters of five countries from wine are nega tive and less than one absolutely, ranging from 0.097 (Chile) to 0.153 (Australia). These five The cross price parameters are reported as the non di agonal numbers of columns (2) (6 ) of Table 6 6. All are positive except that of the France Chile pairing.

PAGE 46

46 Among the conditional cross price parameters, only three are significantly different from zero. All marginal shares parameters are calculated and reported in column (2) of Table 6 7. In the AIDS, the marginal shares equal to They are all significantly different f the U.S. were to consume an additional dollar of imported wine, 34 cents would go towards purchasing French wines. Italy would also benefit 26 cents. While 19 cents and 14 cents of the additional dollar would be allocated to ROW and Australia, respectively. Chile w ould receive only 8 cents of an additional dollar. Thus, the French wine industry would gain the most if an additional dollar were allocated on imported wine. Conditional E xpenditure E lasticities The conditional expenditure elasticities of wine by country of origin are repor ted in column (3) of Table 6 7. As shown by the results, Chile an wine (1.43) h a s the highest expenditure elasticity. Australia (1.10) and ROW (1.07) have conditional elastic expenditure elasticity too. Italy (0.93) and France (0.92) have inelastic ones With these results, the budget share for Chile, Australia and ROW would rise and others would fall if U.S. import expenditure for wine expands by 1% while the quantities demanded of wine would increase by 1.43%, 1.10%, 1.07%, 0.93%, and 0.92% for Chile, Australia, ROW, Italy, and France, respectively Two T ypes of P rice E lasticities Slutsky and Cournot elasticities are reported in columns (4) ( 8 ) of Table 6 7 and columns (2) ( 6 ) of Table 6 8. For Chilean wine no matter whether Slutsky and Cournot, has the most elastic own price elasticity Its Slutsky (Cournot) own price elasticity i s 1.81 ( 1.88 ). Italian and French wine are the least own price inelastic. Conditional

PAGE 47

47 Slutsky (Cournot) own price elasticity for Italian wine is 0.43 ( 0. 69 ) and French wine is 0.41 ( 0. 75 ). Australia n and ROW wine are more own price elastic than those of Italy and France, but less than that of Chile. They have conditional Slutsky (Cournot) own price elasticities 1.20 ( 1.3 4 ) and 0.75 ( 0.9 3 ), respectively. Similar to the previous analysis, we contrast with the difference between the conditional Slutsky and Cournot own price elasticities of each source country. The findings discover that France has the biggest difference. It demonstrates that French imported wine is the most expenditure sensitivity to an own price change in the U.S. import market. These results also indicate various effects for price changes on U.S. wine import demand. Chilean and Au stralia n wine are mostly influenced if the ir own price s increase by 1%. More specifically, the uncompensated quantity demand of wine from Chile and Australia into the U.S. market would decrease 1.8 8 % and 1.3 4 %, respectively. Through the results for Slutsky and Cournot cross price elasticities the U .S import demand for win e from main source countries is inelastic except that of Chile Australia as the absolute value of all other Slutsky (Cournot) cross price elasticities are less than 1. For Chile, the Slutsky (Cournot) cross price elasticity with A ustralia is 1.21 (1.02), which is larger than the cross price elasticities with other countries. It indicates that the U.S. import demand for Chilean wi ne is influenced by Australian wine price the most. The s ame situation is true for Australia Chile. It s hows that they are strong substitutes and Australia holds the price advantage, as the Slutsky (Cournot) cross price elasticity for Chile Australia is larger than that of Australia C hile. Italian and Australia n wines are the only ones that are substitutes f or all wines. A compensated price change in Italian imported wines has a small positive influence on the quantity

PAGE 48

48 demanded for all oth er wines. In terms of Australia and RO W, a relatively large positive e ffect would be true for the quantity d emanded for al l other wines. However, the price changes of the imported wines have more positive effects on the quantity demanded of these source countries. For example, a 1% increase in the price of Chilean wines will increase by 0.51% the compensated quantit y demanded for Australian wines. more, a 1% increase in the price of French wines will increase by 0.46% the compensated quantity d emanded for ROW wines. To sum up, France and Chile are complements and all other pairings are substitutes. If U.S. inc rease s an additional dollar on imported wine, France would gain the most benefit with 34 cents. Besides, Chile has the highest expenditure elasticity, which means that Chile an wine would have the largest percentage change in the quantities demanded if U.S. import expen diture were to increase by 1%. Comparing the difference between Slutsky and Cournot own price elasticities, France has th e biggest difference. At last, o nly Chile Aus tralia h as an elastic result in terms of cross price elasticity. U.S. Import Demand for Spirits by Country of Origin U.S. import demand for spirits by country of origin is estimated based on the general demand system. F ive equation demand system s are estimated from 1992 to 2012 for imported spirits into the U.S. market from France United K ingdom, Mexico, Sweden, and ROW In this part, it is found that the NBR model fits the data best through the result from testing restrictions and LR test s Conditional parameters estimates, conditional expenditure elasticities and two types of pr ice elasticities are presented in this section based on the NBR model

PAGE 49

49 Testing R estriction s and M odel C hoice The log likelihood and ratio values of the four basic demand models and general demand system are reported in Table 6 9 for spirits. The results for spirits indicate that the null hypothesis of homogeneity and symmetry is not rejected for any model at the 0.05 significance level. The model selection results show that all four demand model s are not rejected by the g eneral system while NBR fits the general system best implying NBR model fits the spirits data better than the other three. Accordingly, results using NBR are reported and analyzed further in this section. For the spirits estimation the goodness of fit measure is = 0. 49 indicating that the whole system describes 49 % of the variation in allocation. Parameter E stimates NBR parameters and marginal shares for spirits are reported in Table 6 10 while conditional expenditure, Slutsky ow n price and cross price elasticities are shown in Table 6 11. Cournot own price and cross price elasticities are presented in Table 6 12. The own price parameters are reported along the diagonal of columns (2) ( 6 ) of Table 6 10. The own price parameters of five countries from spirits are negative and less than one absolutely as expect ed ranging from 0.013 (Mexico) to 0.197 (France). All parameters are that for Mexico. The cross price parameters are reported as the non diagonal numbers of columns (2) ( 6 ) of Table 6 10. All are positive except that of the France Mexico and United Kingdom Sweden pairings. Among the conditional cross price parameters, six are significantly different from zero. All margi nal shares parameters are reported in column (7) of Table 6 10 In the NBR model, the marginal shares equal for spirits and are constant They are all

PAGE 50

50 If an additional dollar were spent on imported spirits 40 cents of that additional dollar would be allocated to French spirits ROW would gain 22 cents while th e United Kingdom and Mexico would obtain 19 cents and 11 cents, respectively. Implied by the results, Sweden would b e only allocated 8 cents if U.S. i ncreases its expenditure by an additional dollar on spirits imports. Similar to the circumstance of wine, the French industry would be the most beneficial exporter if an additional dollar were spent on imported spirits. Conditional E xpenditure E lasticitie s The conditional expenditure elasticities of spirits by country of origin are reported in column ( 2 ) of Table 6 11. The budget share of spirits from France would expand the most if U.S import expenditure on spirit s were to go up by 1%. Sweden and Mexico import share would also increase slightly. Countries with inelastic expenditure elasticities are ROW (0.71) and the United Kingdom (0.70). If U.S. import expenditure for spirits increases by one percent the quantities demanded of spirits wou ld increase by 1.68%, 1.05%, 1.03%, 0.71%, and 0.70% f rom France, Sweden, Mexico, ROW, and the United Kingdom, respectively. Two T ypes of P rice E lasticities Slutsky and Cournot elasticities are reported in columns ( 3 ) ( 7 ) of Table 6 11 and columns (2) ( 6 ) of Table 6 12. For France, no matter whether the Slutsky and Cournot, its own price is most elast ic, with its Slutsky (Cournot) own price elasticity being 0.82 ( 1.46 ). Mexican spirits is the least own price inelastic. The c onditional Slutsky (Cournot) ow n price elasticity for Mexican spirits is 0.13 ( 0. 34 ). ROW spirits has an elastic Cournot own price elasticity, but not for Slutsky. Its c onditional Slutsky (Cournot) own price elasticity is 0.56 ( 1.10). The Slutsky (Cournot) elasticities of the other two

PAGE 51

51 sources are inelastic. Through the comparison of the Cournot own price value with the corresponding Slutsky one the findings discover that France has the largest difference It indicates that French imported spirits are the most expenditure sensitivity for the U.S. import market. In addition, results also illustrate that there are various effects for price variations on U.S. spirits import demand. For example, the uncompensated quantity demand of French spirits into the U .S. market would decrease about 1.46 % of its own price increased by one percent Conversely, the uncompensated quantity demand of Mexican spirits into the U.S. market would just shrink about 0. 34 % from a 1.0% increase in its own price Through the results for Slutsky and Cournot cross price elasticities the U.S import demand for spirits from main sourc e countries are inelastic as the absolute value of their Slutsky (Courn ot) cross price elasticities are less than 1. For Mexican spirits the Slutsky (Cour not) cr oss price elasticity with French spirits is 1.21 (1.02), which is larger than the cross price elast icities with other countries. This indicates t hat the U.S. demand for Mexican imported spirits is influenced by the price of French spirits the most. ROW spirits is the only one that a substitute with all other spirit s. A compensated price change in ROW spirit s has a small positive influence on the quantity demanded for all other spirits. Besides, t he price changes of the imported spirits from these sou rce countries have more positive effects on the quantity demanded of ROW For example a 1% increase in the price of French spirits will increase the compensated quantity demanded for ROW spirits by 0.33%. Another interes ting fact is that, when we take the income effect plus the substitution effect into account an increase in the price of spirits produced from France, UK and ROW would decrease the demand for all other spirits.

PAGE 52

52 For example, if there is a 1% increase for th e spirits made from the United Kingdom, the demand from ROW will decrease 0.38%. Additionally, increa se in the prices of other imported spirits would decrease the uncompensated quantity demanded for France, UK and ROW spirits. For example, a 1% increase in the price of Sweden (Mexico) spirits would decrease the quantity (uncompensated) demanded of UK (France) spirits by approximately 0.8% (0.7%). To sum up, France Mexico and United Kingdom Sweden are complements and all other pairings are substitutes. If U.S. increase s an additional dollar on imported spirits, France would gain the most benefit with 40 cents. Besides, France has the highest expenditu re elasticity, which indicates that France would have the largest percentage change in the quantities demand ed if U.S. import expenditure were to increase by 1%. Comparing the difference between Slutsky and Cournot own price elasticities, France has the biggest difference. At last, an increase in the price of spirits from France, United Kingdom, and ROW would de crease the uncompensated demand for all other spirits.

PAGE 53

53 Table 6 1 Test results for log likelihood ratio tests for different restrictions in the Rotterda m, CBS, AIDS, NBR, and general m odel for beer a 2 are 5.99 and 7.81 for 2 and 3 degrees of freedom, respectively, at 0.05 level. b N umbers in parentheses are degrees of freedom for tests. c M odels with homogeneity and symmetry constraints imposed. Table 6 2 Conditional parameter estimates for U.S. import demand for beer by country of origin. Beer, 1992 2012, NBR model Country Price parameters Marginal shares (1) (2) (3) (4) (5) Mexico 0.250 0.083 0.167 0.48 *** (0.152) (0.108) (0.093) (0.047) Netherlands 0.078 0.004 0.30 *** (0.103) (0.057) (0.030) ROW 0.163 0.22 *** (0.086) (0.041) a Numbers in parenthesis are standard errors. b *** indicates this number is statistically different from zero at 0.01 level. c ** indicates the number is statistically different from zero at 0.05 level. d indicates the number is statistically different from zero at 0.1 level. General Rotterdam CBS AIDS NBR Log likelihood value Unrestricted 121.31 120.82 120.48 120.76 121.05 Homogeneity 119.62 119.33 118.39 118.69 119.57 Homogeneity and Symmetry 119.47 119.03 118.38 118.67 119.28 Test statistics Homogeneity 3.37(2 ) 2.99(2 ) 4.19(2 ) 4.14(2 ) 2.96(2 ) Homogeneity and Symmetry 0.29(3 ) 0.60(3 ) 0.02(3 ) 0.02(3 ) 0.59(3 ) Model selection 0.58 (2) 2.46 (2) 1.86 (2) 0.09 (2)

PAGE 54

54 Table 6 3 Marginal shares, conditional expenditure and Slutsky (compensated price elasticities) of U.S. import demand for beer by country of origin at sample mean. Beer, 1992 2012, NBR model Expenditure elasticities Slutsky p rice elasticities Country (1) (2) Mexico (3) Netherlands (4) ROW (5) Mexico 1.33 *** 0.68 0.23 0.46 Netherlands 0.93 *** 0.26 0.24 0.01 ROW 0.69 *** 0.54 0.01 0.52 a *** indicates this number is statistically different from zero at 0.01 level. b ** indicates the number is statistically different from zero at 0.05 level. c indicates the number is statistically different from zero at 0.1 level. Table 6 4 Cournot (unco mpensated price elasticities) of U.S. import dem and for beer by country of origin at sample mean. Beer, 1992 2012, NBR model Cournot p rice elasticities Country Mexico Netherlands ROW (1) (2) (3) (4) Mexico 1.53 *** 0.52 0.27 Netherlands 0.45 0. 86 *** 0.62 *** ROW 0.08 0.56 *** 1.05 ** a *** indicates this number is statistically different from zero at 0.01 level. b ** indicates the number is statistically different from zero at 0.05 level. c indicates the number is statistically different from zero at 0.1 level. Table 6 5 Test results for log likelihood ratio tests for different restrictions in the Rotterdam, CBS, AIDS, NBR, and g eneral model for wine. General Rotterdam CBS AIDS NBR Log likelihood value Unrestricted 262.33 258.09 260.89 261.76 258.26 Homogeneity 260.03 254.73 258.35 259.19 254.98 Homogeneity and Symmetry 252.00 248.11 250.76 251.37 248.05 Test statistics Homogeneity 4.60(4) 6.73(4) 5.07(4) 5.14(4) 6.57(4) Homogeneity and Symmetry 16.06(10) 13.24(10) 15.20(10) 15.65(10) 13.86(10) Model selection 10.60(2) 3.35(2) 1.67(2) 10.10(2) a 2 are 5.99, 9.49 and 18.31 for 2, 4 and 10 degrees of freedom, respectively, at 0.05 level. b N umbers in parentheses are degrees of freedom for tests. c M odels with homogeneity and symmetry constraints imposed.

PAGE 55

55 Table 6 6 Condit ional parameter estimates for U.S. import demand for wine by country of origin a Numbers in parenthesis are standard errors. b *** indicates this number is statistically different from zero at 0.01 level. c ** indicates the number is statistically different from zero at 0.05 level. d indicates the number is statistically different from zero at 0.1 level. Table 6 7 Marginal shares, conditional expenditure and Slutsky (compensated price elasticities) of U.S. import demand for wine by country of origin at sample mean. Wine, 1992 2012, AIDS model Marginal shares Expenditure elasticities Slutsky p rice elasticities Country Italy France Australia Chile ROW (1) (2) (3) (4 ) (5 ) (6 ) (7 ) (8 ) Italy 0.26 *** 0.93 *** 0.43 *** 0.13 0.08 0.13 ** 0.09 France 0.34 *** 0.92 *** 0.09 0.41 *** 0.11 0.01 0.22 *** Australia 0.14 *** 1.10 *** 0.18 0.32 1.20 *** 0.51 *** 0.19 Chile 0.08 *** 1.43 *** 0.67 ** 0.08 1.21 *** 1.81 *** 0.01 ROW 0 .19 *** 1.07 *** 0.14 0.46 *** 0.14 0.00 0.75 *** a *** indicates this number is statistically different from zero at 0.01 level. b ** indicates the number is statistically different from zero at 0.05 level. c indicates the number is statistically different from zero at 0.1 level. Wine, 1992 2012, AIDS model Country Price parameters Income parameters Italy France Australia Chile ROW (1) (2) (3) (4) (5) (6) (7) Italy 0.118 *** 0.035 0.023 0.036 ** 0.024 0.018 (0.037) (0.028) (0.025) (0.015) (0.030) (0.031) France 0.152 *** 0.041 0.004 0.081 *** 0.030 (0.053) (0.035) (0.012) (0.026) (0.061) Australia 0.153 *** 0.065 *** 0.024 0.013 (0.038) (0.014) (0.030) (0.044) Chile 0.097 *** 0.001 0.023 (0.010) (0.014) (0.012) ROW 0.130 *** 0.013 (0.038) (0.029)

PAGE 56

56 Table 6 8 Cournot (uncompensated price elasticities) of U.S. import demand for wine by country of origin at sample mean. Wine, 1992 2012, AIDS model Cournot p rice elasticities Country Italy France Australia Chile ROW (1) (2 ) (3 ) (4 ) (5 ) (6 ) Italy 0.69 *** 0.22 ** 0.04 0.08 0.07 France 0.16 0.75 *** 0.01 0.06 0.06 Australia 0.12 0.08 1.34 *** 0.45 *** 0.00 Chile 0.28 0.61 *** 1.02 *** 1.88 *** 0.24 ROW 0.15 0.07 0.00 0.05 0.93 *** a *** indicates this number is statistically different from zero at 0.01 level. b ** indicates the number is statistically different from zero at 0.05 level. c indicates the number is statistically different from zero at 0.1 level. Table 6 9 Test resul ts for log likelihood ratio tests for different restrictions in the Rotterdam, CBS, AIDS, NBR, and g eneral model for spirits. General Rotterdam CBS AIDS NBR Log likelihood value Unrestricted 278.47 278.07 277.58 277.22 277.98 Homogeneity 275.79 275.07 274.56 274.83 275.43 Homogeneity and Symmetry 272.27 271.33 271.08 271.54 271.89 Test statistics Homogeneity 5.37(4) 6.06(4) 6.04(4) 4.78(4) 5.09(4) Homogeneity and Symmetry 7.04(10) 7.45(10) 6.97(10) 6.57(10) 7.09(10) Model selection 1.47(2) 2.45(2) 1.92(2) 0.71(2) a 2 are 5.99, 9.49 and 18.31 for 2, 4 and 10 degrees of freedom, respectively, at 0.05 level. b N umbers in parentheses are degrees of freedom for tests. c M odels with homogeneity and symmetry constraints imposed.

PAGE 57

57 Table 6 10 Conditional parameter estimates for U.S. import demand for spirits by country of origin. Spirits, 1992 2012, NBR model Country Price parameter Marginal shares France United Kingdom Mexico Sweden ROW (1) (2) (3) (4) (5) (6) (7) France 0.197 *** 0.088 *** 0.023 0.029 *** 0.103 *** 0.40 *** (0.037) (0.021) (0.018) (0.010) (0.030) (0.054) United Kingdom 0.121 *** 0.009 0.018 0.042 0.19 *** (0.026) (0.013) (0.012) (0.025) (0.037) Mexico 0.013 0.017 ** 0.010 0.11 *** (0.016) (0.008) (0.018) (0.034) Sweden 0.049 *** 0.021 0.08 *** (0.008) (0.012) (0.021) ROW 0.l76 *** 0.22 *** (0.039) (0.050) a Numbers in parenthesis are standard errors. b *** indicates this number is statistically different from zero at 0.01 level. c ** indicates the number is statistically different from zero at 0.05 level. d indicates the number is statistically different from zero at 0.1 level. Table 6 11 Marginal sha res, conditional expenditure and Slutsky (compensated price elasticities) of U.S. import demand for spirits by country of origin at sample mean. Spirits, 1992 2012, NBR model Slutsky p rice elasticities Expenditure elasticities Country France United Kingdom Mexico Sweden ROW (1) (2 ) (3 ) (4 ) (5 ) (6 ) (7 ) France 1.68 *** 0.82 *** 0.37 *** 0.09 0.12 *** 0.43 *** United Kingdom 0.70 *** 0.33 *** 0.45 *** 0.03 0.07 ** 0.16 Mexico 1.03 *** 0.22 0.09 0.13 0.17 ** 0.10 Sweden 1.05 *** 0.38 *** 0.23 0.22 ** 0.65 *** 0.28 ROW 0.71 *** 0.33 *** 0.13 0.03 0.07 0.56 *** a *** indicates this number is statistically different from zero at 0.01 level. b ** indicates the number is statistically different from zero at 0.05 level. c indicates the number is statistically different from zero at 0.1 level.

PAGE 58

58 Table 6 12 Cournot (uncompensated price elasticities) of U.S. import demand for spirits by country of origin at sample mean. Spirits, 1992 2012, NBR mo del Cournot p rice elasticities Country France United Kingdom Mexico Sweden ROW (1) ( 2 ) ( 3 ) ( 4 ) ( 5 ) ( 6 ) France 1.46 *** 0.35 *** 0.37 *** 0.08 ** 0.41 *** United Kingdom 0.08 0.91 *** 0.14 ** 0.19 *** 0.38 *** Mexico 0.71 *** 0.46 *** 0.34 0.01 0.54 *** Sweden 0.12 0.78 *** 0.01 0.81 *** 0.36 ** ROW 0.08 0.33 *** 0.14 *** 0.06 1.10 *** a *** indicates this number is statistically different from zero at 0.01 level. b ** indicates the number is statistically different from zero at 0.05 level. c indicates the number is statistically different from zero at 0.1 level.

PAGE 59

59 CHAPTER 7 CONCLUSION S U.S. import demand for beer, wine, and spirits are estimated and U.S. import demand for beer, wine, and spirits by country of origin is estimated using the differential approach. Conditional expenditure, own price, and cross price elasticities of import demand are calculated from the parameters of the models. In this paper, we applied the multistage budgeting and demand theory to explore the import demand for the U.S. alcoholic beverage market. The general demand system proved to be an efficient tool to pick the best from the basic demand models Rotterdam, AIDS and their variants to fit the specific data. According to the results of LR tests and mode l selection, we utilize the AIDS model for aggregated beverages in the first stage. NBR (beer and spirits) and the AIDS (wine) model are selected to best fit the data in the second stage. More specifically, the NBR model suggests that Rotterdam type income and AIDS type price responses better explain U.S. import demand for beer and spirits than do the other three models ; for aggregated beverages and wine, both AIDS type income and price coefficients are found to best fit the import data for these goods. The functional forms of the R otterdam and the AIDS models differ (e.g. income and price terms), which lead to the important difference in results. For example, marginal shares are constant in the Rotterdam and NBR models but are variable in the AIDS and CBS m odels Price elasticities estimation are also found to be functional form specific. An important contribution of this paper is the parameter estimation and elasticity analysis for these beverage commodities. Given the large import budget share for beverag es in total U.S. agricultural imports, it is important to have some elasticity estimates that may be used to drive economic models that consider effects of imports,

PAGE 60

60 consumer preference and public policy requirements for beer, wine and spirits. The key fin dings are obtained from the results. In the first stage, spirits and wine may keep the budget share while beer may decrease the budget share Beer, wine and spirits are substitutes implied by the cros s price parameters. A s light complements effect is fou nd between beer and ROB. Besides, imported w ine would gain the most quantity demanded among the imports of beer, wine, and spirits if the U.S. total bever ages expendi ture were to increase At last, compared with beer and spirits, imported wine h as the strongest expenditur e se nsitivity in U.S. import market. In the second stage, several implications obtained from the results are important. First, some countries may gain the most benefit if U.S import expenditure expands. This is true for Mexico (b eer), Chile (wine) and France (spirits). The value of their expenditure elasticities for each commodity are greater than unity Second, various price strategies may work f or different source countries by each commodity. For example, Chile and Australia may increase their revenues by decreasing their price for their in order to gain increased revenues. Third, all source countries for wine are substitutes except for France Chil e pairing. Compare d with the results of beer and spirits, this relationship is much more competitive between these import ed wines. Fourth, when comparing own price and cross price elasticities, estimation shows that consumers prefer to choose beer than win e and spirits More specifically, a unitary reduction in the price of Mexican beer will increase compensated demand by only 0. 68 %. However,

PAGE 61

61 when income effects are considered, a 1% decrease in the price of Mexican beer would contribute to a 1.53 % expansion in its quantity demanded. This paper still has some important issu es left unanswered. One is to explore whether the use of quarterly or monthly data may better explain the import demand relationship between these expor ters and U.S. and better fit the model chosen by the general demand system. Additionally, if the domestic beverages data were available, the results might be different Due to the huge demand for domestic beverages, domestic beverages data may swamp the imported ones which may lead t o insignificant results for import demand relationship with those exporters Finally, i mport demand estimation for ROB as a group may be highly potential for future research and extension. The U.S. import demand for non alcoholic water, unsweetened water, and fermented beverages may be estimated by country of origin in the future work.

PAGE 62

62 LIST OF REFERENCE S Amemiya, T. ( 1985 ) Advanced e conometrics Cambridge: Harvard University Press. Andayani, S. R. M., & D. S. Tilley. ( 1997 ) Dem and and competition among supply s ources : The Indonesian fruit import m arket. Journal of Agricultural and Applied Economics 29 279 289. Andrikopoulos, A. A., J. A. Brox, & E. Carvalho. ( 1997 ) The d emand for d omestic and i mported a lcoholic b everages in Ontario, Canada: A d ynamic s imultaneous e quation a pproach. Applied Economics 29 945 953. Barnett, W. A., & O. Seck. ( 2008 ) Rotterdam m odel v ersus Almost Ideal Demand System: W ill the b est s pecification p lease s tand u p? Journal of Applied Econ ometrics 23, 795 824. Barten, A. P. ( 1969 ) Maximum li kelihood e stimation of a c omplete s ystem of d emand e quations. European Economic Review 1 7 73. Barten, A. P. ( 1977 ) The s ystems of c onsumer d emand f unctions a pproach: A r eview. Econometrica 45, 23 51. Barten, A. P. ( 1993 ) Consumer a llocation m odels: c hoice of f unctional f orm. Empirical Economics 18, 129 158. Brown, M. G., J Y. Lee, & J. L. Seale, Jr. ( 1994 ) Demand r elationship a mong j uice b everages: A d ifferential d emand s ystem a pproach. Journal of Agricultural and Applied Economics 26 417 429. Brown, M. G., R. M. Behr, & J Y. Lee. ( 1994 ) Conditional d emand and e ndogeneity? A c ase s tudy of d emand for j uice p roducts Journal of Agricultural and Resource Economics 19, 129 140. Carew, R W. J. Florkowski, & S. He. ( 2005 ) Demand for d omestic and i mported t able w ine in British Columbia: A s ource differentiated Almost Ideal Demand System a pproach. Canadian Journal of Agricultural Economics 52 183 199. Clements, K. W., & L. W. Johnson. ( 1983 ) The d emand for b eer, w ine and s pirits: A s ystem wide a pproach. Journal of Business 56, 273 304. Deaton, A., & J. Muellbauer. ( 1980 ) An Almost Ideal Demand System. American Economics Review, 70, 312 326. Eakins, J. M., & L. A. Gallagher. ( 2003 ) Dynamic Almost Ideal Demand Systems: A n e mpirical a nalysis of a lcohol e xpenditure in Ireland. Applied Economics 35, 1025 1036.

PAGE 63

63 Eales, J. S., & L. J. Unnevehr. ( 1988 ) Demand for b eef and c hicken p roducts: Separability and s tructural c hange. American Journ al of Agricultural Economics 70, 521 532. Feleke, S. T., & L. M. Walters. ( 2005 ) Global c offee i mport d emand in a n ew e ra: I mplications for d eveloping c ountries. Review of Applied Economics 1, 223 237. Gallup, Inc. ( 2013 ) U.S. d rinkers d ivide between b eer and w ine as f avorite. Retrieved November 1, 2013 from http://www.gallup.com/poll/163787/drinkers divide beer wine favorite.aspx Gao X., E. Wailes, & G. Cramer. ( 1 995 ) A m icroeconometric m odel a nalysis of US c onsumer d emand for a lcoholic b everages. Applied Economics 27, 59 69. Han, T., & T. I. Wahl. ( 1998 ) r ural h ousehold d emand for f ruit and v egetables. Journal of Agricultural and Applied Economics 30, 141 150. Keller, W. J, & J. Van Driel. ( 1985 ) Differential c onsumer d emand s ystems. European Economic Review 27, 375 390. LaFrance, J. ( 1991 ) When i s e e s eparable d emand m odels? Western Journal of Agricultural Economics 16, 49 6 2. Lee, J. Y., J. L. Seale, Jr., & P.A. Jierwiriyapant. ( 1990 ) .Do t rade a greements h elp US e xports? A s tudy of the Japanese c itrus i ndustry. Agribusiness 6 505 514. Lee, J. Y., M. G. Brown, & J. L. Seale, Jr. ( 1994 ) Model c hoice in c onsumer a nalysis: Taiwan, 1970 89. American Journal of Agricultural Economics 76, 504 512 Lee, Y. J., P. L. Kennedy, & B. M. Hilbun. ( 2009 ) A d emand a nalysis of the Korean w ine m arket u sing an u nrestricted s ource d ifferentiated LA/AIDS m odel Journal of Wine Economics 4, 185 200. National Center for Health Statistics. (2012). Summary h ealth s tatistics for U.S. a dults: National h ealth i nterview s urvey. Retrieved November 1, 2013 from http://www.cdc.gov/nchs/fastats/alcohol.htm Neves, P. D. ( 1987 ) Analysis of c onsumer d emand in Portugal, 1958 1981. Memoire de maitrise en sciences economiques, Universite Catholique de Louvain, Louvain la Neuve. Neves, P. D. ( 1994 ) A c lass of d ifferentia l d emand s ystems. Economics Letters 44 83 86. Seale, J. L., M. A. Marchant, & A. Basso. ( 2003 ) Imports v ersus d omestic p roduction: A d emand s ystem a nalysis of the US r ed wi ne m arket. Review of Agricultural Economics 25, 187 202.

PAGE 64

64 Seale, J. L., Jr., A. L. Sparks, & B. M. Buxton. ( 1992 ) A R otterdam a pplication to i nternational t rade in f resh a pples: A d ifferential a pproach Journal of Agricultural and Resource Economics 17, 138 149. Seale, J. L., Jr., L. Zhang, & M. R. Traboulsi. ( 2013 ) U.S. i mport d emand and s upply r esponse for f resh t omatoes, c antaloupes, o nions, o ranges, and s pinach. Journal of Agricultural and Applied Economics 45, 435 452. Sparks, A. L. ( 1992 ) A s ystem w ide a pproach to i mport d emand for U.S. f resh o ranges. Agribusiness 8, 253 260. Theil, H., & K. W. Clements. ( 1978 ) A d ifferential a pproach to U.S. i mport d emand. Econ. Letters 20, 157 160. United Nations. ( 2012 ) Internationa l T rade Statistics Yearbook. Retrieved November 1, 2013 from http://comtrade.un.org/pb/FileFetch.aspx?docID=4084&type=commodity%20pag es U.S. Department of Agriculture ( 2013 ) Standard q uery, Global Agricultural Trade System, Foreign Agricultural Service Retrieved November 1, 2013 from http://apps.fas.usda.gov/Gats/ExpressQuery1.aspx U.S. Department of Agriculture. (2013). Total v alue o f U.S. a gricultural t rade and t rade balance, m onthly Retrieved November 1, 2013 from http://www.ers.usda.gov/data product s/foreign agricultural trade of the united states (fatus)/us agricultural trade data update.aspx#.U0QDBYafi o U.S. Department of Commerce. ( 2013 ) Historic d ata for p opulation e stimates of U.S. Census Bureau Retrieved February 1, 201 4 from http://www.census.gov/popest/data/historical/index.html Wang J, et al. ( 1996 ) U.S. c onsumer d emand for a lcoholic b everages: Cross s ection e stimation of d emographic and e conomic e ffects Review of Agricultural Economics 18, 477 489. Wang X, & M. Reed. ( 2013 ) Estimation of i mport d emand for f ishery p roducts in the U.S. u sing the s ource differentiated AIDS m odel. Paper presented at the AAEA&CAES Joint Annual Meeting, Washing DC, August 4 6. Worl d Health Organization. ( 2011 ) Global s tatus r eport on a lcohol and h ealth. Retrieved November 1, 2013 from http://www.who.int/substance_abuse/publications/global_alcohol_report/msbgsru profiles.pdf

PAGE 65

65 BIOGRAPHICAL SKETCH Bo Gao was a graduate student of the Master of Science program in Food and Resource Economics Department at the University of Florida. Bo was bor n in Hulunbeier, China in 1990. He attended University of International Relations in China After successfully graduated, he was admitted to the University of Florida, majo red in food and resource economics. He graduated in May 2014. His major area is international economics and applied econometrics. After graduation, he may continue his academic work in agricultural economics or go back to China to seek for a work position related to economics.