Citation
Socioeconomic determinants of at-home seafood consumption

Material Information

Title:
Socioeconomic determinants of at-home seafood consumption a limited dependent variable analysis of existing and latent consumers
Alternate title:
Seafood consumption
Creator:
Keithly, Walter R
Publication Date:
Language:
English
Physical Description:
x, 230 leaves : ill. ; 28 cm.

Subjects

Subjects / Keywords:
Estimated taxes ( jstor )
Family size ( jstor )
Fish ( jstor )
Homes ( jstor )
Households ( jstor )
Income estimates ( jstor )
Mathematical variables ( jstor )
Seafoods ( jstor )
Shellfish ( jstor )
Single status ( jstor )
Dissertations, Academic -- Food and Resource Economics -- UF
Food and Resource Economics thesis Ph. D
Food consumption ( fast )
Seafood ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1985.
Bibliography:
Includes bibliographical references (leaves 226-229).
Additional Physical Form:
Also available online.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Walter R. Keithly, Jr.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
020965746 ( ALEPH )
13992496 ( OCLC )

Downloads

This item has the following downloads:


Full Text












SOCIOECONOMIC DETERMINANTS OF AT-HOME SEAFOOD
CONSUMPTION: A LIMITED DEPENDENT VARIABLE
ANALYSIS OF EXISTING AND LATENT CONSUMERS
















BY

WALTER R. KEITHLY, JR.















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



UNIVERSITY OF FLORIDA

1985
















ACKNOWLEDGEMENTS


Numerous people have helped in making this study and my graduate

program possible. Sincerest appreciation is extended to my chairman,

Fred J. Prochaska. He was always there to give me encouragement and

advice even during strained times. He should be credited with my

success but not failure in the academic field.

Dr. Scott Shonkwiler gave freely of his time in helping me with

questions of a statistical nature. Similarly, Drs. Cato, Kilmer, and

Otwell devoted their time to assure a better quality product than would

otherwise have been the case. Hopefully, their constructive criticisms

of this study make it more useful for the groups for which it is

intended.

I wish also to thank the Food and Resource Economics Department at

the University of Florida and Florida Sea Grant for giving me the

opportunity to pursue a graduate program and for providing financial

support.

I wish to thank Janet Eldred for the typing she did on this

manuscript. It sometimes got complicated sending everything through

the mail.

Finally, I wish to thank my parents for giving me the opportunity

to pursue a graduate career. Without their support I would not have

made it this far. Unfortunately, now they want me to pay them back.





ii
















TABLE OF CONTENTS


Page

ACKNOWLEDGEMENTS ................................................ ii

LIST OF TABLES .................................................. vi

LIST OF FIGURES ........................................ ....... viii

ABSTRACT ........................................................ ix

CHAPTERS

I INTRODUCTION ............................................. 1

Review of Seafood Consumption in the United States ....... 1
Objectives ................................................ 9

II REVIEW OF RELATED WORK AND MODEL DEVELOPMENT ............ 12

Theoretical Considerations of Demand and Consumption ..... 12

Theoretical Considerations of Utility
Maximization ...................................... 12
Price and Income Considerations ..................... 14
Aggregation Considerations .......................... 17

Cross-Section Demand Analysis Considerations ............. 18

Price, Income, and Socioeconomic Variable
Considerations .................................... 18
Statistical Considerations .......................... 22

Related Seafood Consumption Studies ...................... 25
Conceptual Model Development ............................. 31

Econometric Models and Statistical Considerations ........ 39

Econometric Models .................................. 39
Statistical Considerations .......................... 43
Data Source and Considerations ...................... 49






iii










Page

III TOTAL SEAFOOD ANALYSIS ................................... 51

Descriptive Statistics Associated with Total
Seafood Analysis ....................................... 51
Regression Estimates of Total Seafood Analyses ........... 60

Region .............................................. 75
Urbanization ........................................ 78
Season .............................................. 80
Household Life Cycle ................................ 81
Race ................................................ 87
Household Receives Food Stamps ...................... 89
Household Caught Fish for Own Use ................... 90
Employment of the Meal Planner ...................... 90
Sex of the Meal Planner ............................. 90
Household Size ...................................... 91
Education of Meal Planner ........................... 95
Number of Guest Meals ............................... 96
Expenditures on Meal Consumed Away from Home ........ 97
Income before Taxes ................................. 100

Outlook for Increasing At-Home Demand for Seafood
and Implications ....................................... 107

IV SPECIFIC SEAFOOD PRODUCT FORM ANALYSIS ................... 111

Introduction ............................................. 111
Comparison of Consumption Parameters ..................... 114

Region .............................................. 114
Urbanization ........................................ 120
Season .............................................. 121
Household Life Cycle ................................ 121
Race ................................................ 123
Food Stamps ......................................... 124
Fish Caught for Own Use ............................. 124
Employment of the Meal Planner ...................... 124
Sex of the Meal Planner ............................. 125
Family Size ......................................... 125
Education of Meal Planner ........................... 129
Number of Guest Meals ............................... 129
Expenditures on Meals Away from Home ................ 130
Income before Taxes ................................. 130
Other Seafood Expenditures .......................... 135

Outlook for Increasing At-Home Demand for Specific
Product Forms and Implications ......................... 136






iv










Page

V SUMMARY AND IMPLICATIONS FOR FURTHER RESEARCH ............ 139

Summary .................................................. 139
Implications for Further Research ........................ 143

APPENDICES

DEFINITIONS OF SELECTED VARIABLES ........................ 148
DISAGGREGATED SEAFOOD STATISTICS ......................... 151

REFERENCES ...................................................... 226

BIOGRAPHICAL SKETCH ............................................. 230













































v
















LIST OF TABLES


Table Page

3-1 Descriptive statistics of variables in total
seafood models ........................................... 52

3-2 Summary statistics for Tobit analysis of weekly
household expenditures on seafood ........................ 61

3-3 Summary statistics for Tobit analysis of weekly
household consumption of seafood ......................... 67

3-4 Percentage distribution of household head by age
for selected years ....................................... 85

3-5 Estimated effects of changes in household size on
weekly expenditures and at-home seafood consumption ...... 92

3-6 Percent of meals eaten away from home, by type of
meal and selected household characteristics, spring
1965 and spring 1977 ..................................... 98

3-7 Estimated effects of changes in before tax income
on weekly expenditures and quantities of seafood
consumed ................................................. 102

3-8 Estimates of at-home seafood expenditure elastici-
ties with respect to income .............................. 104

3-9 Median family income in constant (1982) dollars for
selected years ........................................... 106

4-1 Descriptive statistics of data used in seafood
product form models ..................................... 112

4-2 Signs of estimated parameter associated with vari-
ables included in Tobit seafood expenditure models ....... 115

4-3 Estimated weekly expenditure elasticities with
respect to family size for specific seafood product
forms .................................................... 127

4-4 Estimated weekly expenditure, quantity and quality
elasticities with respect to before tax income for
specific and total seafood product forms ................. 132

vi










Table Page

B-1 Descriptive statistics of variables in fresh seafood
models ................................................... 151

B-2 Summary statistics for Tobit analysis of weekly
household expenditures on fresh seafood ................ 156

B-3 Summary statistics for Tobit analysis of weekly
household quantity consumption of fresh seafood .......... 161

B-4 Descriptive statistics of variables in frozen
seafood models ........................................... 166

B-5 Summary statistics for Tobit analysis of weekly
household expenditures on frozen seafood ................. 171

B-6 Summary statistics for Tobit analysis of weekly
household quantity consumption of frozen seafood ......... 176

B-7 Descriptive statistics of variables in canned
seafood models ........................................... 181

B-8 Summary statistics for Tobit analysis of weekly
household expenditures on canned seafood ................. 186

B-9 Summary statistics for Tobit analysis of weekly
household quantity consumption of canned seafood ......... 191

B-10 Descriptive statistics of variables in finfish
seafood models ........................................... 196

B-ll Summary statistics for Tobit analysis of weekly
household expenditures on finfish seafood ................ 201

B-12 Summary statistics for Tobit analysis of weekly
household quantity consumption of finfish seafood ........ 206

B-13 Descriptive statistics of variables in shellfish
seafood models ........................................... 211

B-14 Summary statistics for Tobit analysis of weekly
household expenditures on shellfish seafood .............. 216

B-15 Summary statistics for Tobit analysis of weekly
household quantity consumption of shellfish seafood ...... 221









vii
















LIST OF FIGURES


Figure Page

1-1 U.S. annual per capita consumption of commercial
fish and shellfish (edible weight), 1960-83 .............. 4

1-2 U.S. supply of edible fishery products (round
weight), 1960-83 ......................................... 7









































viii
















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


SOCIOECONOMIC DETERMINANTS OF AT-HOME SEAFOOD
CONSUMPTION: A LIMITED DEPENDENT VARIABLE
ANALYSIS OF EXISTING AND LATENT CONSUMERS

By

Walter R. Keithly, Jr.

August, 1985

Chairman: Frederick J. Prochaska
Major Department: Food and Resource Economics

Weekly household at-home seafood consumption in the United States

was analyzed using 1977-1978 Nationwide Food Consumption Survey data.

The cross-sectional consumption study related expenditure and quanti-

ties consumed of total seafood and five specific products (fresh,

frozen, canned, finfish, and shellfish) to a set of socioeconomic and

demographic factors which influence at-home seafood consumption

patterns.

A Tobit procedure was used in the estimation of the various

seafood product equations. The model, though used for statistical

reasons, provided considerable information which was used in examining

existing seafood consumers as well as potential seafood consumers.

The results of the analysis appear logical and useful. For the

most part, the estimated parameters were consistent with theoretical

expectations and/or results of previous studies. Region, urbanization,

race, household size, the stage of household growth and maturity,


ix










number of guest meals, money value of meals consumed away from home,

the household having caught fish, and income were all contributing

factors which helped to explain at-home seafood consumption patterns.

The estimated income elasticities associated with total seafood con-

sumption and consumption of all product categories were positive and

inelastic.

The analysis is distinguished from previous studies in two major

areas. First, the consumption effects were partitioned into those

for existing seafood consumers and those for potential consumers.

Second, consumption effects were separated into quantity and quality

components. These distinctions allow for a separate study of current

consumers and potential seafood consumers and for separation of

consumer expenditures into those for additional volume and those for

different qualities and associated marketing services.

The seafood industry and its support groups may wish to consider

the study when designing and implementing a long term promotion/

marketing program. Factors which determine at-home seafood consumption

are constantly changing. Consideration of these changes is given to

determine possible changes in at-home seafood consumption.



















x

















CHAPTER I
INTRODUCTION


Review of Seafood Consumption in the United States


Annual worldwide per capita consumption of fish and shellfish

averaged 27.1 pounds (live weight equivalent) during the 1975-77 period

(United States Department of Commerce, NOAA, NMFS, 1983). Japan, with

a per capita consumption of fish and shellfish equal to 148.6 pounds

during this period, was the world leader in terms of per capita

consumption. The United States, with a per capita consumption equal to

35.1 pounds, ranked 39th among all reported countries. Among developed

countries of the world, per capita consumption of fish and shellfish in

the United States is slightly less than two-thirds of the average.

However, per capita consumption of fish and shellfish in the United

States is approximately twice that of the undeveloped countries

(Wilson, 1982).

Though per capita consumption of fish and shellfish in the United

States trails that of several countries worldwide, total consumption of

fish and shellfish in the United States exceeds that of most other

countries due to the relatively large population in the United States

compared to other countries. After accounting for population, total

consumption of fish and shellfish in the United States is surpassed

only by Japan, the U.S.S.R., and China.





1







2


The reasons for the relatively low per capita consumption of fish

and shellfish in the U.S. compared to other developed countries of the

world are many and varied. First and foremost, the United States has

traditionally been the world's largest producer of beef and poultry

which has resulted in an abundant supply of these products at

relatively modest prices. The supply of edible fishery products, on

the other hand, has relied heavily upon imports to meet domestic

demand at acceptable prices. Second, in terms of ease of preparation,

fish and shellfish are typically rated poorly when compared to meat and

poultry products (Gillespie and Houston, 1975). This factor, in part,

has resulted in a large institutional and restaurant trade in seafood

products, while at-home consumption as a proportion of the total has

remained relatively low. Third, the demand for fish and shellfish in

the United States has been affected by the market distribution,

perishability, and preservation of these products (Christy and Scott,

1965). As one moves inland from those coastal states recognized as

major seafood producers, the availability of fresh seafood products

falls and the price increases. Finally, there remains a constant

concern among U.S. consumers regarding the quality of seafood being

sold in the various retail outlets. Meat and poultry products must be

inspected and certified by government representatives before sale

while inspection and certification of seafood products by government

representatives remains voluntary on the part of the seafood processor

(Becker, 1983). As such, inspection of seafood products has tradi-

tionally been sporadic and minimal. For example, the most intensive


Based on a regional study in Texas conducted by the authors.






3


Federal seafood inspection program which operates under the auspices

of the National Marine Fisheries Service inspected only about 20

percent of the 2.8 billion pounds of seafood consumed in the U.S. in

1982 (Becker, 1983). Though the average consumer probably does not

realize that seafood requires no federal inspection before sale to the

public, he/she is often reminded of some of the adverse health related

issues associated with consumption of certain seafood products. For

example, periodic newspaper headline scares such as those in the early

to mid 1970s regarding high mercury content in certain finfish species

and those related to occasional outbreaks of cholera resulting from the

consumption of contaminated raw oysters has left the consumer in a

quandry concerning the safety of eating these seafood products.

Though the preceding discussion points a bleak picture of the

future of the seafood industry in the United States, evidence to the

contrary suggests that consumption of seafood will be an increasingly

important component of the American household diet. For example, the

desire among American consumers to increase consumption of lower

calorie, natural, and more nutritious foods will likely translate to

increased seafood consumption (Slavin, 1984). Recent trends in the per

capita consumption of fish and shellfish also suggest that it will be

an increasingly important component of the American household diet.

As illustrated in Figure 1-1, per capita consumption of commercial

fish and shellfish has gradually been trending upwards over the past

two and one-half decades. During the 1960-64 period per capita

consumption of commercial fish and shellfish averaged 10.56 pounds

annually. By the 1979-83 period, annual per capita consumption had

increased 21 percent to 12.78 pounds (United States Department of











14.0 Total

13.0

12.0

11.0

10.0
Fresh and
9.0 Frozen

8.0

7.0

6.0 Canned

S 5.0
0
S4.0

3.0

2.0
Cured
1.0



1960 1965 1970 1975 1980 1983
Year

Figure 1-1. U.S. annual per capita consumption of commercial fish and shellfish (edible weight), 1960-83
SOURCE: United States Department of Commerce, NOAA, NMFS (1983).







5


Commerce, NOAA, NMFS, 1983). Though per capita consumption of poultry

has increased at a much faster rate than that observed for seafood, per

capita consumption of red meats has been declining since 1971

(Vondruska, 1984).

An increase in per capita consumption of fresh and frozen seafood

from 1960 through 1983 has accounted for most of the increase in total

per capita seafood consumption during this period (Figure 1-1).

Averaging 5.82 pounds during the 1960-64 period, per capita consumption

of fresh and frozen fish and shellfish increased 35 percent to 7.86

pounds during the 1979-83 period. By comparison, per capita

consumption of canned fish and shellfish which averaged 4.22 pounds

annually during the 1960-64 period increased only 9 percent to 4.6

pounds during the 1979-83 period. Per capita consumption of cured fish

and shellfish which represents a very small portion of the total per

capita consumption of fish and shellfish has actually declined in

recent decades. Wilson (1982) offers two reasons for the increase in

per capita consumption of fresh and frozen fish and shellfish relative

to that of canned and cured fish and shellfish. First, there has been

an increased availability of fresh and frozen seafood products in

retail outlets in recent years. Much of this increase has resulted

from the introduction and acceptability by the consumer of highly

processed fish and frozen seafood products such as sticks and portions

for which per capita consumption increased 130 percent between the

1960-64 period and the 1979-83 period. The second reason offered by

Wilson for the increased per capita consumption of fresh and frozen

fish and shellfish relates to the recent increase in the number of

restaurants in the United States, especially those specializing in







6


seafood preparation. In support of Wilson's contention, Van Dress

(1983) has estimated that the proportion of eating establishments

specializing in seafood preparation equalled 4.9 percent in 1979

compared to 2.1 percent in 1966.

Two features, already briefly alluded to, distinguish the seafood

industry in the United States from those industries associated with

most other major food products. First, the seafood industry is highly

dependent on an imported seafood product to satisfy domestic demand.

Second, consumption of seafood, as opposed to most other food products,

is highly related to the away-from-home food market.

The role of the international seafood market in meeting the

increasing per capita and total demand for seafood in the United States

can be observed with the aid of Figure 1-2. Between the 1960-64 period

and the 1979-83 period, total consumption of edible fishery products

increased by over 75 percent (Figure 1-2). This increase occurred in a

time period during which the population of the United States increased

only about 23 percent. As the consumption of edible fishery products

trended upwards, the composition of total supply shifted significantly

(Figure 1-2). Domestic supply of edible fishery products remained

relatively stable from 1960 through 1975 at approximately 2.5 billion

pounds annually (round weight). Associated with the passage of the

Magnuson Fisheries Conservation and Management Act of 1976, domestic

supply of edible fishery products in the United States experienced a

sizeable increase, and has averaged approximately 3.25 billion pounds

(round weight) annually since that time. Domestic supply appears to

have stabilized at this higher level in recent years. Overall, a 35.5

percent increase in the domestic supply of edible fishery products












9.0


8.0


7.0
7Total Supply


W 6.0
< Imports

5.0


4.0


o 3.0-

III ...... .. Domestic Landings
2.0


1.0




1960 1965 1970 1975 1980 1983
Year

Figure 1-2. U.S. supply of edible fishery products (round weight), 1960-83
SOURCE: United States Department of Commerce, NOAA, NMFS (various issues).







8


occurred between the 1960-64 and 1979-83 periods. While the domestic

supply of edible fishery products remained relatively stable through-

out the 1960s and early 1970s, considerable growth occurred in the

imports of edible fishery products. Imports of edible fishery products

exceeded that of domestic supply beginning in 1966 and have since con-

tinued to surpass domestic landings (Figure 1-2). Since the passage of

the Magnuson Fisheries Conservation and Management Act of 1976, imports

of edible fishery products have stabilized somewhat, averaging 4.75

billion pounds annually (round weight). Compared to a 35.5 percent

increase in the domestic supply of edible fishery products between the

1960-64 and 1979-83 periods, imports of edible fishery products have

increased by 131 percent, or almost four times that of the increase in

domestic supply. Imports of many of the higher valued seafood products

such as shrimp, scallops, and lobsters and imports of those fishery

products used in the preparation of processed frozen seafood items and

items served by the fast service eating establishments, such as fillets

and steaks frozen in blocks, have risen especially sharply in recent

decades. As the above discussion suggests, the growth in imported

edible fishery products is at least partially in response to a growing

domestic demand not met by domestic supply at acceptable prices.

Furthermore, given the apparent stability of domestic supply, further

increases in domestic demand will have to be met with concurrent

increases in imports.

The other feature that distingushes the seafood industry from

those industries associated with most other food products concerns

the extent of the away-from-home market in the sales of seafood

products. Though no precise data is available indicating the extent







9


of away-from-home versus at-home consumption of fishery products, it

is estimated that anywhere from one-third (Vondruska, 1985) to two-

thirds (Bockstael, 1984) of all seafood is consumed away from home. As

noted by Bockstael (1984) the demand for many species (e.g., crustacea,

fresh finfish) is highly sensitive to changes in income due to the

predominate restaurant trade in these species/products. Consequently,

in recessionary time periods when travelling, vacationing, and hence

restaurant trade are depressed, demand for these species/products will

also be depressed (Bockstael, 1984). During the 1973-74 recession and

the more recent economic slowdown starting in 1979, the depressed

demand for seafood resulted in a lowering of prices to the fishermen to

such an extent that the National Marine Fisheries Service was compelled

to initiate emergency programs such as the Catch America Program with

the intent of increasing consumer demand. Economic slowdowns and/or

recessions are a frequent occurrence in all or most developed

countries. Thus, the cyclical demand for seafood products in the

United States will likely be a recurring theme given the dependence of

seafood consumption on restaurant trade.


Objectives


Given the apparent relationship between seafood consumption and

restaurant trade, demand for seafood can be expected to rise and fall

in a cyclical manner in conjunction with oscillations in the general

economy. One means of alleviating the cyclical nature of the demand

for seafood is to encourage increased at-home consumption of seafood.

This can be done with increased at-home demand either at the expense of

the away-from-home market or independent of the away-from-home market.






10


If as in the first case, at-home consumption of seafood is increased at

the expense of the away-from-home market, little change in the total

demand for seafood may be experienced. If, on the other hand, at-home

demand for seafood is increased independent of the away-from-home

market, total demand for seafood by definition will be increased.

This could put pressure on many of the already heavily fished stocks

(as of 1974 about 62 percent of the economically important fisheries in

the United States were fully utilized or overfished (Eckert, 1979)) and

lead to an even greater demand for imported seafood.

In order to effectively increase at-home demand for seafood, a

thorough understanding of those factors hypothesized to determine

at-home consumption of seafood is required. Only then can the effects

of the at-home seafood market on the away-from-home market and

ultimately on the domestic fishing industry and on import demand for

seafood be fully comprehended. Thus, the overall objective of this

study is to examine and quantify those factors hypothesized to

determine at-home consumption of total seafood and specific product

forms (fresh, frozen, canned, shellfish, and finfish). In relation to

this objective, this study is designed to provide information concern-

ing seafood marketing implications based on empirical findings and

historical trends related to factors which determine seafood

consumption. This information is essential in evaluating fishery

market legislation, management alternatives, long-term trends, and

promotion and marketing programs.

The format of this study proceeds as follows. A review of

literature, the models to be estimated, and a discussion of statistical

considerations and data used in analyzing the models are provided in







11


Chapter II. The empirical results associated with weekly total seafood

consumption are analyzed in Chapter III. In Chapter IV, a discussion

of the results associated with the specific seafood product form models

is presented. In the last chapter of the main text, Chapter V, conclu-

sions of the study are presented as are suggestions for future research

in this area.

















CHAPTER II
THEORETICAL CONSIDERATIONS, REVIEW OF
RELATED WORK, AND MODEL DEVELOPMENT


Cross-sectional consumption studies have become increasingly

accepted and utilized during the past two decades in conjunction with

the increasing availability of appropriate data sets and the advent of

the high speed computer. With the increased volume of literature deal-

ing with the estimation of cross-sectional demand has come increased

sophistication in terms of model development and estimation techniques.

Regardless of the degree of sophistication employed in the analysis of

cross-sectional data, the fact remains that the theory of demand is the

basis for all model development in this area. Given this fact, the

present chapter begins with a review of the theory of demand. This

theory is then adapted to estimation of cross-sectional quantity

demanded and expenditure models. Following this section, a review of

related seafood consumption studies is presented. The chapter then

concludes with a discussion dealing with specification of the models

used in this study, the statistical procedures employed in the estima-

tion of these models, and the data used in the analysis.


Theoretical Considerations of Demand and Consumption


Theoretical Considerations of Utility Maximization


The theory of demand is developed on the postulate that consumers

maximize utility subject to a resource (income) constraint, i.e.,


12






13


max u = u (q) (2.1)

subject to p'q = m

where

u = utility derived from the consumption of q

q = an n-element column vector of quantities of
commodities available in the market place

p = an n-element column vector of market prices

m = consumer income


To maximize u(q) subject to the income constraint, the Lagrangian

expression


L(q,A) = u(q) + \(m p'q) (2.2)


is differentiated with respect to the arguments q and X. The

resultant derivatives are then set equal to zero. The following n + 1

first order conditions are the result of such a procedure:


au p = 0 (i = 1, 2, ., n) (2.3)
3qi

m p'q = 0


The first n equations of (2.3) satisfy the condition that in

equilibrium the consumer has equated the ratio of marginal utilities

derived from the consumption of any two commodities to the ratio of

their respective prices. Additionally, in equilibrium, the marginal

utility derived from the consumption of each commodity divided by its

respective price equals A, the marginal utility of income. The last

equation of (2.3) reinforces the condition that in equilibrium the

consumer has exhausted his total resources (income).







14


Since the solution to (2.3) depends only on prices, income, and

the utility function, (2.3) can be solved yielding n + 1 equations, one

for each of the q.'s and one for X in terms of p and m.


qi = q (p,m) (i = 1, 2, ..., n) (2.4a)
X = X(p,m) (2.4b)


The expressions in (2.4a) are referred to as demand functions, with the

demand for each of the n goods being expressed in terms of its own

price, prices of all substitutes, and income.


Price and Income Considerations


Economic theory suggests that demand functions satisfy certain

restrictions. One restriction is derived from the "fundamental equa-

tion of value" theory which decomposes the effect of a price change

into the substitution and income effects, i.e.,


= ( ) q (ij = 1, 2, .., n) (2.5)
-n)
3pi pi u=const J 3m
j j
A good is said to be normal if it has a downward sloping demand curve

or equivalently, a negative own price elasticity. Otherwise it is a

Giffen good which implies an upward sloping demand curve. The first

restriction which follows from the assumption of a strictly convex

indifference map implies that the own-price substitution effect is
1
negative, i.e.,

9q.
(9i )< 0 (2.6)
p. u=const


See Hicks (1957) for a proof of this and subsequently discussed
restrictions.







15


Hence, for a good to be a Giffen good, the income effect must not only

be negative but also outweigh the own-price substitution effect.

Furthermore, a good whose income effect is negative is referred to as

inferior and conversely, a good whose income effect is positive is

referred to as a superior good. An Engel curve which relates consump-

tion or expenditures on a good with income will thus be downward

sloping in the case of an inferior good and upward sloping in the case

of a superior good.

A second restriction offered by economic theory is that demand

functions (2.4a) are homogeneous of degree zero in all prices and

income. This suggests that an arbitrary scaling of all prices from

(p, m) to (ap, am) will have no effect on quantity demanded of any

good.

Unfortunately, these two restrictions have little to offer when

estimating the demand for a single good. Though the own-price

substitution effect is negative, the demand curve can be upward sloping

given that the good being analyzed is inferior and inferior to the

extent that the income effect outweighs the own-price substitution

effect. Thus, no restrictions can be imposed, a priori, on the sign of
2
the own-price coefficient. Furthermore, since a good can be either

inferior or superior, no restrictions can be imposed, a priori, on the

sign of the income coefficient. The second restriction is also of

little value in the estimation of single demand equations since rarely

if ever are all prices included in a single demand equation. Though

these restrictions offer little value in the estimation of single

2
This is of little concern in empirical demand analysis since few
if any goods have ever been shown to be Giffen goods.






16


demand equations, they do show the importance of including prices and

income in single demand equations.

Though the above restrictions offer little assistance in the

formulation of single equation demand models, they are of value in the

formulation of complete systems of demand equations. There are

additional restrictions offered by economic theory which are useful

in the formulation of complete systems of demand equations. First,

as illustrated by the budget constraint (equation 2.1), the sum of

expenditures on individual commodities must equal income in

equilibrium. A second restriction suggests that the sum of marginal

expenditures is unity, i.e.,


n 3q
E P m = 1 (i = 2, ..., n) (2.7)
i=l


This restriction guarantees that an increase in income is associated

with increased expenditures on at least one good. Third, the zero

homogeneity of the demand functions yields another n restrictions on

the slope coefficients of the following form:


aqi n Dq
m +- + P = 0 (i,j = 1, 2, ..., n) (2.8)
j=l J j


Finally, another in(n 1) restrictions can be obtained through

symmetry conditions


3q 3q 3q. 3q.
S+ q = + q (i,j = 1, 2, ... n) (2.9)
3p j m p. i 3m
j 1







17


Though these restrictions are vital in the formulation of complete

systems of demand equations, again they are of little value in the

formulation of single demand equations.


Aggregation Considerations


Though the demand functions developed earlier and specified in

equation (2.4a) provide the foundation for demand analyses, estimation

of the equations as specified in equation (2.4a) is generally difficult

because the equations refer to only a specific individual or household

and hence variation in the data is not present. Empirical demand

studies, on the other hand, are generally conducted by aggregating

consumption and income across all households for several different

time periods or, alternatively, treating the household as the main

consuming unit while estimating the demand equations for a cross-

section of households. The first approach, referred to as a time-

series analysis of demand, is used to study the effect of changes in

prices and income on consumption through time. With the second

approach, referred to as a cross-sectional demand analysis, differences

among household units are explicitly accounted for in the estimation

procedures. Thus, determining the effects of differences in socio-

economic factors and income across households on consumption patterns

becomes the primary objective of cross-sectional studies. Given the

cross-sectional nature of the present study, the discussion to follow

centers on the estimation of cross-sectional demand functions.







18


Cross-Sectional Demand Analysis Considerations


Price, Income, and Socioeconomic Variable Considerations


Two features of cross-sectional demand analysis of concern in

doing applied research in this field relate to proper model specifica-

tion and method of estimation. In terms of model specification,

concern has traditionally centered around variable selection. In

practice, researchers have estimated demand functions (Equation 2.4a)

when using cross-sectional data as follows:


iJ = qij(mj, Z) (i = 1, 2, ... n) (2.10)
(J = 1, 2, ..., k)

where

qiJ = quantity of the ith commodity demanded by the
Jth household

mj = income of Jth household

Zj = a column vector of socioeconomic characteristics
particular to the Jth household


The demand functions specified in (2.10) differ from those given by

(2.4a) in three aspects. First, a vector of socioeconomic characteris-

tics (Zj) whose values are specific to the Jth household has been

included in the latter equation. Second, the vector of prices (p) has

been excluded from the latter equation. Third, all variables have been

subscripted denoting the Jth household.

Given the specification of the cross-sectional demand functions

(2.10), estimation entails the use of a sample of households differing

in socioeconomic characteristics and income rather than that of a

representative household as presented in (2.4a). Differences in






19


characteristics of households, thought to influence tastes and prefer-

ences and thus expenditures and/or quantity demanded of commodity i,

are controlled for by way of the vector of socioeconomic character-

istics (Z ) particular to each household. Correct specification of the

vector Z continues to create controversy even though considerable

groundwork in this area was provided more than two decades ago by Prais

and Houthakker (1955) and Burk (1961). The problem with specifying

the vector Zj "correctly" is that it can potentially vary depending

upon the commodity being analyzed. For example, the socioeconomic

characteristics determining household demand for alcoholic beverages

may depend on a completely different set of cultural factors than

those factors determining household demand for milk. Thus, a review

of both economic literature and literature from disciplines specific

to the commodity being analyzed is essential for the correct model

specification. Prais and Houthakker and Burk discuss the importance

of examining factors such as family composition, social class,

religion, and demographic characteristics when analysing household

demand for food. These exploratory studies have subsequently been

extended and refined in an attempt to more closely model the actions of

the household. Two areas specific to the estimation of the household

demand for food frequently addressed by subsequent researchers are

those pertaining to measurement of household composition (e.g.,

Blokland, 1976; Muellbauer, 1974; Buse and Salathe, 1978; Murphy and

Staples, 1979) and the incorporation of the opportunity cost of time

of the homemaker (e.g., Mincer, 1963; Prochaska and Schrimper, 1973;

Redman, 1980).







20


With respect to household composition, the authors generally

attempted to standardize the household in order to show the impact that

different household members, varying in terms of sex and age, had on

the consumption behavior of the household. The one exception to this

rule is provided by Murphy and Staples, who attempted to explain

differences in consumption among different households by examining the

different stages of a household's growth and maturity.

With respect to the opportunity cost of time, it has long been

recognized that a commodity, when consumed at home, is not in the same

form as when purchased. Rather, value is added to the commodity after

it is purchased to transform it into some "new" commodity suitable for

consumption. Mincer (1963) cognizant of this fact demonstrated that

estimated income elasticities for a variety of commodities will tend to

be biased if the opportunity cost of time is omitted. Prochaska and

Schrimper (1973) analyzed away-from-home consumption with respect to

the opportunity cost of time of the meal preparer. Redman (1980)

analyzed the impact of women's time on away-from-home consumption and

for prepared meals based upon the concept of the opportunity cost of

time as presented by Gronau (1977). Though conceptually similar, the

studies of Prochaska and Schrimper and Redman differ somewhat in the

treatment of the opportunity cost of time. Prochaska and Schrimper

estimated a wage rate for the meal preparer based upon a set of

arguments (education and age) and then included the estimated wage rate

as an argument in the demand equation for away-from-home consumption.

Redman, on the other hand, introduced those arguments hypothesized to

affect women's opportunity cost of time (education of homemaker, age of






21


homemaker, age of children, employment status of homemaker) directly

into the demand functions for meals away from home and prepared food.

The rationale for excluding prices from cross-sectional demand

analyses (equation 2.10) is based on the concept that households

observe the same price for any given commodity at a point in time.

Though the validity of this assumption has been addressed (e.g.,

Mincer, 1963) and relaxed in applied research (e.g., Capps, 1982; Cox

et al., 1984), the exclusion of prices from cross-sectional demand

analyses remains the prevailing practice. Those who include prices in

a cross-sectional demand study do so on the premise that prices can be

expected to vary in some systematic manner across space and time (given

a sufficient period over which the data were collected). However, when

prices are included in the model, the interpretation of the estimated

price coefficients becomes exceedingly difficult due to price varia-

tions independent of shifts in supply. For example, price variations

can result from differences in average prices per unit due to quality

variation, price discounts associated with larger purchases, and/or

price variations resulting from added services provided with the

"basic" commodity. When price is excluded from the estimated model,

variables which relate to these price variations are included in the

model. However, often the parameters associated with these variables

represent a composition of two effects: a direct effect associated

with the specific variable being analyzed and a price effect. For

example, assume the average price paid per unit of a commodity, qi,

to be positively related with income, m. Hence, as household income

increases total expenditures on that commodity, equaling piqi, will

tend to increase proportionately more than the increase in q..







22


As discussed by George and King (1971), the increase in average price

per unit of the commodity associated with an increase in household

income can be viewed as a demand for quality or services. Furthermore,

an estimate of the quality elasticity for commodity qi can be defined

as the difference between the expenditure elasticity for that commodity

with respect to income and the quantity elasticity for that commodity

with respect to income (George and King, 1971). Given that other price

variations related to space, package size, etc., have been adequately

accounted for by inclusion of variables such as region, urbanization,

family size, etc., in the estimated equation, the quality elasticity

hence measures the percentage change in average price paid for a

commodity with respect to a percentage change in income3 and can be of

considerable interest "as a measure of consumers' desire for improved

quality or services, given the present average or standard quality"

(George and King, 1971, p. 72). Though the "quality/services" concept

as generally discussed is associated only with income, it can easily be

extended to reflect other economic variables which take continuous

values.


Statistical Considerations


The statistical considerations most often addressed with respect

to cross-sectional demand analyses concern the functional form of the

demand equations and the treatment of nonconsumers in the analyses.

Though considerable research has been conducted in an effort to find

the "ideal" functional form (e.g., Prais and Houthakker, 1955;


3For proof of the relationship see George and King (1971, p. 72).






23


Leser, 1963), no single functional form has emerged as clearly superior

under all conditions. There are, however, certain criteria that should

be considered when selecting and judging the appropriateness of a given

functional form (Brown and Deaton, 1972; Tomek, 1977; Hassan et al.,

1977). First, the functional form should allow for the possibility

that the commodity will not be consumed given an income below some

initial level. Second, the functional form should allow for a declin-

ing marginal propensity to consume with increased income. Third, the

functional form should allow for a satiety level which provides an

upper bound on quantity consumed. Finally, simplicity and convenience

of estimation need to be considered. Though these criteria are valid

when considering quantity consumption functions, they appear to be

overly restrictive when considering expenditure consumption functions.

Given the additional aspect of demand for quality and services, there

is no reason to necessarily expect the marginal propensity of expendi-

tures associated with some commodities to decline with increased

income. Similarly, a satiety level associated with expenditures on

some commodities is not necessarily expected, a priori.

As a commodity becomes more narrowly defined in an analysis, the

percentage of individual households not consuming that commodity

naturally increases. Therefore, a decision whether to include in the

analysis those households not consuming the commodity being analyzed

needs to be made. Currie et al. (1972) provide a good discussion

concerning under what conditions it is logical to exclude (include)

nonconsuming households from (in) the analysis. Basically, the issue

reduces to the following premise: if the consuming and nonconsuming

households can be considered as having identical behavioral patterns,






24



then there exists no rationale for excluding the latter group from the

analysis. The parameter estimates associated with such an analysis can

then be interpreted as reflecting the average of both consumers and

nonconsumers. In order to determine whether consumers and nonconsumers

represent a homogeneous group, the reasons why nonconsumers may be

present in a given sample are addressed by Currie et al.

The first reason provided by Currie et al. for observing non-

consumers in the sample proceeds as follows. Assume the distinction

between consumers and nonconsumers can be adequately summarized in

terms of some qualitative variable(s), say religion. In this case, two

options are available to the researcher. First, the researcher can

exclude the nonconsumers from the analysis, in which case the results

should be interpreted as exclusive of that religious segment of the

population. Alternatively, the researcher has the option of including

all households in the analysis by explicitly accounting for the

qualitative difference of religion.

As a second example, the authors consider the case in which non-

consumers can be explained as a result of a difference in the level of

some quantitative variable, say income, between them and the consuming

group. This being the case, there is no reason to expect behavioral

differences between the consuming and nonconsuming groups if incomes

were equal. With an increase in the level of income, nonconsumers

should enter the market and react in a similar manner to that of the

consuming group. Hence there is no rationale, from a theoretical

standpoint, for excluding the nonconsuming group from the analysis

under this condition.






25


As a final example, the authors consider the case in which non-

consumers are observed only because the time period represented by the

survey does not cover a sufficient span of time necessary to observe

consumption by most or all households. As in the previous case, there

exists no reason to expect nonconsumers to exhibit a different

behavioral pattern from that of consumers and hence there exists no

economic reason for excluding them from the analysis.

Summarizing these three cases, if the differences between con-

sumers and nonconsumers can adequately be accounted for, then there

exists no reason, from a methodological viewpoint, for excluding the

nonconsumers. Though there is no methodological rationale for exclud-

ing the nonconsumers, caution must be taken when including nonconsumers

in the analysis because of statistical problems. As discussed later,

use of ordinary least squares when the data includes a large percentage

of nonconsumers will generally be inappropriate because of resulting

biased and inconsistent estimates of the true parameters. An appro-

priate statistical technique to be used in conjunction with problems of

this nature will be presented towards the end of this chapter.


Related Seafood Consumption Studies


When compared to cross-sectional demand studies on those food

commodities which comprise a large percentage of the consumers food

budget, cross-sectional demand analyses for seafood products tend to be

somewhat limited. This probably reflects a lack of consideration of

"nontraditional" agricultural commodities in such surveys until recent

years.







26


Purcell and Raunikar (1968) provided one of the first comprehen-

sive demand studies for seafood and seafood products. Though their

results may be of limited use in providing an understanding of current

U.S. seafood consumption patterns due to the regional specificity and

age of the study (the data consisted of quarterly observations on 160

households in Atlanta, Georgia, over the 1958 through 1962 period),

the study does provide some valuable information. In the analysis,

the following set of arguments were used to explain expenditures on

seafood: race, household composition, annual household income,

seasonality, a trend variable, gifts, and price. The results indi-

cated that all variables with the exception of price and seasonality

were statistically significant in explaining expenditures on seafood.

Using data based on the 1972-74 BLS Consumer Expenditure Survey,

Salathe (1979), Capps (1982), and Perry (1981) each analyzed the

consumer demand for seafood and/or seafood products. Salathe expressed

expenditures on total seafood and two specific forms (canned fish and

fresh/frozen fish) as a function of household income and household

size. The analysis presented by Salathe may not be directly applicable

to this study for at least two reasons. First, arguments other than

household income and size undoubtedly influence expenditures on seafood

and seafood products. To the extent that the excluded variables are

correlated with household income and household size, the estimated

parameters associated with these two variables will tend to be biased.

The second reason that the results presented by Salathe may not be

directly applicable to this study relates to the econometric technique

employed by the author. In the analysis, Salathe included both con-

sumers and nonconsumers of seafood products and proceeded to estimate







27


the expenditure equations via ordinary least squares. Given the

relatively large proportion of nonconsuming households of seafood

products in the 1972-74 Consumer Expenditure Survey, the statistical

technique used by Salathe is probably inappropriate, which will also

lead to biased estimates of the true parameters. Given the possible

bias of the parameter estimates presented by Salathe, they must be

interpreted with caution and viewed as only a rough approximation of

the true underlying parameters. Salathe found the expenditure

elasticities with respect to income for aggregate seafood and its two

components to be highly inelastic, ranging from 0.21 to 0.38. The

estimated household size elasticities of expenditures on aggregate

seafood and its two components were somewhat less inelastic, ranging

from 0.36 to 0.57.

The analysis presented by Capps is more complete than that of

Salathe's in some areas. The strength of the model developed by Capps

lies in the specification of the seafood expenditure equation which was

expressed as a function of region, urbanization, race, marital status,

education, occupation, tenure class, employment status of the female

household head, season, household size, household income, and price.

The drawback of the model presented by Capps lies in the exclusion of

those households who reported no expenditures on seafood during the

survey period. The consequence of excluding the nonpurchasing house-

holds from the analysis is a loss of valuable information which could

potentially help to explain why some households purchased seafood

during the survey period while others did not. In addition, the

exclusion of nonpurchasers from the analysis indicates that the results

must be interpreted with respect to only those households purchasing







28


seafood products. Capps' study does provide considerable information

for assistance in determining which variables should be included in a

seafood consumption equation. Using a quadratic expenditure equation

Capps found that region, urbanization, race, martial status, household

size, household income, and price all contributed in a statistically

significant manner in explaining seafood expenditures. In agreement

with the results provided by Salathe, Capps found the income elasticity

of seafood expenditures to be extremely inelastic, equalling 0.1651.

With respect to family size, Capps found an elasticity of 0.2296 which

is somewhat less than that reported by Salathe.

Perry's analysis of seafood expenditures was by far the most

complete of those utilizing the 1972-74 BLS Consumer Expenditure

Survey. In addition to specifying a rather complete model describing

expenditures on total seafood and specific product forms (shellfish,

canned fish, whole fish, and filleted/steaks fish) in terms of

variables introduced into the equations, the analysis incorporated all

households in the survey. Furthermore, to avoid the likelihood of

biased estimates of the true parameters associated with using ordinary

least squares when a large concentration of zero observations for the

dependent variable is presented in the data, Perry estimated the

equations via a Tobit procedure. This procedure provides asymptoti-

cally consistent estimates of true parameters given a correct model

specification. The variables included in the various seafood/seafood

product expenditure equations were household income, race, urbaniza-

tion, expenditures on food consumed away from home, occupation of

household head, education level of household head, and household

composition. By estimating separate expenditure functions for the four







29


seafood/seafood products classifications by region and different income

groups, Perry in total estimated 85 equations explaining seafood/

seafood product expenditures. Though there are certain advantages to

estimating separate equations for different income groups, regions,

etc., the value of such a study in terms of answering national policy

questions becomes increasingly limited with increased refinements.

Given the relatively large number of equations estimated by Perry, a

discussion of the results necessitates generality. Those variables

found significant most often in explaining seafood expenditures were

income, race, and household composition. Other factors, most notably

urbanization and education, were important determinants of seafood

expenditures only in isolated instances.

The results presented by Perry were for the most part consistent

with those reported by other researchers. As in the studies conducted

by Salathe and Capps, Perry reported the income elasticities of

seafood/seafood products to be extremely inelastic. By region, the

income elasticities for total seafood expenditures ranged from a low of

0.069 in the South to a high of 0.204 in the Northeast. In terms of

specific seafood product forms, income elasticities were generally

statistically significant by region for shellfish, canned fish, and

filleted/steaks fish and insignificant for whole fish. The reported

income elasticities for shellfish expenditures were consistently higher

than that for the other individual products and ranged from a low of

0.069 in the South to a high of 0.344 in the Northeast.

In a recent study employing the 1977-78 Nationwide Food

Consumption Survey data, Haidacher et al. (1982) analyzed expenditures

and quantity demanded for total seafood, shellfish, and finfish.







30


Including all households in the analysis, the authors estimated the

various equations using ordinary least squares. As discussed in the

context of the previous studies, this estimation procedure may be

inappropriate when a large proportion of the households did not consume

the commodity. Given the large percentage of households not reporting

consumption of seafood products in the 1977-78 Nationwide Food Consump-

tion Survey, caution should be used in the interpretation of the

results reported by Haidacher et al. In the analysis, the authors

expressed expenditures and quantities consumed as functions of region,

race, urbanization, income, household size, household composition,

season, and the number of guest meals. Given the similarity between

the study by Haidacher et al. and those by Capps and Perry, one would

expect the results reported in the respective studies to be similar,

which in fact was the case. The income elasticity for total seafood

expenditures as reported by Haidacher et al. equalled 0.16 which was

the same as that reported by Capps and within the range of those

reported by Perry for the various regions. Income elasticities for

shellfish and finfish expenditures were given as 0.73 and 0.03,

respectively. These estimates are in agreement with those presented by

Perry to the extent that the income elasticity of shellfish expendi-

tures tended to be somewhat higher than that of finfish expenditures.

However, in absolute magnitude, the income elasticity of shellfish

expenditures reported by Haidacher et al. was two to three times the

size of that reported by Perry. Some of the discrepency in results

probably reflects the different statistical techniques employed in the

two studies.







31


In addition to the expenditure elasticities reported by Haidacher

et al., the authors also report the quantity elasticities with respect

to income. The quantity elasticities with respect to income were

positive for shellfish and negative for finfish and total seafood. The

negative quantity elasticity with respect to income for total seafood

in conjunction with a positive expenditure elasticity implies a posi-

tive quality elasticity for total seafood which equalled 0.20. This

consists of a high estimate of the quality elasticity associated with

shellfish consumption (0.59) and a relatively low estimate of the

quality elasticity associated with finfish consumption (0.10).

Summarizing the research to date, evidence suggests inelastic

income expenditure and quantity elasticities for total seafood and

specific product forms. However, none of the studies conducted to

date has made complete use of all data and/or available statistical

options. An extension of the work provided by the authors discussed in

this chapter is the basis of the next two chapters. The remainder of

this chapter lays the groundwork for the models to be estimated.


Conceptual Model Development


The first task associated with specifying a cross-sectional con-

sumption model involves that of defining the set of arguments compris-

ing the column vector of socioeconomic characteristics, Zj, given in

equation (2.10). The concept of consumer demand in conjunction with

the seafood expenditure/quantity consumption studies discussed in the

previous section were of assistance in meeting this objective. The

respective expenditure and quantity equations were specified as a







32


function of the following set of variables (household subscripts have

been deleted for notational convenience):


EXPi = f(Z1, Z2, ..., Z13, M, S) (2.11a)

Q. = f(Z1, Z2, ..., Z13, M, S) (2.11b)


The variables EXPi and Qi refer to the expenditures on and quantity

consumed of total seafood or specific product form by the Jth

household, respectively. The values these two variables take range

from zero upwards, with the frequency of observed zero values varying

with respect to the specific product form.

The region of the country (Northeast, North Central, South,

West), the degree of urbanization (Central City, Suburban, Nonmetro),

and the season during which the household was interviewed (Spring,
4
Summer, Fall, Winter) are defined as Zl, Z2, and Z3, respectively.

The rationale for including these variables in (2.11a and 2.11b) is

two-fold. First, the prices associated with total seafood and the

specific product forms vary by region (Zl), urbanization (Z2), and

season (Z3) due to differences in aggregate demand and supply. As

such, households in different regions and/or levels of urbanization or

interviewed in different seasons encounter different prices for the

same product. Second, consumption of total seafood and the specific

product forms is likely to differ among households by region, urbaniza-

tion, and season for reasons independent of a price effect, such as

tastes and preferences associated with cultural or institutional

factors.


See Appendix A for a description of these and the following
variables used in this study.







33


The remaining socioeconomic variable entered into equations

(2.11a) and (2.11b), Z4-Z13, M, and S were included to specifically

account for variations in seafood consumption among different house-

holds resulting from underlying socioeconomic differences which are

expected to influence tastes and preferences. The measurement of

household composition, Z4, used in this study is a modification of the

household life cycle classification proposed by Murphy and Staples

(1979) and is more directly applicable to the study than measurements

generally proposed for reasons discussed below. For purposes of this

study, households were stratified according to ten mutually exclusive

life cycle classifications: young single without children, young

married without children, young single with children, young married

with children, middle aged single without children, middle aged married

without children, middle aged single with children, middle aged married

with children, older single, and older married (where young is defined

as the head of household being less than 35 years old, middle aged is

defined as head of household being from 35 years old to 65 years old,

and elder is defined as head of household being equal to or greater

than 65 years old). There were two reasons for using the household

life cycle measurement of family composition in this study as opposed

to a more traditional measurement. First, it is useful to investigate

the reasons for the changes in apparent consumption of total seafood

and specific product forms over the past two decades. Available time

series data pertaining to household composition are related to the

family life cycle measurement of household composition more closely

than with the other measurements. Thus, the life cycle classification

was a preferable measurement of household composition in terms of






34


examining changes in seafood consumption through time. The second

reason for using the household life cycle measurement of family compo-

sition is based on the hypothesis that household consumption of total

seafood and specific product forms is more directly related to the life

cycle classification of the household than to other measurements of

family composition. For example, households with young children may

avoid the purchase and consumption of specific seafood product forms

that are known to have bones. While most measures of household

composition do not consider the case in which certain households are

unlikely to consume a given commodity as a result of certain charac-

teristics of the household members, the life cycle classification of

the household does to some extent account for this possibility by

stratifying households into mutually exclusive categories according to

given characteristics of the household.

The race of the household head (Z5), found in previous studies to

be of importance in explaining household consumption of seafood, was

included in the analysis to account for variations in tastes and

preferences among households of different races which would lead to

differences in at-home seafood consumption. For purposes of this study

the race of the household head was assigned to one of the categories:

White, Black, or "Other," where "Other" refers to any ethnic origin

other than that of White or Black.

Food stamps (Z6) in essence are an additional source of income to

households which can be used for the purchase of most food items.

Similarly, a household having caught fish for its own use (Z7) has a


"Other" represents an all-inclusive term referring to households
of various ethnic origins such as Asian, Indian, etc.







35


home produced good intended for consumption. As such, there should be

a positive relationship between the catching of fish and at-home con-
6
sumption of seafood.

Employment of the meal planner (Z8), the sex of the meal planner

(Z9), family size (Z10), and the education level of the meal planner

(Zll) are expected to affect the opportunity cost of the meal planner's

time. Following Gronau's (1977) premise, an increase in the education

level of the meal planner should result in an increase in the oppor-

tunity cost of his/her time, ceteris paribus. Similarly, meal planners

employed outside the home are expected to have a higher opportunity

cost of time than their counterparts. The planning of household meals

has traditionally been associated with the female members of the

household. Increases in the family size are expected to be associated

with increases in the opportunity cost of the meal planner's time,

ceteris paribus. Thus, an increase in family size is expected to

result in an increased consumption of the highly processed seafood

product forms such as canned seafood products relative to the non-

processed seafood products such as fresh seafood, ceteris paribus.

Though increases in the education level of the meal planner and

family size are expected to increase the opportunity cost of time of

the meal planner and hence result in a movement of household consump-

tion patterns towards heavily processed seafood products, results

supporting this contention are likely to be masked by offsetting

factors. For example, increases in the education level of the meal


A price representing the market price for a similar product in a
given region and season was assigned in those cases where the seafood
product was not purchased in the market.







36


planner are likely to be associated with an increased awareness of the

nutritional value associated with consumption of seafood products.

This increased nutritional awareness and resultant increased consump-

tion of seafood products are likely to offset the expected decline in

consumption associated with the increase in the opportunity cost of

time resulting from additional education. Similarly, though increases

in family size are expected to result in an increase in the opportunity

cost of time of the meal planner and hence a potential decline in

at-home consumption of seafood, increases in family size by definition

necessitates increased consumption in total. Hence the expected

decline in household consumption of seafood resulting from an increase

in the opportunity cost of time of the meal planner associated with an

increase in family size may be offset by increased consumption necessi-

tated by an increase in family size.

The number of guest meals served from home food supplies (Z12),

found to be a significant factor by Capps (1982) in explaining expendi-

ture on seafood consumed at home, was introduced into the analysis to

account for the expected increase in consumption of total seafood and

especially those seafood product forms most likely to be served when

entertaining guests. Since the less processed seafood product forms

are generally associated with a higher quality product and hence viewed

as more preferred items, increased weekly consumption of these product

forms is expected to be positively related with increases in the

number of guest meals.

Although Perry (1981) concluded that the money value of away-from-

home consumption was generally unimportant in explaining seafood

expenditures for at-home consumption, a similar variable was included







37


in the present analysis for two reasons. First, increases in the money

value of meals consumed away from home7 (Z13) may imply a lower need to

consume meals at home, ceteris paribus. The second reason for includ-

ing the money value of meals consumed away from home is that a large

proportion of total seafood consumption reportedly occurs in the away-

from-home food market. Furthermore, the away-from-home trade as a

percentage of the total varies substantially from one seafood product

to another. For example, consumption of certain shellfish species such

as shrimp and lobster occurs largely in institutional and restaurant

outlets, while the proportion of other seafood products, such as canned

tuna, consumed in the away-from-home market is considerably less

(Vondruska, 1985). In general, it is expected that those seafood

product forms difficult to prepare at home, such as fresh seafood, are

most often consumed in the away-from-home market. To the extent that

consumption of seafood products away-from-home substitutes for consump-

tion of similar products at home, the money value of meals consumed

away from home and at-home consumption of seafood are expected to be

negatively related.

The effect of income on consumption in general and on at-home

consumption of seafood in particular has been discussed extensively

throughout this chapter. In this study, before tax income (M) was

used as a proxy for the resources available to the household for the


Excludes the value of snacks purchased and consumed away from
home.







38


purchase of seafood products.8 No distinction was made with respect

to the sources of income and their separate effects on consumption of

total seafood and specific product forms. Though increases in house-

hold income have been found to be related to increases in expenditures

on total seafood consumed at home (e.g., Perry, 1981; Capps, 1982),

less evidence exists concerning the relationship between income and

at-home consumption of specific seafood product forms. For example,

the estimated income parameter associated with a specific seafood

product form may be either positive or negative depending upon whether

that product form is considered to be a normal or inferior good and may

in fact even vary among different segments of the population.

Substitutes for at-home consumption of seafood and the specific

product forms, denoted as S in equations (2.11a and 2.11b), follow from

the concept of the demand function provided in equation (2.4a). In

most cross-sectional studies of this nature, substitutes are omitted

from the analysis because prices of substitutes encountered by any

given household should be the same prices as those encountered by any

other household. In the estimation of the total seafood consumption

models no substitutes were specified. However, with respect to the

models for specific product forms, consumption of the alternative

seafood product forms were considered as appropriate substitutes.

For example, shellfish consumption by a given household was related to

consumption of finfish by that same household. Similarly, consumption

of fresh seafood was related to the consumption of the summation of

8
Though an arguement could be made to use total food expenditures
rather than income as an explanatory variable in the analysis, the
latter variable was used because it is more directly applicable for
answering policy oriented questions.







39


frozen and canned seafood. In a very strict sense, one might consider

that consumption of one seafood product is simultaneously related to

consumption of other seafood products by way of the budget constraint

(equation 2.1). However, given the very small proportion of the con-

sumer food dollar being allocated to seafood purchases, the problem of

simultaneity with respect to these variables is probably negligable.

In fact, the Longwood Research Group Limited (1984) concluded that

heavy users of one category of seafood tended to be heavy users of

other types of seafood. This is in contrast to what one would expect

to find if in fact the budget constraint played a major role in

determining substitutability among alternative seafood product forms.


Econometric Models and Statistical Considerations


Econometric Models


The conceptual models developed in the previous section (equations

2.11a and 2.11b) were fully specified to include the actual variables

included in the estimated relationships. Incorporating these changes

and making a similar change in the notation yielded the following

weekly expenditure and at-home quantity consumption equations:


EXPi = a0 + alX1 + a2X2 + + a33X33 + U1 (2.12a)

Qi = B0 + B X1 + B2X2 + + 33X33 + U2 (2.12b)

where

EXPi = weekly household expenditures on at-home consumption
of total seafood and specific product forms


Qi = weekly household at-home quantity consumed of
total seafood and specific product forms







40


X1-X3 = region of the country in which household resides

X1 = Northeastern region
X2 = North Central region
X3 = Southern region
Western region (base region)

X4-X5 = degree of urbanization in which household resides

X4 = central city
X5 = suburban (metro)
nonmetro (base urbanization)

X6-X8 = season during which household was interviewed

X6 = spring quarter
X7 = summer quarter
X8 = fall quarter
winter quarter (base season)

X9-X17 = household life cycle stage

X9 = young single adult without children
X10 = young married adults without children
XI1 = young single adult with children
X12 = young married adults with children
X13 = middle aged single adult without children
X14 = middle aged married adults without children
X15 = middle aged single adult with children
X16 = middle aged married adults with children
X17 = elderly single adult
elderly married adult (base life cycle stage)

X18-X19 = race of respondent

X18 = White
X19 = Other than White or Black
Black (base race)

X20 = household presently receiving food stamps (equals 1
if household is presently receiving food stamps,
O otherwise)

X21 = household caught fish for own use (equals 1 if household
caught fish for own use, 0 otherwise)

X22 = meal planner employed outside the home (equals 1 if meal
planner employed outside the home, 0 otherwise)

X23 = sex of meal planner (equals 1 if meal planner is female,
0 otherwise)







41


X24 = family size (total number living in household)

X25 = family size squared

X26 = number of years of schooling of meal planner

X27 = number of guest meals served from household food
supply in previous 7 days

X28 = dollar value of meals purchased and consumed away
from home (excluding snacks)

X29 = before tax income (thousand dollars)

X30 = before tax income squared

X31 = interaction between before tax income and race
(X18 *X29)

X32 = interaction between before tax income and family size
(X24 *X29)

X33 = expenditures on (or quantity consumed) of alternative
product forms

a a2, C3,
33 = estimated coefficients associated with weekly
expenditure equations

1, B2, a3,
33 = estimated coefficients associated with weekly
quantity equations

U1 = normally distributed random disturbance specific to
the expenditure equations

U2 = normally distributed random disturbance specific to
the quantity consumed equations


Though most of the independent variables included in equations

(2.12a) and (2.12b) enter in a binary manner (X1-X23), family size

(X24), education level of the meal planner (X26), guest meals (X27),

money value of meals consumed away from home (X28), before tax income

(X29), and expenditures (or quantities) of alternative seafood product

forms (X33) enter the equations in a continuous manner. Among this

latter group of variables, family size and income were specified in a







42


quadratic form. Income was specified in a quadratic form for three

reasons. First, the quadratic specification of the income variable

allows for a declining/increasing marginal propensity to consume/

purchase with increased income. Second, the quadratic specification

of the income variable allows for a satiety level providing an upper

bound on quantity consumed yet at the same time does not restrict

expenditures to behave in a similar manner. Third, the quadratic

specification of the income variable is easily modelled. Though the

quadratic specification of the income variable provides no assurance

that the commodity in question is not purchased/consumed given an

income below some threshold value, as will be demonstrated shortly,

the statistical technique used in the analysis does associate a lower

income with a lower probability of purchasing/consuming the commodity.

Family size was introduced into the analysis in a quadratic specifica-

tion to account for possible economies to scale in the purchasing and

consumption of seafood and specific product forms associated with

increased family size.

Two interactions (X31, X32) were introduced into the analysis.

The first interaction, that between White households and the linear

income term, allows for differences in the marginal propensity to

purchase/consume total seafood and specific product forms among house-

holds of different races. Similarly, the interaction between the

linear family size term and linear income term allows for differences

in the marginal propensity to purchase/consume among households of

different sizes. An argument could probably be made for the intro-

duction of other interaction terms, in addition to the two specified.

However, the use of too many interaction terms would have probably






43


resulted in severe collinearity problems among the regressors.

Therefore, those used were only those considered most appropriate.


Statistical Considerations


The model developed in the previous section can be expressed in

matrix form as follows:


Yt= XtB + Ut if XtB + Ut > 0 (2.13)

yt = 0 if Xt + Ut < 0

t = 1, 2, N

where,

yt = dependent variable

Xt = vector of independent variables

Ut = error term assumed iid N(O, 2 )


This model specification referred to as the Tobit model after its

founder is well known and used extensively in economic studies of a

cross-sectional nature (see Amemiya (1984) for a good review of the

basic model and its uses). Given the specification in (2.13), an

assumption is implicitly made that an underlying stochastic index equal

to XtB + Ut is observed only when strictly positive. In other words,

yt will only be positive given a value of Xt3 + Ut greater than zero.

Otherwise, yt will equal zero. For example, assume two households with

identical attributes with the exception of income. Furthermore, assume

that the household with the higher income consumed seafood while the

household with the lower income did not consume seafood. This would

imply that the first household with the higher income had exceeded that

threshold level required to consume seafood (i.e., XtB + Ut > 0), while






44


the second household with the lower income had not crossed that

threshold level (i.e., XtB + Ut < 0). Factors such as those in Xt

probably influence at-home consumption of seafood and thus the Tobit

procedure is appropriate for this analysis.

As shown by Greene (1981), OLS estimates of (2.13) are both biased

and inconsistent due to the non-normality of the expected error terms.

Thus, some estimated procedure other than that of OLS must be used if

unbiased or at least consistent parameter estimates are to be obtained.

Since the original work by Tobin (1958), several methods of estimating

equation (2.13) which assure consistent estimates of the true

parameters have been developed and used (see Amemiya (1984) for a

discussion of the different methods). Since the different methods are

widely known and should in all cases provide the same parameter esti-

mates assuming a unique maximum, the different approaches to estimating

(2.13) will not be discussed.

Though the uses and methods of estimation of the Tobit model are

well known and documented, less well known is the amount and types of

information that can be obtained from the Tobit estimates. The types

of information that the Tobit model provide are discussed here and

used extensively in the next two chapters. The uses of the Tobit

model, first presented by McDonald and Moffitt (1980), are the basis

for the ensuing discussion with some modifications to their work added

towards the end of the discussion. Those equations which will be

modified are assigned the letter (a) after the numerical numbering.

The modified equations are assigned the letter (b).

The unconditional expected value of the dependent variable in

equation (2.13) was shown by Tobin (1958) to equal






45


E(y) = XBF(Z) + of(Z) (2.14a)

where

Z = XB/o

f(Z) = unit normal density function

F(Z) = cumulative normal distribution functionl0


The unconditional expected value of the dependent variable represents

the expected value of the dependent variable associated with all

observations. Furthermore, as shown by Amemiya (1973), the conditional

expected value of the dependent variable for observations above the

limit (i.e., positive observations), y* is given by


E(y*) = E(yly > 0)

= XB + E(uly > 0) (2.15a)

= X + of(Z)/F(Z)


The relationship between the unconditional expected value of the

dependent variable (expected value associated with all observations)

and that of the conditional expected value of the dependent variable

(expected value associated with positive observations) is given as

follows:


E(y) = E(y*) F(Z) (2.16a)



9 1 -Z2/2
Defined as e
2r2
B'X
10 0 1 -Z2/2
10Defined as -- e
-00 27







46


Thus, the unconditional expected value of the dependent variable is

equal to the conditional expected value of the dependent variable

adjusted by the probability of observing a positive value of the

dependent variable, F(Z). Differentiating this relationship with

respect to an exogenous variable, Xi, gives the effect of a change in

the dependent variable resulting from a change in Xi.


3E(y)/3Xi = F(Z) (aE(y*)/3Xi) + E(y*) (DF(Z)/DXi) (2.17a)


Furthermore, it can be shown that the two partial derivatives on the

right-hand side of equation (2.17a) are equal to


aF(Z)/3Xi = f(Z)Bi/o (2.18a)

and

3E(y*)/=Xi = Bi + (o/F(Z)) af(Z)/MXi (2.19a)

(of(Z)/F(Z)2) aF(Z)/3Xi


which upon reduction yields


3E(y*)/3X. = 8.[1 Zf(Z)/F(Z) f(Z)2/F(Z)2] (2.20a)
1 1

Finally, note that after substitution of (2.20a) and (2.18a) into

equation (2.17a) and upon rearrangement of the terms, one arrives at

the following expression:


3E(y)/aXi = F(Z)3i (2.21a)


a much simplier expression than that of equation (2.13).

For purposes of this study, a modification of the above equations

was necessary. Notably, the expressions given above are based on the

assumption that all of the variables in the vector Xt are continuous






47


in nature. However, several of the variables evaluated in the present

study were binary in nature (X1-X23). The main consequence of such

a specification pertains to the evaluation of XB given a binary

variable. Specifically, the probability of observing a positive

value of the dependent variable now becomes conditional on the binary

variable being evaluated. Thus, the value of the standard normal,

(XI|Xi)/o, and hence the value of the cumulative normal density

function F(Z Xi) and the value of the unit normal density function

f(Z Xi) all become dependent on the binary variable being evaluated.

Hence, equations (2.14a) through (2.21a) need to be adjusted accord-
11
ingly when discussing binary variables. These adjustments are


E(ylXi) = (XBIXi) F(ZijXi) + of(ZilXi) (2.14b)
E(y*JXi) = (XaIXi) + of(ZiXi)/F(ZijXi) (2.15b)

E(ylXi) = E(y*IXi) F(ZilXi) (2.16b)

aE(ylXi)/aXi = F(ZiXi) ( IE(y*jXi)/aXi)

+ E(y'iXi) (aF(ZiIXi)3Xi) (2.17b)

aF(Zilxi)/aXi = f(Zi Xi)Bi/o (2.18b)

aE(y* Xi)/X = Bi + (o/F(ZilXi)) af(ZilXi)/3Xi

(of(ZiXi)/F(ZiIXi)2) 3F(Zixi)/9Xi (2.19b)

E(y* IXi)/3Xi = i[1 (Zi|Xi) f(ZilXi)/F(ZiIXi)
f(ZiXi)2/F(Zi Xi)2] (2.20b)

aE(y[Xi)/aXi = F(ZilXi)Bi (2.21b)


Technically, it would be preferable to use the concept of a
limit rather than a derivative in evaluating equations (2.17b) through
(2.21b) due to the discrete nature of Xi in each of the equations.






48


For evaluation of the continuous variables in the model, use of equa-

tions (2.14a) through (2.21a) is valid.

The total change in the dependent variable y given a change is Xi

as specified in equation (2.17a) can be broken into two components.

The first component [F(Z)(aE(y)/3.Xi] represents the change in the

expected value of the dependent variable if above the limit (positive)

weighted by the probability of being above the limit (positive). The

second component [E(y*)(aF(Z)/aXi)] represents the change in the prob-

ability of being above the limit (positive) weighted by the expected

value of the dependent variable if above. The breakdown of the total

change of the dependent variable into these two components is probably

very consistent with the actions of consumers in the market and is thus

useful in studying the consumption patterns of households.

It is interesting to note the simularity between the Tobit

estimates and OLS estimates. From equation (2.21a) it can be observed

that the effect of a change in Xi on the dependent variable y is equal

to Bi only when F(Z) is equal to one. This is in fact what one would

expect since as F(Z) approaches one, OLS estimates should be obtained.

Multiplying equation (2.17a) by Xi/E(y) gives the Tobit

elasticity, ni, which equals


EF(Z) X
i [ F(Z) Y + E(y*) F(Z E)) (2.22)
ax ax E(y)
1 1


Substituting equation (2.16a) for E(y) and making the appropriate

reductions provides the following specification of the elasticity of y

with respect to Xi:






49

ax ax
i E (y*) + i F(Z)
\i = ((y ) (F- ) (- ) (2.23)
i E(y*) ) X F(Z) ax ) (2.23)
1 1


The total elasticity is comprised of two components. The first compo-

nent measure the conditional elasticity associated with the nonlimit

observations. The second term measures the elasticity of the prob-

abilitiy of participation associated with a change in Xi.


Data Source and Considerations


The 1977-78 Nationwide Food Consumption Survey (NFCS) provides the

data used in the analysis. This is the most recent of the household

food consumption surveys periodically conducted under the auspices of

the United States Department of Agriculture. The survey encompassed

approximately 15,000 households throughout the 48 continguous states

and contained detailed information on characteristics for each

household. In addition, the survey contained detailed information

pertaining to expenditures on and the corresponding quantities of a

continum of foods consumed at home (measured at the level at which the

foods came into the kitchen) by each of the households surveyed. The

survey was conducted over a one year period (April 1977 through March

1978) and was stratified according to a variety of factors including

season, region, and urbanization in an attempt to have the sample

represent the universe of households in the continental United States

as accurately as possible. Though information on 14,930 households

was provided on the original NFCS data tapes distributed through the

Department of Commerce's National Technical Information Service, only

10,689 observations of the original 14,930 were retained for the

current analysis. Of the 4,241 deleted observations, approximately 92






50


percent were deleted due to the household's refusal to report annual

income. The remaining 8 percent were deleted due to the omission of

other relevant information. In light of the fact that households

reporting incomplete income information often provide poor or incom-

plete expenditure information (Buse, 1979), it was deemed appropriate

to delete these observations. While omission of those observations for

which information is missing could potentially lead to sampling bias

(Maddala, 1977), a comparison of those households not reporting income

with those reporting income indicates that this was not a serious

problem in the present study.12





























12
Among those households not reporting income, 50.2 percent of the
households consumed seafood at home during the one week interview
period compared to 50.8 percent among those households reporting
income. Average weekly consumption among those households consuming
seafood and not reporting income equalled 1.94 pounds compared to 1.91
pounds among those households reporting income.

















CHAPTER III
TOTAL SEAFOOD ANALYSIS


The discussion of the results associated with the total seafood

models is given in two sections. First, descriptive statistics

comparing/contrasting consumers and nonconsumers of seafood as

established by the survey data are presented and briefly discussed.

In the second section, the results of the Tobit analysis associated

with the total seafood models are presented and discussed.


Descriptive Statistics Associated with
Total Seafood Analysis


Descriptive statistics of the 10,689 households included in the

analysis are presented in Table 3-1. Though the information presented

in this table is of a descriptive nature without any attempt to

separate partial effects, the information does provide an overall

comparison of those households which consumed seafood during the one

week survey period versus those households which did not.

The statistics provided in Table 3-1 are assigned to one of four

categories. The first category, labelled total sample, gives the mean

values of all variables used in the analysis. For example, as

indicated in the table, 24.8 percent of the households in the analysis

resided in the Northeast region (XI). The second category, labelled

nonlimit observations, provides the mean values of all variables for

those households consuming seafood at home. For example, of those



51













Table 3-1. Descriptive statistics of variables in total seafood models


(1) (2) (3) (4)
Category Consumers & Consumers Nonconsumers Proportion
nonconsumers (nonlimit (limit of category
(total sample) observations) observations) consuming


----------------------- Mean (percent)a.................

Region:

Northeastern (XI) 0.248 0.299 0.195 0.612
North Central (X2) 0.241 0.213 0.270 0.449
Southern (X3) 0.342 0.305 0.379 0.453
Western (base) 0.169 0.183 0.156 0.550

Urbanization:

Central City (X4) 0.305 0.330 0.280 0.550
Suburban (X5) 0.355 0.374 0.335 0.535
Nonmetro (base) 0.340 0.296 0.385 0.442

Season:

Spring (X6) 0.238 0.233 0.243 0.497
Summer (X7) 0.232 0.234 0.229 0.512
Fall (X8) 0.268 0.259 0.278 0.491
Winter (base) 0.262 0.274 0.250 0.531













Table 3-1. Continued


(1) (2) (3) (4)
Category Consumers & Consumers Nonconsumers Proportion
nonconsumers (nonlimit (limit of category
(total sample) observations) observations) consuming


----------------------- Mean (percent)a ------

Household life cycle:

Young single w/o children (X9) 0.055 0.045 0.065 0.416
Young married w/o children (X10) 0.058 0.054 0.061 0.473
Young single with children (X11) 0.035 0.035 0.034 0.508
Young married with children (X12) 0.151 0.158 0.143 0.532
Middle aged single w/o children (X13) 0.075 0.063 0.088 0.427
Middle aged married w/o children (X14) 0.115 0.119 0.111 0.526
Middle aged single with children (X15) 0.056 0.062 0.050 0.562
Middle aged married with children (X16) 0.261 0.299 0.223 0.582
Elderly single (X17) 0.099 0.073 0.126 0.375
Elderly married (base) 0.095 0.092 0.099 0.492

Race of respondent:

White (X18) 0.852 0.831 0.874 0.495
Other (X19) 0.030 0.035 0.024 0.592
Black (base) 0.118 0.134 0.102 0.577


) ? '













Table 3-1. Continued


(1) (2) (3) (4)
Category Consumers & Consumers Nonconsumers Proportion
nonconsumers (nonlimit (limit of category
(total sample) observations) observations) consuming

a
----------------------- Mean (percent)a

Receives food stamps:

Yes (X20) 0.074 0.077 0.071 0.529
No (base) 0.926 0.923 0.929 0.506

Caught fish for own use:

Yes (X21) 0.234 0.264 0.202 0.573
No (base) 0.766 0.736 0.798 0.488

Employment of meal planner:

Yes (X22) 0.465 0.454 0.476 0.496
No (base) 0.535 0.546 0.524 0.518

Sex of meal planner:

Female (X23) 0.908 0.932 0.882 0.521
Male (base) 0.092 0.068 0.118 0.370














Table 3-1. Continued


(1) (2) (3) (4)
Category Consumers & Consumers Nonconsumers Proportion
nonconsumers (nonlimit (limit of category
(total sample) observations) observations) consuming


--------------------- Actual mean values----------------------

Family size:

Total number in household (X24) 2.946 3.153 2.733
Total number squared (X25) 11.457 12.803 10.067 ---

Education of meal planner:

Years (X26) 11.732 11.941 11.517

Guest meals:

Number of meals (X27) 1.141 1.256 1.021

Meals away from home:
Dollars (X28) 11.658 11.605 11.714

Income before taxes:

Thousand dollars (X29) 14.109 15.080 13.107
Thousand dollars squared (X30) 324.590 357.590 290.516




L') '














Table 3-1. Continued


(1) (2) (3) (4)
Category Consumers & Consumers Nonconsumers Proportion
nonconsumers (nonlimit (limit of category
(total sample) observations) observations) consuming


--------------------- Actual mean values----------------------

Interaction terms:

Income and race (X31) 12.745 13.394 12.075 ---
Income and family size (X32) 46.628 52.460 40.601 ---

Weekly expenditures and quantity:

EXP 1.487 2.927 0.000
Q 0.971 1.911 0.000

Number of households: 10,689 5,430 5,259

Percent of households: 100 50.8 49.2

a
The data provided in this table associated with the binary variables (X1-X23) should be interpreted as
representing proportions rather than percentages. To obtain percentages, multiply data by 100.







57 *


households consuming seafood at home, 29.9 percent resided in the /

Northeast region. The third category, labelled limit observations,

provides the mean value of all variables for those households reporting

a zero level of at-home consumption of seafood during the one week

survey period. On this basis, the information contained in Table 3-1

suggests that of those households reporting no at-home seafood consump-

tion 19.5 percent resided in the Northeast region. The final category

in Table 3-1, labelled proportion consuming, gives the value of the

proportion of households consuming seafood during the survey period

associated with each of the binary variables. For example, 61.2

percent of the households residing in the Northeast region of the

United States consumed seafood during the interview period.

Approximately one half (50.8 percent) of the 10,689 households

included in the analysis consumed seafood at home during the survey

period. Among households consuming seafood at home, average expendi-

tures and consumption of seafood equalled $2.93 and 1.91 pounds,

respectively. For the total sample, average at-home weekly consumption

of seafood was 0.97 pounds valued at $1.49 or approximately one-half

the volume and value estimated for consumers only. Placed on a yearly -

basis, at-home consumption of seafood by an average household thus

equals just over 50 pounds, or about 18 pounds per capita assuming an

Saverage of 2.75 members per household. This value equals about one-

half of the approximately 35 pounds (round weight) annual reported per

capita consumption of seafood in the United States during the 1975-77

period.

With respect to region, households in the Northeastern region of

the United States (XI) had a higher probability of consuming seafood at







58


home than did those households in either the Northcentral region (X2),

Southern region (X3), or Western region (base). Over 61 percent of the

Northeastern households consumed seafood at home compared with 44.9

percent, 45.3 percent, and 55 percent of the households residing in the

Northcentral, Southern, and Western regions, respectively.

Little variation in the proportion of households consuming seafood

at home was observed across seasons. Though a slightly larger propor-

tion of households consumed seafood at home during the summer (X7) and

winter (base) quarters than in either the spring (X6) or fall (X8)

quarters, the differences would probably not be significant if a

statistical test were to be conducted.

An examination of the different family life cycles (X9-X17)

revealed that households with children were more likely to consume

seafood at home than those households without children. On average

(weighted), 55.7 percent of the households with children consumed

seafood compared to 45.3 of those househoolds without children.

Similarly, at-home seafood consumption tended to increase in prob-

ability with increased family size (X24) which generally reflects '

increased number of children.

The race of the household respondent (X18, X19) appears to be an

important consideration in determining the probability of at-home

seafood consumption. White households (X18) had a significantly lower

observed probability of at-home seafood consumption than that of either

Black households (base) or households of other ethnic origins (X19).


It was assumed (though not verified) that elderly households
(X17, base) had no children living at home.







59


Households who caught fish for their own use (X21) had a much

higher probability of consuming seafood at home than did households not

reporting fish catches (base). Among those households who caught fish

for their own use, 57.3 percent reported consuming fish at home during

the interview period. This figure compares to only 48.8 percent among

those households who did not catch fish.

The employment status of the meal planner (X22) and the sex of

the meal planner (X23) had the anticipated effects on consumption of

seafood at home. Employment of the meal planner leads to a slight

decrease in probability of at-home seafood consumption. A female meal

planner (X23), on the other hand, appears to greatly enhance the

expectancy of that household consuming seafood at home.

With respect to the continuous variables included in the models

(X24, X26, X27, X28, X29), differences in the mean values between

consumers and nonconsumers for a given variable should indicate

increased (decreased) probability of at-home seafood consumption with

respect to that variable. For example, the average family size (X24)

of seafood consumers (3.153) greatly exceeds that of nonconsumers

(2.733). Similarly, the mean values for consumers with respect to

education (X26), number of guest meals (X27), and annual income before "
'i
taxes (X29) exceed the mean values for nonconsumers. Mean values for

continuous explanatory variables in the analysis are larger for con-

sumers than nonconsumers with the exception of expenditures on meals

away from home. Large differences between the mean values of the non-

consumers and consumers are especially apparent with respect to family

size (X24), number of guest meals (X27), and annual income before taxes

(X29). Thus, one would expect the probability of at-home seafood






60


consumption to increase significantly with changes in the value of

these variables.


Regression Estimates of Total Seafood Analyses


Notwithstanding the general usefulness of the descriptive statis-

tics just presented, there are several inherent weaknesses associated

with these types of statistics. The foremost weakness associated with

these types of statistics is that they do not control for confounded

effects among different variables. Thus, one cannot separate the

effect of one exogenous variable from that of another when examining

changes in the dependent variable. For example, the relatively low

probability of seafood consumption among households without children

when compared to those with children may be due to some factor such as

larger expenditures on meals consumed away from home among the former

group of households. The Tobit regression parameters presented in this

section can be considered as partial effects in that the confounded

effects among exogenous variables have been controlled for.

Results of the Tobit analysis relating to total at-home seafood

consumption are presented in Tables 3-2 and 3-3.2 The first column

gives a listing of the variables used in the analysis. The second

column in each table gives the Tobit parameter estimates associated

with each of the exogenous variables. The asymptotic t-values asso-

ciated with the parameter estimates are presented in the third column.

The relatively large sample size employed in this study should assure

that the asymptotic t-values are representative of the true values.


The Tobit model used for this analysis was developed by the Rand
Corporation and is referred to as LIMDEP. Documentation of the model
is given by Phelps (1972).












Table 3-2. Summary statistics for Tobit analysis of weekly household expenditures on total seafooda


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
ai F(Zi) 3E(EXP) 3E(EXP-) ,F(Zi)
axi 3Xi


(1) (2) (3) (4) (5) (6)

Region:

Northeastern (Xl). 0.8147 4.903 .5469 0.4456 0.1738
North Central (X2) -0.9383 -5.412 .4061 -0.3810 -0.1216
Southern (X3) -0.3578 -2.146 .4523 -0.1618 -0.0547
Western (base) -- --- .4812 --

Urbanization:

Central City (X4) 1.0843 7.507 .5198 0.5636 0.2104
Suburban (X5) 0.4888 3.628 .4717 0.2306 0.0802
Nonmetro (base) --- --- .4322 --- --

Season:

Spring (X6) 0.0154 0.101 .4771 0.0073 0.0026
Summer (X7) 0.1428 0.937 .4873 0.0696 0.0248
Fall (X8) -0.1414 -0.959 .4644 -0.0657 -0.0226
Winter (base) --- --- .4759












Table 3-2. Continued


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
ai F(Z.) ;E(EXP)b 9E(EXP-) .F(Zi)
aXi xi


(1) (2) (3) (4) (5) (6)

Household life cycle:

Young single w/o
children (X9) -0.6320 -1.803 .4605 -0.2910 -0.0996 0
Young married w/o
children (X10) -0.6892 -2.292 .4559 -0.3142 -0.1068
Young single with
children (X11) -1.2180 -3.235 .4137 -0.5039 -0.1604
Young married with
children (X12) -1.0171 -3.853 .4297 -0.4370 -0.1428
Middle aged single
w/o children (X13) -0.5456 -1.758 .4674 -0.2550 -0.0884
Middle aged married
w/o children (X14) 0.1697 0.698 .5252 0.0891 0.0336
Middle aged single
with children (X15) -0.0579 -0.188 .5069 -0.0293 -0.0108
Middle aged married
with children (X16) -0.3747 -1.437 .4812 -0.1567 -0.0637
Elderly single (X17) -0.5676 -1.976 .4656 -0.2643 -0.0911
Elderly married (base) --- --- .5116 ---












Table 3-2. Continued


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
ai F(Zi) E(EXP)b 3E(EXP-) F(Zi)
Mxi 3xi


(1) (2) (3) (4) (5) (6)

Race of respondent:

White (X18) -1.2718 -4.635 .4504 -0.8553 -0.2883
Other (X19) -0.3428 -0.988 .6051 -0.2074 -0.0883
Black (base) --- -- .6061 --- --

Receives food stamps:

Yes (X20) 0.2103 0.870 .4915 0.1034 0.0371
No (base) --- --- .4745 ---

Caught fish for own use:

Yes (X21) 1.3842 10.723 .5580 0.7724 0.3084
No (base) --- --- .4464 -- --

Employment of meal planner:

Yes (X22) -0.1207 -0.964 .4705 -0.0568 -0.0199
No (base) --- --- .4803 --











Table 3-2. Continued


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
a. F(Zi) 3E(EXP)b 3E(EXP) F(Zi)
1 i aXi


(1) (2) (3) (4) (5) (6)

Sex of meal planner:

Female (X23) 0.5679 2.425 .4790 0.2720 0.0958
Male (base) --- --- .4417 --- ---

Family size:

Total number in
household (X24) 0.4689 2.721 .4759 0.2062 0.0722
Total number
squared (X25) -0.0169 -1.036 ---

Education of meal planner:

Years (X26) 0.0858 3.982 .4759 0.0408 0.0143

Guest meals:

Number of meals (X27) 0.1649 7.831 .4759 0.0785 0.0275











Table 3-2. Continued


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
a F(Zi) EE(EXP)b E(EXP*) .F(Zi)
aXi 3Xi


(1) (2) (3) (4) (5) (6)

Meals away from home:
Dollars (X28) -0.0120 -3.598 .4759 -0.0057 -0.0020

Income before taxes:

Thousand dollars (X29) 0.0817 3.572 .4759 0.02518 0.0088
Thousand dollars
squared (X30) -0.0003 -2.619

Interaction terms:

Income and race (X31) -0.0416 -2.086 ---
Income and family
size (X32) 0.0047 1.577 ---












Table 3-2. Continued


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
ai F(Zi) aE(EXP)b 3E(EXP*) F(Zi)
3Xi 3Xi


(1) (2) (3) (4) (5) (6)

Other seafood:

Dollars (X33) NAc NA NA NA NA

Constant:

O0 -3.2808 -5.902 ---


The value of XP at the means of all Xi's is equal to -0.29854; a = 4.93351.

bThe effects of the interaction and/or squared terms have been accounted for in the construction of the
linear ai terms associated with those variables.

cNot applicable.











Table 3-3. Summary statistics for Tobit analysis of weekly household quantity consumption of total seafood


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
SF(Zi) aE(Q)b aE(Q*) F(Z.)
axi ax


(1) (2) (3) (4) (5) (6)

Region:

Northeastern (X1) 0.5888 4.382 .5010 0.2950 0.1073
North Central (X2) -0.5883 -5.197 .3842 -0.2260 -0.0688
Southern (X3) -0.0060 -0.044 .4414 -0.0026 -0.0009
Western (base) --- --- .4420 --

Urbanization:

Central City (X4) 0.7393 6.343 .4807 0.3553 0.1256
Suburban (X5) 0.4234 3.895 .4490 0.1901 0.0639
Nonmetro (base) --- --- .4071 ---

Season:

Spring (6) 0.1739 1.417 .4535 0.0789 0.0267
Summer (7) 0.2623 2.132 .4623 0.1213 0.0416
Fall (X8) 0.0215 0.181 .4383 -0.0094 0.0031
Winter (X9) --- .4361 --











Table 3-3. Continued


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
Bi F(Zi) E(, 8E(Q-) F(Zi)
ax1 3Xi


(1) (2) (3) (4) (5) (6)

Household life cycle:

Young single w/o
children (X9) -0.5569 -1.967 .4255 -0.2370 -0.0770
Young married w/o
children (X10) -0.5615 -2.312 .4251 -0.2387 -0.0774
Young single with
children (X11) -0.9970 -3.289 .3825 -0.3814 -0.1157
Young married with
children (X12) -0.7776 -3.653 .4038 -0.3140 -0.0988
Middle aged single
w/o children (X13) -0.4500 -1.797 .4630 -0.2084 -0.0862
Middle aged married
w/o children (X14) 0.1682 0.858 .4980 0.8038 0.0303
Middle aged single
with children (X15) -0.0317 -0.128 .4780 -0.0152 -0.0054
Middle aged married
with children (X16) -0.2881 -1.369 .4522 -0.1303 -0.0440
Elderly single (X17) -0.4718 -2.036 .4777 -0.2254
Elderly married (base) --- --- .4826 --- --












Table 3-3. Continued


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
Bi F(Zi) aE(Q)b aE(Q*) F(Zi)
3Xi 3Xi


(1) (2) (3) (4) (5) (6)

Race of respondent:

White (X18) -1.3990 -6.344 .4181 -0.7838 -0.2516
Other (X19) -0.7943 -2.839 .5263 -0.4180 -0.1579
Black (base) --- --- .6051 -

Receives food stamps:

Yes (X20) 0.2263 1.164 .4677 0.1058 0.0366
No (base) --- --- .4457 -

Caught fish for own use:

Yes (X21) 1.2961 12.451 .5426 0.7033 0.2724
No (base) --- --- .4129 ---

Employment of meal planner:

Yes (X22) -0.0252 -0.249 .4455 -0.0112 -0.0037
No (base) -- --- .4480 ---











Table 3-3. Continued


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
6i F(Zi) aE(Q)b aE(Q*) F(Zi)
3Xi Xi


(1) (2) (3) (4) (5) (6)

Sex of meal planner:

Female (X23) 0.4437 2.344 .4499 0.1996 0.0672
Male (base) --- --- .4060 --- ---

Family size:

Total number in
household (X24) 0.3759 2.710 .4469 0.1579 0.0530
Total number
squared (X25) -0.0117 -0.896 --- ---

Education of meal planner:

Years (X26) 0.0662 3.809 .4469 0.0269 0.0099

Guest meals:

Number of meals (X27) 0.1201 7.069 .4469 0.0037 0.0180











Table 3-3. Continued


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
.i F(Zi) E(Q)b 3E(Qa ) F(Zi)
axi axi



(1) (2) (3) (4) (5) (6)

Meals away from home:
Dollars (X28) -0.0131 -4.837 .4469 -0.0059 -0.0020

Income before taxes:

Thousand dollars (X29) 0.0487 2.637 .4469 0.0129 0.0043
Thousand dollars
squared (X30) -0.0002 -2.146 ---

Interaction terms:

Income and race (X31) -0.0295 -1.839 ---
Income and family
size (X32) 0.0034 1.411 --











Table 3-3. Continued


Expected total
Parameter Asymptotic Expected change resulting Expected change
Category estimates t-ratio probability from change in Xi among consuming units
Bi F(Zi) aE(Q)b E(Q*) .F(Zi)
3Xi 3Xi


(1) (2) (3) (4) (5) (6)

Other seafood:

Pounds (X33) NAc NA NA NA NA
-rj
Percent of households:

80 -2.5837 -5.772 ---


aThe value of Xa at the means of all Xi's is equal to -0.52941; a = 3.96192.

bThe effects of the interaction and/or squared terms have been accounted for in the construction of the
linear Bi terms associated with those variables.
Not applicable.
Not applicable.






73


In column 4 of each table, the vlue of the cumulative normal

distribution function associated with each variable is presented.

The value of this function varies with each of the discrete variables

due to variations in E(XB[Xi). For discrete variables in the models,

the values provided for the cumulative normal distribution function are

interpreted as the expected probability of observing a positive level

of the dependent variable given the occurrence of Xi, holding all non-

mutually exclusive variables at their mean levels. Mutually exclusive

variables are set equal to zero. For example, to determine the

expected probability of seafood consumption by average Northeastern

households, the value of X1 is set equal to one, the values for X2 and

X3 are set equal to zero, and the values for all remaining variables

are set at their respective means. For continuous variables, the value

provided for the cumulative normal distribution function represents the

probability of observing a positive level of the dependent variable

given the mean values for all exogenous variables. Of course, the

value of the cumulative normal distribution function varies with

changes in the level of exogenous variables.

Multiplication of the appropriate parameter estimates, given by

those values in column 2 by their respective expected probabilities of

occurrence (column 4) provides the unconditional or total expected

change in the dependent variable due to a change in Xi. These esti-

mates are given in column 5 of Tables 3-2 and 3-3. The unconditional

or total effect of a change in expenditures or quantity consumed with

respect to a change in the independent variable Xi can be decomposed

into two parts. The first part represents the change in the value of

expenditures or quantity consumed among existing consumers weighted by







74


the probability of being a consumer. The second part represents the

change in the probability of being a consumer weighted by the expected

value of expenditure or quantity consumed among consuming households.

The values for the first component, the conditional expected change in

expenditures or quantity consumed resulting from a change in Xi as

derived in equations (2.20a) and (2.20b), are provided in column 6.

Thus, by definition the values for the second component of the total

or unconditional change can be obtained by subtracting the values in

column 6 from those in column 5.

The parameter estimates associated with the at-home total seafood

expenditure and consumption equations appear satisfactory and reason- .

able based on several criteria. First, the estimates, with few

exceptions, conform to either theoretical expectations and/or results

of previous research. Second, the relative magnitudes of the estimate

probabilities of observing a positive level of seafood consumption

associated with each of the categories of binary variables are for the

most part in agreement with the observed probabilities given in column
/
5 of Table 3-1. Finally, a high proportion of the parameter estimates,

for a cross-sectional study of this nature, were statistically signifi-

cant (at a 10 percent significant level) in explaining weekly household

consumption of seafood at home. More detail is given to these factors

in the discussion of the individual explanatory variables.

Before providing an in depth discussion of the results, a few

general findings are discussed here. First, with few exceptions, the

results associated with the expenditure model were found to be con-

sistent with the results pertaining to the quantity consumed model.

Second, as indicated by the results of both models, the change among







75


consuming households resulting from a change in Xi consistently

averaged from 30-40 percent of the total change. This implies that

approximately 35 percent of the change in at-home consumption of

seafood with respect to a change in Xi is due to increased/decreased

consumption of seafood by those households currently consuming seafood

as opposed to entry or exit among households. Thus, approximately 65

percent of the total change is attributable to entry/exit into (from)

the at-home seafood market by households. This finding has significant)

implications to the seafood industry and its support groups who would

like to know the relative merits and associated costs of increasing

at-home seafood consumption by enticing new consumers into the

at-home seafood market as opposed to increasing consumption among
/
currently consuming households.


Region


Total weekly expenditures on seafood consumed at home were esti-

mated to be highest among the households residing in the Northeastern

region of the United States (XI) (Table 3-2) which is consistent with

the results presented by Perry (1981) and Capps (1982). Similarly,

total weekly quantity of seafood consumed by households residing in the

Northeastern region was estimated to be higher than that for other

regions (Table 3-3). Based on the information provided in column 5 of

Table 3-2, the total expected expenditures on seafood consumed at home

for a household in the Northeastern region were estimated to exceed

those for a household in the base region (West) by $0.45. Similarly,

expenditures among Northeastern households were estimated to exceed

expenditures among North Central households (X2) and Southern







76


households (X3) by $0.83 and $0.61, respectively. The information

provided in column 5 of Table 3-3 suggests the total expected weekly

consumption of seafood by households in the Northeastern region

exceeded that of households in the North Central region, Southern

region, and Western region by 0.52 pounds, 0.30 pounds, and 0.30

pounds, respectively, ceteris paribus.

The relatively high estimates associated with at-home seafood

expenditures for a household in the Northeastern region compared to

households in other regions of the United States are the result of two

factors. First, the estimated probabilities of a household having

positive weekly consumption expenditures and quantities, given in

column 4 of Tables 3-2 and 3-3, exceed those associated with any other

region. Second, the expected expenditures among consuming households

residing in the Northeast exceeded those of households in other

regions, as noted in the last column of Tables 3-2 and 3-3.

Given the estimated differences in at-home seafood expenditures

among households residing in the different regions, establishing

probable causes for these differences may be beneficial to the seafood

industry and its support groups. Traditionally, the Northeast region

has had an extensive fishing industry. This factor, in conjunction

with the close proximity of most of the Northeastern States to the

ocean, has resulted in a relatively steady supply of fresh seafood to

the households in this region. Transporting fresh seafood products

from coastal states to the inland states, such as many of those located

in the North Central region, is risky and expensive. O'Rourke (1977)

categorizes the U.S. seafood marketing system into two distinct

segments. The first segment, defined by O'Rourke (p. 239) as the






77


coastal fisheries ". .. are exploited by small, ill-equipped or part-

time fishermen, delivering their product to dockside warehouses for

fresh distribution to communities within a 50-mile radius." O'Rourke

further claims that the New England fisheries deliver much of their

specialty catches in this manner. Because of this factor ". .. large

areas of the continental U.S. have below average consumption of many

fish and shellfish species." The second segment which may account

for two-thirds of the U.S. consumption of seafood is comprised of

". .. large, highly capitalized canners or prepackers selling

standardized, breaded, and heavily promoted products through nationwide

retail chains or institutional outlets." The relative unavailability

and expense of certain fresher seafood products in the North Central

region probably explains to some extent the relatively low estimates

of at-home seafood expenditures and consumption in this region compared

to the other regions of the United States. Furthermore, one would

expect regional differences to be relatively large for fresh seafood

products and somewhat less for the more processed seafood products sold

either frozen or canned. The validity of this hypothesis is examined

in the next chapter when results pertaining to the seafood product

forms are analyzed.

Given the differences and probable causes for these differences as

relating to at-home consumption of seafood, what are their implications

to the seafood industry and its support group? Most obvious and

probably the most important from a policy standpoint lies in the

realization that the North Central region provides a relatively large

and untapped source by which to increase national at-home seafood

consumption. To this extent, it may be beneficial to the seafood






78


industry and the support groups to look for ways of reducing costs and

preserving the freshness of seafood when transporting fresh seafood to

the inland regions of the country. Unfamiliarity with many seafood

products among North Central households may also help to explain the

differences in the at-home seafood consumption patterns across regions.

Thus, promotion aimed at familiarizing these households with the dif-

ferent available products may prove useful in the short run. In the

long run, as the regional structure of the population shifts due to

increased population mobility and the transportation of fresh seafood

products becomes more economically feasible due to improved methods of

preserving the freshness of seafood, regional differences in seafood

consumption will probably decline naturally. For example, between 1950

and 1980, the proportion of the U.S. population living in the Northeast

and Midwest sections of the United States declined by 17 percent and 12

percent, respectively, while the proportion of the U.S. population

residing in the South and West increased by 7 percent and 44 percent,

respectively (United States Department of Commerce, Bureau of the

Census, 1984). Additional declines in the Northeast and Midwest

sections of the country are expected until at least the year 2000.

With these demographic shifts in population should come an exchange of

knowledge among households concerning different types of seafood and

methods of preparation which may eventually lead to the disappearance

of regional differences in at-home seafood consumption patterns.


Urbanization


Households in the central city (X4) were estimated to have higher

weekly consumption of seafood at home than those households in either







79


suburban areas (X5) or nonmetro areas (base), ceteris paribus.

Similarly, households in suburban areas were estimated to have higher

weekly at-home consumption of seafood than those households in nonmetro

areas, ceteris paribus. Central city households had total expected

weekly expenditures (quantity consumed) equal to $0.564 (0.355 pounds)

in excess of those households in nonmetro areas. Households in

suburban areas had total expected weekly expenditures (quantities)

equal to only $0.231 (0.190 pounds) greater than households in nonmetro

areas.

Since proximity to the coast was a major factor leading to the

development of many of the larger cities in the United States, house-

holds in the larger cities may have greater access to a larger variety

and quality of seafood than those households in either suburban or

nonmetro areas. For example, New York City, Boston, and New Orleans

have major fish markets acting as central locations from which distri-

bution of seafood products to other localities is coordinated. Due to

a decline in the accessibility of moderate cost quality seafood as one ,

moves away from the distribution centers, one would expect that the

probability of a household purchasing and hence consuming seafood to

decline in relation to distance from the distribution center. As noted

in column 4 of Tables 3-2 and 3-3, the estimated probabilities of

observing a positive level of expenditures and consumption of seafood

for a household residing in a central city area exceed those of a

household residing in a suburban or nonmetro area. These estimated

probabilities are consistent with the observed probabilities of

observing a consuming household given in column 4 of Table 3-1.







80


Season


At-home quantity of seafood consumed was estimated to be greatest

in the summer quarter (X7), followed by the spring quarter (X6), fall

quarter (X8), and winter quarter (base). Weekly expenditures, however,

were found to be much more constant across quarters with no statistical

differences noted for any season, as judged by the nonsignificance of

the asymptotic t-values associated with the parameter estimates (columni

3 of Table 3-2). A quantity change not reflected by a corresponding

expenditure change suggests that price must also be changing.

During the one year period in which the survey data used in this

analysis were collected (April 1977-March 1978), the price for seafood

consumed at home as measured by the Consumer Price Index increased by j

7.6 percent (United States Department of Agriculture, 1983). Given the

7.6 percent increase in the price of seafood consumed at home, the

question arises as to why the estimated weekly expenditures on seafood

consumed at home did not show a similar increase. In fact, weekly

expenditures were estimated to be lower (though only marginally) in the

later half of the survey year than in the first half. A possible

explanation put forward to answer this question is that households may

have reacted to the increase in the price of seafood consumed at home

by reducing quantities purchased leaving weekly household expenditures

on seafood unchanged. This would imply the quantity elasticity with

respect to price must equal approximately unity.







81


Household Life Cycle


The household life cycle category (X9-X17) appears to be very

useful in explaining weekly household at-home seafood consumption as

judged by the number of statistically significant parameter estimates.

Household life cycle estimates suggest that the composition of the

household, independent of size, explains both expenditures and quanti-

ties of seafood consumed at home in a rather systematic and logical

manner. Furthermore, the manner in which consumption of seafood can be

explained via the household life cycle is as expected, given the

current understanding of the at-home seafood market. For example,

there appears to be a general tendency for increased consumption of

seafood associated with the maturing of the household. Households with

the household head less than 35 years of age (X9-X12) consistently

consumed less seafood than households in more mature life cycle

categories. Households with the household head from 35 through

64 years of age (X13-X16) generally consumed less seafood than house-

holds comprised of an elderly married couple (base). Given that the

difference in household size between that of an elderly individual

(X17) and that of an elderly married couple (base) has been accounted

for by the variables representing household size (X24 and X25), the

estimated difference in at-home seafood consumption between these two

groups of households may represent differences in eating habits. For

example, elderly individuals may not wish to cook only for themselves,

especially those items requiring any amount of preparation time. This

would preclude them from consuming all but canned seafood which takes a

minimal amount of preparation before being suitable for consumption.






82


Overall, households comprised of young adults with children (X11-

X12) had the lowest at-home seafood consumption of any of the life

cycle categories, ceteris paribus. A representative household

comprised of a young single adult with children (X11) was estimated to

consume on average 0.144 fewer pounds of seafood at home than that of a

household comprised of a young single adult without children (X9) with

lower corresponding expenditures equal to about $0.21, ceteris paribus.

Analogously, a household comprised of a young married couple with

children (X12) was estimated to consume 0.0753 fewer pounds of seafood

with corresponding expenditures of about $0.12 less than that of a

household comprised of a young married couple without children, ceteris

paribus. Though these differences may appear small it should be kept

in mind that the analysis is based on a one week period. Extrapolating

to a one year period, a household comprised of a young single adult

with children is expected to consume almost 7.5 fewer pounds of seafood

at home than a young single adult without children, ceteris paribus.

With respect to the life cycle categories pertaining to middle

aged heads of households (X13-X16) the results do not appear to provide

any systematic trends. Households comprised of a single middle aged

adult without children (X13) had lower consumption and corresponding

expenditures than did those households comprised of a single middle

aged adult with children (X15). On the other hand, households con-

sisting of a middle aged couple without children (X14) tended to have

higher consumption of seafood than did those households consisting of

a middle aged couple with children.

The results pertaining to the household life cycle can be used

to design and implement a seafood promotion/marketing strategy.







83


For example, the relatively low estimates of at home seafood consump-

tion among those households categorized in the younger life cycle

categories establish the premise that seafood promotion/marketing

targeted towards this segment of the population may provide the seafood

industry and its support groups with greater net returns than that of

targeting households categorized in more mature stages of their life

cycle. Of course, the validity of this premise depends on the relative

costs associated with promoting seafood to households in the different

life cycles relative to the returns per dollar expended. However,

before targeting this group of households the reasons why this group of

households exhibits a relatively low level of seafood consumption needs

to be addressed. A couple of reasons can be offered to help explain

these results. First, the meal planner in the "younger" households

probably tends to be somewhat less expeieced attpreparing meals than

the meal planner in more mature households. Due to this factor, these

meal planners are more likely to avoid cooking a meal which involves

any much of preparation. Seafood, especially fresh seafood, has a

reputation for being difficult to properly prepare. Thus, seafood may

not be prepared and consumed as often by "younger" households as would

be expected among more mature households.

A second explanation is specific only to those younger households

with children (X11-X12). This segment of the population exhibited the

lowest weekly seafood consumption among the different categories in

the household life cycle. The children in this household group will

generally be of a lower average age than in households categorized in

the more mature life cycles. Hence, these children will place a

larger burden on the meal planner's time than would older children in






84


the more mature household life cycles. A related factor pertains to

the hesitency among parents of serving seafood to younger children out

of fear that the bones in the fish may injure the children or that the

children will have problems eating certain types of seafood (such as

shellfish). All of these reasons suggest a marketing strategy aimed at

promoting highly processed/ready-to-eat types of seafood products to

this segment of the population. Additionally, one would expect to find

fresh seafood consumption, which requires the most preparation time and

with which bones are most frequently associated, to be lower among

younger households, especially those with children, than among other

households. However, for frozen and canned seafood for which prepara-

tion time is minimal and bones are generally not a problem, one would

expect to observe little differences in consumption patterns among the

households categorized in the different life cycles. This hypothesis

will be examined in greater detail in the following chapter.

The characteristics and composition of the American household is

in a continual state of transition. Knowledge of this transition in

conjunction with the information provided by the results pertaining to

the household life cycle category can further aid the seafood industry

and its support groups in the planning stage of a long term marketing

strategy.

The first factor the seafood industry and its support groups

may wish to consider when planning a long term seafood marketing

strategy is the changing age structure of the American household.

Table 3-4 provides some statistics on the age distribution of

household heads for selected years. The statistics suggest that

younger households (those with household heads less than 35 years of


















Table 3-4. Percentage distribution of household head by age for selected years


Year
Age of house-
hold head
1960 1965 1970 1975 1980 1982


------------------------- Percent -------------------------

< 35 years 25.4 25.5 25.3 29.2 31.0 30.4

35-64 years 60.9 60.5 55.2 50.8 48.5 48.8

> 65 years 13.7 14.0 19.5 20.1 20.5 20.7



SOURCE: United States Department of Commerce, Bureau of the Census (various issues).







86


age) and older households (those with household heads greater than 65

years of age) have become increasingly important segments of the popu-

lation over the past two decades at the expense of the middle aged

households. The results of the Tobit analysis suggest, however, that

at-home seafood consumption among younger households (X9-X12) and

households comprised of an elderly individual (X17) tends to be among

the lowest of any household life cycle category. The seafood industry

and its support groups thus may want to consider these age structure

changes when planning a seafood marketing strategy.

A second factor the seafood industry and its support groups may

want to consider when planning a long term seafood marketing strategy

is the growing proportion of households in the United States that have

no children. In 1960, 43.1 percent of all households in the United

States had no children of their own under 18 years of age. By 1982,

this proportion had increased to 49.2 percent (United States

Department of Commerce, Bureau of the Census, 1984). The results of

the Tobit analysis suggest that, at least among certain age groups,

weekly at home seafood consumption differs depending upon whether

children are present in the household, certis paribus.

A final factor the seafood industry and its support groups may

wish to consider when planning a long term promotional strategy is the

growing proportion of single adult with children households in the

United States. In 1970, 84.9 percent of all children under 18 years

of age were living with both parents. By 1982, the proportion had

fallen to 75 percent (United States Department of Commerce, Bureau of

the Census, 1984). A single parent with children can be expected to

have more constraints on his/her time than would be the case if both







87


parents were present in the household. Thus, households in this group

should have a higher demand for seafood products for which little

preparation time is required relative to households in which both

parents are present.


Race


Black households (base) had significantly higher at-home seafood

expenditures and quantities consumed than did White households (X18).

Similarly, Black households were estimated to consume greater quanti-

ties of seafood than households of "Other" ethnic origins (X19) but

their weekly expenditures were not significantly different given the

insignificant t-value in column 3 of Table 3-2. Total weekly at-home

seafood consumption by a typical White household was estimated to be

0.78 pounds less than that of a similar Black household while expendi-

tures by that same White household were estimated to be $0.86 less than

that of a similar Black household. Among households of some "Other"

ethnic origin, weekly at-home consumption of seafood was estimated to

be 0.42 pounds less than that of a similar Black household while

expenditures by this group were only $0.21 less than that of a similar

Black household, certis paribus.

The estimated probability of consuming seafood at home was

substantially lower among White households than among either Black

households or households of "Other" ethnic origins. Among White house-

holds, the estimated probability of consuming seafood at home was 0.42,

compared to 0.61 among Black households, and 0.53 among households of

"Other" ethnic origins (column 4, Table 3-3). The large differences

associated with the probability of consumption among the different







88


races probably reflect cultural factors leading to differences in

tastes and preferences.

The proportion of Black households to that of White households has

been gradually trending upwards. In 1970, Black families represented

9.5 percent of all families in the United States. By 1982, the propor-

tion of Black families to that of the total had increased to 10.5

percent (United States Department of Commerce, Bureau of the Census,

various issues). This represents more than a 10 percent increase in

the proportion of Black families in just over a decade. Households of

Spanish origin, though representing a relatively small proportion of

the total number of households in the United States, also represent an,

increasingly important component of the population. Representing

approximately 3.9 percent of the total number of families in the United

States in 1970, the proportion of households in the United States of

Spanish origin has increased almost 40 percent in just over a decade

and represented 5.4 percent of the total number of families in 1982

(United States Department of Commerce, Bureau of the Census, various

issues). As the statistics tend to highlight, though the proportion of

Black households and those of "Other" ethnic origins still represents a

small proportion of the total number of households in the United

States, they do represent an increasingly important component of the

population. As such, their needs and wants in terms of seafood for

at-home consumption may want to be considered by the seafood industry

and its support groups when developing a long term marketing strategy.


Persons of Spanish origin may be of any race. i

'







89


Household Receives Food Stamps


Households receiving food stamps (X20) were estimated to have

total at-home consumption of seafood equal to 0.11 pounds valued at

$0.10 in excess of those households not receiving food stamps.

Caution, however, needs to be exercised when discussing the importance

of food stamps to at-home seafood consumption since receiving food

stamps did not have a statistically significant effect on at-home

consumption of seafood in either the expenditure model or the quantity

model.

Though not documented, it is a commonly held belief among seafood

dealers that food stamps are an important factor in a households's

decision to purchase and consume seafood. The results of the Tobit

analysis, however, tend to refute this idea. Since eligibility to

collect food stamps is related to household income, it is likely that

Black and elderly households are major recipients of food stamps. In

fact, about 30 percent of all food stamp recipients were Black in 1982

(United States Department of Commerce, Bureau of the Census, 1985).

As discussed earlier, Black households and households consisting of an

elderly couple tend to consume more seafood at home than that of the

"average" household. The combination of these two events may have

resulted in the association of food stamps with seafood purchases.

Furthermore, receiving food stamps may influence the time of month

that households collecting food stamps purchase seafood. If receiving

food stamps tends to result in these households purchasing seafood

during a narrow span of time, seafood dealers may incorrectly associate







90


this with an overall increase in seafood purchases among this group of

consumers.


Household Caught Fish for Own Use


As expected, households which caught fish (X21) consumed a greater

quantity of seafood than did their counterparts. Similarly, weekly

expenditures on seafood consumed at home were higher among households
4
who caught fish than among those who did not. As evidenced by the

estimated probabilities given in column 4 of Table 3-3, having caught

fish was second only to race among the discrete variable in determining

whether a household consumed seafood.


Employment of the Meal Planner


Employment of the meal planner (X22) was not statistically

significant in explaining weekly expenditures or quantities of seafood

consumed at home. The estimated signs of the respective coefficients,

however, were negative as expected given the increased opportunity of

the meal planner's time when employed. In the next chapter, considera-

tion will be given to the effect of employment by the meal planner on

consumption of those products which require the most preparation time.


Sex of the Meal Planner


The sex of the meal planner was significantly related to at-home

consumption of seafood, ceteris paribus. Households with female meal


If the fish consumed during the one week interview period
consisted of that which had been caught rather than purchased, the
price assigned to that product was based on the average retail price of
a comparable product in that region and season.




Full Text
103
the 1977-78 survey. The information in Table 3-7 provides two reasons
why the saturation level of at-home consumption of seafood is expected
to occur only at very high levels of income. First, the change in
consumption among participating households declines very slowly with
increases in income at least within the relevant range. Second, the
expected probability of consuming seafood increases with income, though
at a declining rate, throughout the range of income reported by most
households in the 1977-78 survey.
Following the specification of the Tobit elasticity, given in
equation (2.23), the elasticity of expenditures on seafood consumed at
home with respect to annual before tax income equals^
nEXP = 0.0953+0.1436 = 0.2389
Similarly, the quantity elasticity of seafood consumed at home with
respect to before tax income equals^
Nq = 0.0496 + 0.1318 = 0.1814
The estimate of the weekly at-home seafood expenditure elasticity with
respect to income, 0.2389, is well within the range of estimates given
in previous studies, the results of which are summarized in Table 3-8.
The current estimate of the at-home seafood expenditure elasticity thus
adds to the growing amount of research that indicates that at-home
consumption of seafood is very unresponsive to changes in income.
^Evaluated at the means of all variables.
^Evaluated at the means of all variables.


I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate, in scope and quality, as a dissertation for the degree of
Doctor of Philosophy.
W. 'Steven Otwell
Associate Professor of
and Human Nutrition
Food Science
This dissertation was submitted to the Graduate Faculty of the College
of Agriculture and to the Graduate School, and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy
August, 1985
Dean, Graduate School


104
Table 3-8. Estimates of at-home seafood expenditure elasticities
with respect to income3
Study
Expenditure
elasticity
Method of estimation
Capps (1982)
0.1651
0LS nonlimit observations
Salathe (1979
0.3568b
0LS all observations
0.2407
Perry (1981)
0.0609c
Tobit
0.2040
Haidacher et al. (1982)
0.16
0LS all observations
All studies with the exception of Haidacher et al. were based on the
1973-74 household consumption survey. The Haidacher et al. study was
based on the 1977-78 household consumption survey.
bThe first estimate given for Salathe's study was based on data from
June 1972 to June 1973 while the second estimate was based on data
from July 1973 to July 1974.
Q
The two estimates associated with Perry's study gives the range among
the different regions.


93
Expected seafood expenditures among consuming households
consistently increased though at a declining rate with each additional
household member (Table 3-5). This suggests that among consuming
households each additional household member resulted in a small decline
in both expenditures and quantities consumed for household members.
These declines may relate to price discounts associated with larger
purchases and less waste per household member with increases in
household size. j
The expected probability of observing a positive level of both
expenditures and quantity consumed increased with each additional
household member. The observed probabilities, presented in Table 3-1,
bear out the fact that households consuming seafood on a weekly basis
were in fact approximately 13 percent larger than those households not
consuming seafood. Though the expected probability of consuming
seafood increased with each additional household member, it did so at
a declining rate.
It is useful to address why the expected probability of consuming
seafood at home increased with household size, ceteris paribus. One
hypothesis put foreward to answer this question relates to the expected
increase in variety of foods consumed associated with increases in
household size. Assuming each member of any given household has an
individual preference function which contributes to a household
consumption function, increases in family size suggests an increased
probability that at least one member prefers seafood. This preference
will then translate to an increased probability that the household will
consume seafood.


Table B-10. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Mean (percent)3
Receives food stamps:
Yes (X20)
0.074
0.080
0.069
0.512
No (base)
0.926
0.920
0.931
0.471
Caught fish for own use:
Yes (X21)
0.234
0.274
0.207
0.535
No (base)
0.766
0.736
0.793
0.455
Employment of meal planner:
Yes (X22)
0.465
0.449
0.480
0.458
No (base)
0.535
0.551
0.520
0.488
Sex of meal planner:
Female (X23)
0.908
0.933
0.884
0.487
Male (base)
0.092
0.067
0.116
0.345
198


Table B-l. Continued
(1)
(2)
(3)
(A)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Mean (percent)a
Household life cycle:
Young single w/o children (X9)
0.055
0.036
0.059
0.109
Young married w/o children (X10)
0.058
0.053
0.059
0.152
Young single with children (Xll)
0.035
0.037
0.034
0.176
Young married with children (X12)
0.151
0.110
0.159
0.121
Middle aged single w/o children (X13)
0.075
0.080
0.074
0.178
Middle aged married w/o children (X14)
0.115
0.143
0.110
0.207
Middle aged single with children (X15)
0.056
0.064
0.055
0.190
Middle aged married with children (X16)
0.261
0.269
0.260
0.172
Elderly single (X17)
0.099
0.096
0.100
0.162
Elderly married (base)
0.095
0.112
0.090
0.196
Race of respondent:
White (X18)
0.852
0.713
0.880
0.139
Other (X19)
0.030
0.040
0.027
0.222
Black (base)
0.118
0.247
0.093
0.349
152


Table B-6. Continued
Category
Parameter
estimates
*1
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
8 E(Q)b
8X.
Expected change
among consuming units
8E(Q*) F(Z.)
8X
(1)
Income before taxes:
(2)
(3)
(4)
(5)
(6)
Thousand dollars (X29)
Thousand dollars
0.0623
2.560
.1379
0.0037
0.0007
squared (X30)
Interaction terms:
-0.0002
-1.591
Income and race (X31)
Income and family
-0.0344
-1.614



size (X32)
Other seafood:
0.0006
0.213
Pounds (X33)
Constant:
-0.0512
-2.343
.1379
-0.0071
-0.0013
e0
-6.8526
-11.419



aThe value of X8 at the means of all X^'s is equal to -3.94703; a = 3.61119.
kThe effects of the interaction and/or squared terms have been accounted for in the construction of the
associated linear terms.
180


Table B-9. Continued
Category
Parameter
estimates
*1
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X
8E(Q)b
3X .
i
i
Expected change
among consuming units
E(Q) F(Z.)
(1) (2) (3)
Race of respondent:
White (X18) 0.1082 1.435
Other (X19) 0.1500 1.672
Black (base)
Receives food stamps:
Yes (X20) -0.0304 -0.476
No (base)
Caught fish for own use:
Yes (X21) 0.0239 0.706
No (base)
Employment of meal planner:
-0.418
(A) (5) (6)
.3632 0.0307 0.0092
.3669 0.0550 0.0166
.3264
.3409 -0.0104 -0.0029
.3300
.3557 0.0082 0.0025
.3483
.3483 -0.0047
.3520
Yes (X22)
No (base)
-0.0134
-0.0013
193


Table Page
B-l Descriptive statistics of variables in fresh seafood
models 151
B-2 Summary statistics for Tobit analysis of weekly
household expenditures on fresh seafood 156
B-3 Summary statistics for Tobit analysis of weekly
household quantity consumption of fresh seafood 161
B-4 Descriptive statistics of variables in frozen
seafood models 166
B-5 Summary statistics for Tobit analysis of weekly
household expenditures on frozen seafood 171
B-6 Summary statistics for Tobit analysis of weekly
household quantity consumption of frozen seafood 176
B-7 Descriptive statistics of variables in canned
seafood models 181
B-8 Summary statistics for Tobit analysis of weekly
household expenditures on canned seafood 186
B-9 Summary statistics for Tobit analysis of weekly
household quantity consumption of canned seafood 191
B-10 Descriptive statistics of variables in finfish
seafood models 196
B-ll Summary statistics for Tobit analysis of weekly
household expenditures on finfish seafood 201
B-12 Summary statistics for Tobit analysis of weekly
household quantity consumption of finfish seafood 206
B-13 Descriptive statistics of variables in shellfish
seafood models 211
B-14 Summary statistics for Tobit analysis of weekly
household expenditures on shellfish seafood 216
B-15 Summary statistics for Tobit analysis of weekly
household quantity consumption of shellfish seafood 221
vii



PAGE 1

SOCIOECONOMIC DETERMINANTS OF AT-HOME SEAFOOD CONSUMPTION: A LIMITED DEPENDENT VARIABLE ANALYSIS OF EXISTING AND LATENT CONSUMERS BY WALTER R. KEITHLY JR. A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1985

PAGE 2

ACKNOWLEDGEMENTS Numerous people have helped in making this study and my graduate program possible. Sincerest appreciation is extended to my chairman, Fred J. Prochaska. He was always there to give me encouragement and advice even during strained times. He should be credited with my success but not failure in the academic field. Dr. Scott Shonkwiler gave freely of his time in helping me with questions of a statistical nature. Similarly, Drs. Cato, Kilmer, and Otwell devoted their time to assure a better quality product than would otherwise have been the case. Hopefully, their constructive criticisms of this study make it more useful for the groups for which it is intended I wish also to thank the Food and Resource Economics Department at the University of Florida and Florida Sea Grant for giving me the opportunity to pursue a graduate program and for providing financial support I wish to thank Janet Eldred for the typing she did on this manuscript. It sometimes got complicated sending everything through the mail Finally, I wish to thank my parents for giving me the opportunity to pursue a graduate career. Without their support I would not have made it this far. Unfortunately, now they want me to pay them back. i i

PAGE 3

TABLE OF CONTENTS Page ACKNOWLEDGEMENTS ii LIST OF TABLES vi LIST OF FIGURES viii ABSTRACT ix CHAPTERS I INTRODUCTION 1 Review of Seafood Consumption in the United States 1 Objectives 9 II REVIEW OF RELATED WORK AND MODEL DEVELOPMENT 12 Theoretical Considerations of Demand and Consumption 12 Theoretical Considerations of Utility Maximization 12 Price and Income Considerations 14 Aggregation Considerations 17 Cross-Section Demand Analysis Considerations 18 Price, Income, and Socioeconomic Variable Considerations 18 Statistical Considerations 22 Related Seafood Consumption Studies 25 Conceptual Model Development 31 Econometric Models and Statistical Considerations 39 Econometric Models 39 Statistical Considerations A3 Data Source and Considerations 49 iii

PAGE 4

Page III TOTAL SEAFOOD ANALYSIS 51 Descriptive Statistics Associated with Total Seafood Analysis 51 Regression Estimates of Total Seafood Analyses 60 Region 75 Urbanization 78 Season 80 Household Life Cycle 81 Race 87 Household Receives Food Stamps 89 Household Caught Fish for Own Use 90 Employment of the Meal Planner 90 Sex of the Meal Planner 90 Household Size 91 Education of Meal Planner 95 Number of Guest Meals 96 Expenditures on Meal Consumed Away from Home 97 Income before Taxes 100 Outlook for Increasing At-Home Demand for Seafood and Implications 107 IV SPECIFIC SEAFOOD PRODUCT FORM ANALYSIS Ill Introduction Ill Comparison of Consumption Parameters 114 Region 114 Urbanization 120 Season 121 Household Life Cycle 121 Race 123 Food Stamps 124 Fish Caught for Own Use 124 Employment of the Meal Planner 124 Sex of the Meal Planner 125 Family Size 125 Education of Meal Planner 129 Number of Guest Meals 129 Expenditures on Meals Away from Home 130 Income before Taxes 130 Other Seafood Expenditures 135 Outlook for Increasing At-Home Demand for Specific Product Forms and Implications 136 iv

PAGE 5

Page V SUMMARY AND IMPLICATIONS FOR FURTHER RESEARCH 139 Summary 139 Implications for Further Research 143 APPENDICES DEFINITIONS OF SELECTED VARIABLES 148 DISAGGREGATED SEAFOOD STATISTICS 151 REFERENCES 226 BIOGRAPHICAL SKETCH 230 v

PAGE 6

LIST OF TABLES Table Page 3-1 Descriptive statistics of variables in total seafood models 52 3-2 Summary statistics for Tobit analysis of weekly household expenditures on seafood 61 3-3 Summary statistics for Tobit analysis of weekly household consumption of seafood 67 3-4 Percentage distribution of household head by age for selected years 85 3-5 Estimated effects of changes in household size on weekly expenditures and at-home seafood consumption 92 3-6 Percent of meals eaten away from home, by type of meal and selected household characteristics, spring 1965 and spring 1977 98 3-7 Estimated effects of changes in before tax income on weekly expenditures and quantities of seafood consumed 102 3-8 Estimates of at-home seafood expenditure elasticities with respect to income 104 39 Median family income in constant (1982) dollars for selected years 106 41 Descriptive statistics of data used in seafood product form models 112 4-2 Signs of estimated parameter associated with variables included in Tobit seafood expenditure models 115 4-3 Estimated weekly expenditure elasticities with respect to family size for specific seafood product forms 127 4-4 Estimated weekly expenditure, quantity and quality elasticities with respect to before tax income for specific and total seafood product forms 132 vi

PAGE 7

Table Page B-l Descriptive statistics of variables in fresh seafood models 151 B-2 Summary statistics for Tobit analysis of weekly household expenditures on fresh seafood 156 B-3 Summary statistics for Tobit analysis of weekly household quantity consumption of fresh seafood 161 B-4 Descriptive statistics of variables in frozen seafood models 166 B-5 Summary statistics for Tobit analysis of weekly household expenditures on frozen seafood 171 B-6 Summary statistics for Tobit analysis of weekly household quantity consumption of frozen seafood 176 B-7 Descriptive statistics of variables in canned seafood models 181 B-8 Summary statistics for Tobit analysis of weekly household expenditures on canned seafood 186 B-9 Summary statistics for Tobit analysis of weekly household quantity consumption of canned seafood 191 B-10 Descriptive statistics of variables in finfish seafood models 196 B-ll Summary statistics for Tobit analysis of weekly household expenditures on finfish seafood 201 B-12 Summary statistics for Tobit analysis of weekly household quantity consumption of finfish seafood 206 B-13 Descriptive statistics of variables in shellfish seafood models 211 B-14 Summary statistics for Tobit analysis of weekly household expenditures on shellfish seafood 216 B-15 Summary statistics for Tobit analysis of weekly household quantity consumption of shellfish seafood 221 vii

PAGE 8

LIST OF FIGURES Figure Page 1-1 U.S. annual per capita consumption of commercial fish and shellfish (edible weight), 1960-83 4 1-2 U.S. supply of edible fishery products (round weight), 1960-83 7 viii

PAGE 9

Abstract of Dissertation Presented to the Graduate School of of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SOCIOECONOMIC DETERMINANTS OF AT-HOME SEAFOOD CONSUMPTION: A LIMITED DEPENDENT VARIABLE ANALYSIS OF EXISTING AND LATENT CONSUMERS By Walter R. Keithly, Jr. August, 1985 Chairman: Frederick J. Prochaska Major Department: Food and Resource Economics Weekly household at-home seafood consumption in the United States was analyzed using 1977-1978 Nationwide Food Consumption Survey data. The cross-sectional consumption study related expenditure and quantities consumed of total seafood and five specific products (fresh, frozen, canned, finfish, and shellfish) to a set of socioeconomic and demographic factors which influence at-home seafood consumption patterns. A Tobit procedure was used in the estimation of the various seafood product equations. The model, though used for statistical reasons, provided considerable information which was used in examining existing seafood consumers as well as potential seafood consumers. The results of the analysis appear logical and useful. For the most part, the estimated parameters were consistent with theoretical expectations and/or results of previous studies. Region, urbanization, race, household size, the stage of household growth and maturity, ix

PAGE 10

number of guest meals, money value of meals consumed away from home, the household having caught fish, and income were all contributing factors which helped to explain at-home seafood consumption patterns. The estimated income elasticities associated with total seafood consumption and consumption of all product categories were positive and inelastic The analysis is distinguished from previous studies in two major areas. First, the consumption effects were partitioned into those for existing seafood consumers and those for potential consumers. Second, consumption effects were separated into quantity and quality components. These distinctions allow for a separate study of current consumers and potential seafood consumers and for separation of consumer expenditures into those for additional volume and those for different qualities and associated marketing services. The seafood industry and its support groups may wish to consider the study when designing and implementing a long term promotion/ marketing program. Factors which determine at-home seafood consumption are constantly changing. Consideration of these changes is given to determine possible changes in at-home seafood consumption. x

PAGE 11

CHAPTER I INTRODUCTION Review of Seafood Consumption in the United States Annual worldwide per capita consumption of fish and shellfish averaged 27.1 pounds (live weight equivalent) during the 1975-77 period (United States Department of Commerce, NOAA, NMFS, 1983). Japan, with a per capita consumption of fish and shellfish equal to 148.6 pounds during this period, was the world leader in terms of per capita consumption. The United States, with a per capita consumption equal to 35.1 pounds, ranked 39th among all reported countries. Among developed countries of the world, per capita consumption of fish and shellfish in the United States is slightly less than two-thirds of the average. However, per capita consumption of fish and shellfish in the United States is approximately twice that of the undeveloped countries (Wilson, 1982). Though per capita consumption of fish and shellfish in the United States trails that of several countries worldwide, total consumption of fish and shellfish in the United States exceeds that of most other countries due to the relatively large population in the United States compared to other countries. After accounting for population, total consumption of fish and shellfish in the United States is surpassed only by Japan, the U.S.S.R., and China. 1

PAGE 12

The reasons for the relatively low per capita consumption of fish and shellfish in the U.S. compared to other developed countries of the world are many and varied. First and foremost, the United States has traditionally been the world's largest producer of beef and poultry which has resulted in an abundant supply of these products at relatively modest prices. The supply of edible fishery products, on the other hand, has relied heavily upon imports to meet domestic demand at acceptable prices. Second, in terms of ease of preparation, fish and shellfish are typically rated poorly when compared to meat and poultry products (Gillespie and Houston, 1975). 1 This factor, in part, has resulted in a large institutional and restaurant trade in seafood products, while at-home consumption as a proportion of the total has remained relatively low. Third, the demand for fish and shellfish in the United States has been affected by the market distribution, perishability, and preservation of these products (Christy and Scott, 1965). As one moves inland from those coastal states recognized as major seafood producers, the availability of fresh seafood products falls and the price increases. Finally, there remains a constant concern among U.S. consumers regarding the quality of seafood being sold in the various retail outlets. Meat and poultry products must be inspected and certified by government representatives before sale while inspection and certification of seafood products by government representatives remains voluntary on the part of the seafood processor (Becker, 1933). As such, inspection of seafood products has traditionally been sporadic and minimal. For example, the most intensive ^ Based on a regional study in Texas conducted by the authors.

PAGE 13

3 Federal seafood inspection program which operates under the auspices of the National Marine Fisheries Service inspected only about 20 percent of the 2.8 billion pounds of seafood consumed in the U.S. in 1982 (Becker, 1983). Though the average consumer probably does not realize that seafood requires no federal inspection before sale to the public, he/she is often reminded of some of the adverse health related issues associated with consumption of certain seafood products. For example, periodic newspaper headline scares such as those in the early to mid 1970s regarding high mercury content in certain finfish species and those related to occasional outbreaks of cholera resulting from the consumption of contaminated raw oysters has left the consumer in a quandry concerning the safety of eating these seafood products. Though the preceding discussion points a bleak picture of the future of the seafood industry in the United States, evidence to the contrary suggests that consumption of seafood will be an increasingly important component of the American household diet. For example, the desire among American consumers to increase consumption of lower calorie, natural, and more nutritious foods will likely translate to increased seafood consumption (Slavin, 1984). Recent trends in the per capita consumption of fish and shellfish also suggest that it will be an increasingly important component of the American household diet. As illustrated in Figure 1-1, per capita consumption of commercial fish and shellfish has gradually been trending upwards over the past two and one-half decades. During the 1960-64 period per capita consumption of commercial fish and shellfish averaged 10.56 pounds annually. By the 1979-83 period, annual per capita consumption had increased 21 percent to 12.78 pounds (United States Department of

PAGE 14

4

PAGE 15

5 Commerce, NOAA, NMFS, 1983). Though per capita consumption of poultry has increased at a much faster rate than that observed for seafood, per capita consumption of red meats has been declining since 1971 (Vondruska, 1984). An increase in per capita consumption of fresh and frozen seafood from 1960 through 1983 has accounted for most of the increase in total per capita seafood consumption during this period (Figure 1-1). Averaging 5.82 pounds during the 1960-64 period, per capita consumption of fresh and frozen fish and shellfish increased 35 percent to 7.86 pounds during the 1979-83 period. By comparison, per capita consumption of canned fish and shellfish which averaged 4.22 pounds annually during the 1960-64 period increased only 9 percent to 4.6 pounds during the 1979-83 period. Per capita consumption of cured fish and shellfish which represents a very small portion of the total per capita consumption of fish and shellfish has actually declined in recent decades. Wilson (1982) offers two reasons for the increase in per capita consumption of fresh and frozen fish and shellfish relative to that of canned and cured fish and shellfish. First, there has been an increased availability of fresh and frozen seafood products in retail outlets in recent years. Much of this increase has resulted from the introduction and acceptability by the consumer of highly processed fish and frozen seafood products such as sticks and portions for which per capita consumption increased 130 percent between the 1960-64 period and the 1979-83 period. The second reason offered by Wilson for the increased per capita consumption of fresh and frozen fish and shellfish relates to the recent increase in the number of restaurants in the United States, especially those specializing in

PAGE 16

6 seafood preparation. In support of Wilson's contention, Van Dress (1983) has estimated that the proportion of eating establishments specializing in seafood preparation equalled 4.9 percent in 1979 compared to 2.1 percent in 1966. Two features, already briefly alluded to, distinguish the seafood industry in the United States from those industries associated with most other major food products. First, the seafood industry is highly dependent on an imported seafood product to satisfy domestic demand. Second, consumption of seafood, as opposed to most other food products, is highly related to the away-f rom-home food market. The role of the international seafood market in meeting the increasing per capita and total demand for seafood in the United States can be observed with the aid of Figure 1-2. Between the 1960-64 period and the 1979-83 period, total consumption of edible fishery products increased by over 75 percent (Figure 1-2). This increase occurred in a time period during which the population of the United States increased only about 23 percent. As the consumption of edible fishery products trended upwards, the composition of total supply shifted significantly (Figure 1-2). Domestic supply of edible fishery products remained relatively stable from 1960 through 1975 at approximately 2.5 billion pounds annually (round weight). Associated with the passage of the Magnuson Fisheries Conservation and Management Act of 1976, domestic supply of edible fishery products in the United States experienced a sizeable increase, and has averaged approximately 3.25 billion pounds (round weight) annually since that time. Domestic supply appears to have stabilized at this higher level in recent years. Overall, a 35.5 percent increase in the domestic supply of edible fishery products

PAGE 18

8 occurred between the 1960-64 and 1979-83 periods. While the domestic supply of edible fishery products remained relatively stable throughout the 1960s and early 1970s, considerable growth occurred in the imports of edible fishery products. Imports of edible fishery products exceeded that of domestic supply beginning in 1966 and have since continued to surpass domestic landings (Figure 1-2). Since the passage of the Magnuson Fisheries Conservation and Management Act of 1976, imports of edible fishery products have stabilized somewhat, averaging 4.75 billion pounds annually (round weight). Compared to a 35.5 percent increase in the domestic supply of edible fishery products between the 1960-64 and 1979-83 periods, imports of edible fishery products have increased by 131 percent, or almost four times that of the increase in domestic supply. Imports of many of the higher valued seafood products such as shrimp, scallops, and lobsters and imports of those fishery products used in the preparation of processed frozen seafood items and items served by the fast service eating establishments, such as fillets and steaks frozen in blocks, have risen especially sharply in recent decades. As the above discussion suggests, the growth in imported edible fishery products is at least partially in response to a growing domestic demand not met by domestic supply at acceptable prices. Furthermore, given the apparent stability of domestic supply, further increases in domestic demand will have to be met v/ith concurrent increases in imports. The other feature that distingushes the seafood industry from those industries associated with most other food products concerns the extent of the away-f rom-home market in the sales of seafood products. Though no precise data is available indicating the extent

PAGE 19

9 of away-f rom-home versus at-home consumption of fishery products, it is estimated that anywhere from one-third (Vondruska, 1985) to twothirds (Bockstael, 1984) of all seafood is consumed away from home. As noted by Bockstael (1984) the demand for many species (e.g., Crustacea, fresh finfish) is highly sensitive to changes in income due to the predominate restaurant trade in these species/products Consequently, in recessionary time periods when travelling, vacationing, and hence restaurant trade are depressed, demand for these species/products will also be depressed (Bockstael, 1984). During the 1973-74 recession and the more recent economic slowdown starting in 1979, the depressed demand for seafood resulted in a lowering of prices to the fishermen to such an extent that the National Marine Fisheries Service was compelled to initiate emergency programs such as the Catch America Program with the intent of increasing consumer demand. Economic slowdowns and/or recessions are a frequent occurrence in all or most developed countries. Thus, the cyclical demand for seafood products in the United States will likely be a recurring theme given the dependence of seafood consumption on restaurant trade. Objectives Given the apparent relationship between seafood consumption and restaurant trade, demand for seafood can be expected to rise and fall in a cyclical manner in conjunction with oscillations in the general economy. One means of alleviating the cyclical nature of the demand for seafood is to encourage increased at-home consumption of seafood. This can be done with increased at-home demand either at the expense of the away-f rom-home market or independent of the away-f rom-home market.

PAGE 20

10 If as in the first case, at-home consumption of seafood is increased at the expense of the away-f rom-home market, little change in the total demand for seafood may be experienced. If, on the other hand, at-home demand for seafood is increased independent of the away-f rom-home market, total demand for seafood by definition will be increased. This could put pressure on many of the already heavily fished stocks (as of 1974 about 62 percent of the economically important fisheries in the United States were fully utilized or overfished (Eckert, 1979)) and lead to an even greater demand for imported seafood. In order to effectively increase at-home demand for seafood, a thorough understanding of those factors hypothesized to determine at-home consumption of seafood is required. Only then can the effects of the at-home seafood market on the away-f rom-home market and ultimately on the domestic fishing industry and on import demand for seafood be fully comprehended. Thus, the overall objective of this study is to examine and quantify those factors hypothesized to determine at-home consumption of total seafood and specific product forms (fresh, frozen, canned, shellfish, and finf ish) In relation to this objective, this study is designed to provide information concerning seafood marketing implications based on empirical findings and historical trends related to factors which determine seafood consumption. This information is essential in evaluating fishery market legislation, management alternatives, long-term trends, and promotion and marketing programs. The format of this study proceeds as follows. A review of literature, the models to be estimated, and a discussion of statistical considerations and data used in analyzing the models are provided in

PAGE 21

11 Chapter II. The empirical results associated with weekly total seafood consumption are analyzed in Chapter III. In Chapter IV, a discussion of the results associated with the specific seafood product form models is presented. In the last chapter of the main text, Chapter V, conclusions of the study are presented as are suggestions for future research in this area.

PAGE 22

CHAPTER II THEORETICAL CONSIDERATIONS, REVIEW OF RELATED WORK AND MODEL DEVELOPMENT Cross-sectional consumption studies have become increasingly accepted and utilized during the past two decades in conjunction with the increasing availability of appropriate data sets and the advent of the high speed computer. With the increased volume of literature dealing with the estimation of cross-sectional demand has come increased sophistication in terms of model development and estimation techniques. Regardless of the degree of sophistication employed in the analysis of cross-sectional data, the fact remains that the theory of demand is the basis for all model development in this area. Given this fact, the present chapter begins with a review of the theory of demand. This theory is then adapted to estimation of cross-sectional quantity demanded and expenditure models. Following this section, a review of related seafood consumption studies is presented. The chapter then concludes with a discussion dealing with specification of the models used in this study, the statistical procedures employed in the estimation of these models, and the data used in the analysis. Theoretical Considerations of Demand and Consumption Theoretical Considerations of Utility Maximization The theory of demand is developed on the postulate that consumers maximize utility subject to a resource (income) constraint, i.e., 12

PAGE 23

13 u (q) ^ (2.1) subject to p'q = m max u where u = utility derived from the consumption of q q = an n-element column vector of quantities of commodities available in the market place p = an n-element column vector of market prices m = consumer income To maximize u(q) subject to the income constraint, the Lagrangian expression L(q,X) = u(q) + \(m p'q) (2.2) is differentiated with respect to the arguments q and A. The resultant derivatives are then set equal to zero. The following n + first order conditions are the result of such a procedure: |H_ Xp = 0 (i = 1, 2, ., n) (2.3) 3q i i m p q = 0 The first n equations of (2.3) satisfy the condition that in equilibrium the consumer has equated the ratio of marginal utilities derived from the consumption of any two commodities to the ratio of their respective prices. Additionally, in equilibrium, the marginal utility derived from the consumption of each commodity divided by its respective price equals A, the marginal utility of income. The last equation of (2.3) reinforces the condition that in equilibrium the consumer has exhausted his total resources (income).

PAGE 24

14 Since the solution to (2.3) depends only on prices, income, and the utility function, (2.3) can be solved yielding n + 1 equations, one for each of the q.'s and one for X in terms of p and m. i q i = q (p,m) (i = 1, 2, n) (2.4a) X = X( p ,m) (2.4b) The expressions in (2.4a) are referred to as demand functions, with the demand for each of the n goods being expressed in terms of its own price, prices of all substitutes, and income. Price and Income Considerations Economic theory suggests that demand functions satisfy certain restrictions. One restriction is derived from the "fundamental equation of value" theory which decomposes the effect of a price change into the substitution and income effects, i.e., dq. 3q. 9q -A = _„ — 1 (i,j = 1, 2, n) (2.5) 3p. dp. u=const J 3m A good is said to be normal if it has a downward sloping demand curve or equivalently a negative own price elasticity. Otherwise it is a Giffen good which implies an upward sloping demand curve. The first restriction which follows from the assumption of a strictly convex indifference map implies that the own-price substitution effect is 1 negative i.e., 9q. (t-M <0 (2.6) dp u=const i ''"See Hicks (1957) for a proof of this and subsequently discussed restrictions

PAGE 25

15 Hence, for a good to be a Giffen good, the income effect must not only be negative but also outweigh the own-price substitution effect. Furthermore, a good whose income effect is negative is referred to as inferior and conversely, a good whose income effect is positive is referred to as a superior good. An Engel curve which relates consumption or expenditures on a good with income will thus be downward sloping in the case of an inferior good and upward sloping in the case of a superior good. A second restriction offered by economic theory is that demand functions (2.4a) are homogeneous of degree zero in all prices and income. This suggests that an arbitrary scaling of all prices from (p, m) to (otp, am) will have no effect on quantity demanded of any good Unfortunately, these two restrictions have little to offer when estimating the demand for a single good. Though the own-price substitution effect is negative, the demand curve can be upward sloping given that the good being analyzed is inferior and inferior to the extent that the income effect outweighs the own-price substitution effect. Thus, no restrictions can be imposed, a priori, on the sign of 2 the own-price coefficient. Furthermore, since a good can be either inferior or superior, no restrictions can be imposed, a priori, on the sign of the income coefficient. The second restriction is also of little value in the estimation of single demand equations since rarely if ever are all prices included in a single demand equation. Though these restrictions offer little value in the estimation of single 2 This is of little concern in empirical demand analysis since few if any goods have ever been shown to be Giffen goods.

PAGE 26

16 demand equations, they do show the importance of including prices and income in single demand equations. Though the above restrictions offer little assistance in the formulation of single equation demand models, they are of value in the formulation of complete systems of demand equations. There are additional restrictions offered by economic theory which are useful in the formulation of complete systems of demand equations. First, as illustrated by the budget constraint (equation 2.1), the sum of expenditures on individual commodities must equal income in equilibrium. A second restriction suggests that the sum of marginal expenditures is unity, i.e., n 3q. E p = 1 (i = 1, 2, n) (2.7) 1 = 1 This restriction guarantees that an increase in income is associated with increased expenditures on at least one good. Third, the zero homogeneity of the demand functions yields another n restrictions on the slope coefficients of the following form: dq n 3 q Finally, another |n(n 1) restrictions can be obtained through symmetry conditions 9q. 3q 3q. 3q tt+ q, = ~+ q, -rr 1 (i 'J = l > 2 > • n > < 2 9 ) 3Pj J am 3p^ i 3m

PAGE 27

17 Though these restrictions are vital in the formulation of complete systems of demand equations, again they are of little value in the formulation of single demand equations. Aggregation Considerations Though the demand functions developed earlier and specified in equation (2.4a) provide the foundation for demand analyses, estimation of the equations as specified in equation (2.4a) is generally difficult because the equations refer to only a specific individual or household and hence variation in the data is not present. Empirical demand studies, on the other hand, are generally conducted by aggregating consumption and income across all households for several different time periods or, alternatively, treating the household as the main consuming unit while estimating the demand equations for a crosssection of households. The first approach, referred to as a timeseries analysis of demand, is used to study the effect of changes in prices and income on consumption through time. With the second approach, referred to as a cross-sectional demand analysis, differences among household units are explicitly accounted for in the estimation procedures. Thus, determining the effects of differences in socioeconomic factors and income across households on consumption patterns becomes the primary objective of cross-sectional studies. Given the cross-sectional nature of the present study, the discussion to follow centers on the estimation of cross-sectional demand functions.

PAGE 28

18 Cross-Sectional Demand Analysis Considerations Price, Income, and Socioeconomic Variable Considerations Two features of cross-sectional demand analysis of concern in doing applied research in this field relate to proper model specification and method of estimation. In terms of model specification, concern has traditionally centered around variable selection. In practice, researchers have estimated demand functions (Equation 2.4a) when using cross-sectional data as follows: qiJ = qij(nj, Zj) (i = 1, 2, n) (2.10) (J = 1, 2, k) where q^j = quantity of the ith commodity demanded by the Jth household mj = income of Jth household Zj = a column vector of socioeconomic characteristics particular to the Jth household The demand functions specified in (2.10) differ from those given by (2.4a) in three aspects. First, a vector of socioeconomic characteristics (Zj) whose values are specific to the Jth household has been included in the latter equation. Second, the vector of prices (p) has been excluded from the latter equation. Third, all variables have been subscripted denoting the Jth household. Given the specification of the cross-sectional demand functions (2.10), estimation entails the use of a sample of households differing in socioeconomic characteristics and income rather than that of a representative household as presented in (2.4a). Differences in

PAGE 29

19 characteristics of households, thought to influence tastes and preferences and thus expenditures and/or quantity demanded of commodity i, are controlled for by way of the vector of socioeconomic characteristics (Zj) particular to each household. Correct specification of the vector Zj continues to create controversy even though considerable groundwork in this area was provided more than two decades ago by Prais and Houthakker (1955) and Burk (1961). The problem with specifying the vector Zj "correctly" is that it can potentially vary depending upon the commodity being analyzed. For example, the socioeconomic characteristics determining household demand for alcoholic beverages may depend on a completely different set of cultural factors than those factors determining household demand for milk. Thus, a review of both economic literature and literature from disciplines specific to the commodity being analyzed is essential for the correct model specification. Prais and Houthakker and Burk discuss the importance of examining factors such as family composition, social class, religion, and demographic characteristics when analysing household demand for food. These exploratory studies have subsequently been extended and refined in an attempt to more closely model the actions of the household. Two areas specific to the estimation of the household demand for food frequently addressed by subsequent researchers are those pertaining to measurement of household composition (e.g., Blokland, 1976; Muellbauer, 1974; Buse and Salathe, 1978; Murphy and Staples, 1979) and the incorporation of the opportunity cost of time of the homemaker (e.g., Mincer, 1963; Prochaska and Schrimper, 1973; Redman, 1980).

PAGE 30

20 With respect to household composition, the authors generally attempted to standardize the household in order to show the impact that different household members, varying in terms of sex and age, had on the consumption behavior of the household. The one exception to this rule is provided by Murphy and Staples, who attempted to explain differences in consumption among different households by examining the different stages of a household's growth and maturity. With respect to the opportunity cost of time, it has long been recognized that a commodity, when consumed at home, is not in the same form as when purchased. Rather, value is added to the commodity after it is purchased to transform it into some "new" commodity suitable for consumption. Mincer (1963) cognizant of this fact demonstrated that estimated income elasticities for a variety of commodities will tend to be biased if the opportunity cost of time is omitted. Prochaska and Schrimper (1973) analyzed away-f rom-home consumption with respect to the opportunity cost of time of the meal preparer. Redman (1980) analyzed the impact of women's time on away-f rom-home consumption and for prepared meals based upon the concept of the opportunity cost of time as presented by Gronau (1977). Though conceptually similar, the studies of Prochaska and Schrimper and Redman differ somewhat in the treatment of the opportunity cost of time. Prochaska and Schrimper estimated a wage rate for the meal preparer based upon a set of arguments (education and age) and then included the estimated wage rate as an argument in the demand equation for away-f rom-home consumption. Redman, on the other hand, introduced those arguments hypothesized to affect women's opportunity cost of time (education of homemaker, age of

PAGE 31

21 homemaker, age of children, employment status of homemaker) directly into the demand functions for meals away from home and prepared food. The rationale for excluding prices from cross-sectional demand analyses (equation 2.10) is based on the concept that households observe the same price for any given commodity at a point in time. Though the validity of this assumption has been addressed (e.g., Mincer, 1963) and relaxed in applied research (e.g., Capps, 1982; Cox et al., 1984), the exclusion of prices from cross-sectional demand analyses remains the prevailing practice. Those who include prices in a cross-sectional demand study do so on the premise that prices can be expected to vary in some systematic manner across space and time (given a sufficient period over which the data were collected). However, when prices are included in the model, the interpretation of the estimated price coefficients becomes exceedingly difficult due to price variations independent of shifts in supply. For example, price variations can result from differences in average prices per unit due to quality variation, price discounts associated with larger purchases, and/or price variations resulting from added services provided with the "basic" commodity. When price is excluded from the estimated model, variables which relate to these price variations are included in the model. However, often the parameters associated with these variables represent a composition of two effects: a direct effect associated with the specific variable being analyzed and a price effect. For example, assume the average price paid per unit of a commodity, q^, to be positively related with income, m. Hence, as household income increases total expenditures on that commodity, equaling P^q^> will tend to increase proportionately more than the increase in q..

PAGE 32

22 As discussed by George and King (1971), the increase in average price per unit of the commodity associated with an increase in household income can be viewed as a demand for quality or services. Furthermore, an estimate of the quality elasticity for commodity q^ can be defined as the difference between the expenditure elasticity for that commodity with respect to income and the quantity elasticity for that commodity with respect to income (George and King, 1971). Given that other price variations related to space, package size, etc., have been adequately accounted for by inclusion of variables such as region, urbanization, family size, etc., in the estimated equation, the quality elasticity hence measures the percentage change in average price paid for a commodity with respect to a percentage change in income-^ and can be of considerable interest "as a measure of consumers' desire for improved quality or services, given the present average or standard quality" (George and King, 1971, p. 72). Though the "quality/services" concept as generally discussed is associated only with income, it can easily be extended to reflect other economic variables which take continuous values. Statistical Considerations The statistical considerations most often addressed with respect to cross-sectional demand analyses concern the functional form of the demand equations and the treatment of nonconsumers in the analyses. Though considerable research has been conducted in an effort to find the "ideal" functional form (e.g., Prais and Houthakker, 1955; 3 For proof of the relationship see George and King (1971, p. 72).

PAGE 33

23 Leser, 1963), no single functional form has emerged as clearly superior under all conditions. There are, however, certain criteria that should be considered when selecting and judging the appropriateness of a given functional form (Brown and Deaton, 1972; Tomek, 1977; Hassan et al 1977). First, the functional form should allow for the possibility that the commodity will not be consumed given an income below some initial level. Second, the functional form should allow for a declining marginal propensity to consume with increased income. Third, the functional form should allow for a satiety level which provides an upper bound on quantity consumed. Finally, simplicity and convenience of estimation need to be considered. Though these criteria are valid when considering quantity consumption functions, they appear to be overly restrictive when considering expenditure consumption functions. Given the additional aspect of demand for quality and services, there is no reason to necessarily expect the marginal propensity of expenditures associated with some commodities to decline with increased income. Similarly, a satiety level associated with expenditures on some commodities is not necessarily expected, a priori. As a commodity becomes more narrowly defined in an analysis, the percentage of individual households not consuming that commodity naturally increases. Therefore, a decision whether to include in the analysis those households not consuming the commodity being analyzed needs to be made. Currie et al. (1972) provide a good discussion concerning under what conditions it is logical to exclude (include) nonconsuming households from (in) the analysis. Basically, the issue reduces to the following premise: if the consuming and nonconsuming households can be considered as having identical behavioral patterns,

PAGE 34

24 then there exists no rationale for excluding the latter group from the analysis. The parameter estimates associated with such an analysis can then be interpreted as reflecting the average of both consumers and nonconsumers In order to determine whether consumers and nonconsumers represent a homogeneous group, the reasons why nonconsumers may be present in a given sample are addressed by Currie et al. The first reason provided by Currie et al for observing nonconsumers in the sample proceeds as follows. Assume the distinction between consumers and nonconsumers can be adequately summarized in terms of some qualitative variable(s), say religion. In this case, two options are available to the researcher. First, the researcher can exclude the nonconsumers from the analysis, in which case the results should be interpreted as exclusive of that religious segment of the population. Alternatively, the researcher has the option of including all households in the analysis by explicitly accounting for the qualitative difference of religion. As a second example, the authors consider the case in which nonconsumers can be explained as a result of a difference in the level of some quantitative variable, say income, between them and the consuming group. This being the case, there is no reason to expect behavioral differences between the consuming and nonconsuming groups if incomes were equal. With an increase in the level of income, nonconsumers should enter the market and react in a similar manner to that of the consuming group. Hence there is no rationale, from a theoretical standpoint, for excluding the nonconsuming group from the analysis under this condition.

PAGE 35

25 As a final example, the authors consider the case in which nonconsumers are observed only because the time period represented by the survey does not cover a sufficient span of time necessary to observe consumption by most or all households. As in the previous case, there exists no reason to expect nonconsumers to exhibit a different behavioral pattern from that of consumers and hence there exists no economic reason for excluding them from the analysis. Summarizing these three cases, if the differences between consumers and nonconsumers can adequately be accounted for, then there exists no reason, from a methodological viewpoint, for excluding the nonconsumers. Though there is no methodological rationale for excluding the nonconsumers, caution must be taken when including nonconsumers in the analysis because of statistical problems. As discussed later, use of ordinary least squares when the data includes a large percentage of nonconsumers will generally be inappropriate because of resulting biased and inconsistent estimates of the true parameters. An appropriate statistical technique to be used in conjunction with problems of this nature will be presented towards the end of this chapter. Related Seafood Consumption Studies When compared to cross-sectional demand studies on those food commodities which comprise a large percentage of the consumers food budget, cross-sectional demand analyses for seafood products tend to be somewhat limited. This probably reflects a lack of consideration of "nontraditional" agricultural commodities in such surveys until recent years

PAGE 36

26 Purcell and Raunikar (1968) provided one of the first comprehensive demand studies for seafood and seafood products. Though their results may be of limited use in providing an understanding of current U.S. seafood consumption patterns due to the regional specificity and age of the study (the data consisted of quarterly observations on 160 households in Atlanta, Georgia, over the 1958 through 1962 period), the study does provide some valuable information. In the analysis, the following set of arguments were used to explain expenditures on seafood: race, household composition, annual household income, seasonality, a trend variable, gifts, and price. The results indicated that all variables with the exception of price and seasonality were statistically significant in explaining expenditures on seafood. Using data based on the 1972-74 BLS Consumer Expenditure Survey, Salathe (1979), Capps (1982), and Perry (1981) each analyzed the consumer demand for seafood and/or seafood products. Salathe expressed expenditures on total seafood and two specific forms (canned fish and fresh/frozen fish) as a function of household income and household size. The analysis presented by Salathe may not be directly applicable to this study for at least two reasons. First, arguments other than household income and size undoubtedly influence expenditures on seafood and seafood products. To the extent that the excluded variables are correlated with household income and household size, the estimated parameters associated with these two variables will tend to be biased. The second reason that the results presented by Salathe may not be directly applicable to this study relates to the econometric technique employed by the author. In the analysis, Salathe included both consumers and nonconsumers of seafood products and proceeded to estimate

PAGE 37

27 the expenditure equations via ordinary least squares. Given the relatively large proportion of nonconsuraing households of seafood products in the 1972-74 Consumer Expenditure Survey, the statistical technique used by Salathe is probably inappropriate, which will also lead to biased estimates of the true parameters. Given the possible bias of the parameter estimates presented by Salathe, they must be interpreted with caution and viewed as only a rough approximation of the true underlying parameters. Salathe found the expenditure elasticities with respect to income for aggregate seafood and its two components to be highly inelastic, ranging from 0.21 to 0.38. The estimated household size elasticities of expenditures on aggregate seafood and its two components were somewhat less inelastic, ranging from 0.36 to 0.57. The analysis presented by Capps is more complete than that of Salathe 's in some areas. The strength of the model developed by Capps lies in the specification of the seafood expenditure equation which was expressed as a function of region, urbanization, race, marital status, education, occupation, tenure class, employment status of the female household head, season, household size, household income, and price. The drawback of the model presented by Capps lies in the exclusion of those households who reported no expenditures on seafood during the survey period. The consequence of excluding the nonpurchasing households from the analysis is a loss of valuable information which could potentially help to explain why some households purchased seafood during the survey period while others did not. In addition, the exclusion of nonpurchasers from the analysis indicates that the results must be interpreted with respect to only those households purchasing

PAGE 38

23 seafood products. Capps' study does provide considerable information for assistance in determining which variables should be included in a seafood consumption equation. Using a quadratic expenditure equation Capps found that region, urbanization, race, martial status, household size, household income, and price all contributed in a statistically significant manner in explaining seafood expenditures. In agreement with the results provided by Salathe, Capps found the income elasticity of seafood expenditures to be extremely inelastic, equalling 0.1651. With respect to family size, Capps found an elasticity of 0.2296 which is somewhat less than that reported by Salathe. Perry's analysis of seafood expenditures was by far the most complete of those utilizing the 1972-74 BLS Consumer Expenditure Survey. In addition to specifying a rather complete model describing expenditures on total seafood and specific product forms (shellfish, canned fish, whole fish, and filleted/steaks fish) in terms of variables introduced into the equations, the analysis incorporated all households in the survey. Furthermore, to avoid the likelihood of biased estimates of the true parameters associated with using ordinary least squares when a large concentration of zero observations for the dependent variable is presented in the data, Perry estimated the equations via a Tobit procedure. This procedure provides asymptotically consistent estimates of true parameters given a correct model specification. The variables included in the various seaf ood/seaf ood product expenditure equations were household income, race, urbanization, expenditures on food consumed away from home, occupation of household head, education level of household head, and household composition. By estimating separate expenditure functions for the four

PAGE 39

29 seafood/seafood products classifications by region and different income groups, Perry in total estimated 85 equations explaining seafood/ seafood product expenditures. Though there are certain advantages to estimating separate equations for different income groups, regions, etc., the value of such a study in terms of answering national policy questions becomes increasingly limited with increased refinements. Given the relatively large number of equations estimated by Perry, a discussion of the results necessitates generality. Those variables found significant most often in explaining seafood expenditures were income, race, and household composition. Other factors, most notably urbanization and education, were important determinants of seafood expenditures only in isolated instances. The results presented by Perry were for the most part consistent with those reported by other researchers. As in the studies conducted by Salathe and Capps, Perry reported the income elasticities of seafood/seafood products to be extremely inelastic. By region, the income elasticities for total seafood expenditures ranged from a low of 0.069 in the South to a high of 0.204 in the Northeast. In terms of specific seafood product forms, income elasticities were generally statistically significant by region for shellfish, canned fish, and filleted/steaks fish and insignificant for whole fish. The reported income elasticities for shellfish expenditures were consistently higher than that for the other individual products and ranged from a low of 0.069 in the South to a high of 0.344 in the Northeast. In a recent study employing the 1977-78 Nationwide Food Consumption Survey data, Haidacher et al. (1982) analyzed expenditures and quantity demanded for total seafood, shellfish, and finfish.

PAGE 40

30 Including all households in the analysis, the authors estimated the various equations using ordinary least squares. As discussed in the context of the previous studies, this estimation procedure may be inappropriate when a large proportion of the households did not consume the commodity. Given the large percentage of households not reporting consumption of seafood products in the 1977-78 Nationwide Food Consumption Survey, caution should be used in the interpretation of the results reported by Haidacher et al In the analysis, the authors expressed expenditures and quantities consumed as functions of region, race, urbanization, income, household size, household composition, season, and the number of guest meals. Given the similarity between the study by Haidacher et al. and those by Capps and Perry, one would expect the results reported in the respective studies to be similar, which in fact was the case. The income elasticity for total seafood expenditures as reported by Haidacher et al equalled 0.16 which was the same as that reported by Capps and within the range of those reported by Perry for the various regions. Income elasticities for shellfish and finfish expenditures were given as 0.73 and 0.03, respectively. These estimates are in agreement with those presented by Perry to the extent that the income elasticity of shellfish expenditures tended to be somewhat higher than that of finfish expenditures. However, in absolute magnitude, the income elasticity of shellfish expenditures reported by Haidacher et al. was two to three times the size of that reported by Perry. Some of the discrepency in results probably reflects the different statistical techniques employed in the two studies.

PAGE 41

31 In addition to the expenditure elasticities reported by Haidacher et al., the authors also report the quantity elasticities with respect to income. The quantity elasticities with respect to income were positive for shellfish and negative for finfish and total seafood. The negative quantity elasticity with respect to income for total seafood in conjunction with a positive expenditure elasticity implies a positive quality elasticity for total seafood which equalled 0.20. This consists of a high estimate of the quality elasticity associated with shellfish consumption (0.59) and a relatively low estimate of the quality elasticity associated with finfish consumption (0.10). Summarizing the research to date, evidence suggests inelastic income expenditure and quantity elasticities for total seafood and specific product forms. However, none of the studies conducted to date has made complete use of all data and/or available statistical options. An extension of the work provided by the authors discussed in this chapter is the basis of the next two chapters. The remainder of this chapter lays the groundwork for the models to be estimated. Conceptual Model Development The first task associated with specifying a cross-sectional consumption model involves that of defining the set of arguments comprising the column vector of socioeconomic characteristics, Zj given in equation (2.10). The concept of consumer demand in conjunction with the seafood expenditure/quantity consumption studies discussed in the previous section were of assistance in meeting this objective. The respective expenditure and quantity equations were specified as a

PAGE 42

32 function of the following set of variables (household subscripts have been deleted for notational convenience): EI? = f(Zl, Z2, Z13, M, S) (2.11a) Q = f(Zl, Z2, Z13, M, S) (2.11b) i The variables EXP^ and refer to the expenditures on and quantity consumed of total seafood or specific product form by the Jth household, respectively. The values these two variables take range from zero upwards, with the frequency of observed zero values varying with respect to the specific product form. The region of the country (Northeast, North Central, South, West), the degree of urbanization (Central City, Suburban, Nonmetro), and the season during which the household was interviewed (Spring, Summer, Fall, Winter) are defined as Zl Z2, and Z3, respectively.^ The rationale for including these variables in (2.11a and 2.11b) is two-fold. First, the prices associated with total seafood and the specific product forms vary by region (Zl), urbanization (Z2) and season (Z3) due to differences in aggregate demand and supply. As such, households in different regions and/or levels of urbanization or interviewed in different seasons encounter different prices for the same product. Second, consumption of total seafood and the specific product forms is likely to differ among households by region, urbanization, and season for reasons independent of a price effect, such as tastes and preferences associated with cultural or institutional factors ^See Appendix A for a description of these and the following variables used in this study.

PAGE 43

33 The remaining socioeconomic variable entered into equations (2.11a) and (2.11b), Z4-Z13, M, and S were included to specifically account for variations in seafood consumption among different households resulting from underlying socioeconomic differences which are expected to influence tastes and preferences. The measurement of household composition, Z4, used in this study is a modification of the household life cycle classification proposed by Murphy and Staples (1979) and is more directly applicable to the study than measurements generally proposed for reasons discussed below. For purposes of this study, households were stratified according to ten mutually exclusive life cycle classifications: young single without children, young married without children, young single with children, young married with children, middle aged single without children, middle aged married without children, middle aged single with children, middle aged married with children, older single, and older married (where young is defined as the head of household being less than 35 years old, middle aged is defined as head of household being from 35 years old to 65 years old, and elder is defined as head of household being equal to or greater than 65 years old). There were two reasons for using the household life cycle measurement of family composition in this study as opposed to a more traditional measurement. First, it is useful to investigate the reasons for the changes in apparent consumption of total seafood and specific product forms over the past two decades. Available time series data pertaining to household composition are related to the family life cycle measurement of household composition more closely than with the other measurements. Thus, the life cycle classification was a preferable measurement of household composition in terms of

PAGE 44

34 examining changes in seafood consumption through time. The second reason for using the household life cycle measurement of family composition is based on the hypothesis that household consumption of total seafood and specific product forms is more directly related to the life cycle classification of the household than to other measurements of family composition. For example, households with young children may avoid the purchase and consumption of specific seafood product forms that are known to have bones. While most measures of household composition do not consider the case in which certain households are unlikely to consume a given commodity as a result of certain characteristics of the household members, the life cycle classification of the household does to some extent account for this possibility by stratifying households into mutually exclusive categories according to given characteristics of the household. The race of the household head (Z5), found in previous studies to be of importance in explaining household consumption of seafood, was included in the analysis to account for variations in tastes and preferences among households of different races which would lead to differences in at-home seafood consumption. For purposes of this study the race of the household head was assigned to one of the categories: White, Black, or "Other," where "Other" refers to any ethnic origin other than that of White or Black. ^ Food stamps (Z6) in essence are an additional source of income to households which can be used for the purchase of most food items. Similarly, a household having caught fish for its own use (Z7) has a ^"Other" represents an all-inclusive term referring to households of various ethnic origins such as Asian, Indian, etc.

PAGE 45

35 home produced good intended for consumption. As such, there should be a positive relationship between the catching of fish and at-home con6 sumption of seafood. Employment of the meal planner (Z8), the sex of the meal planner (Z9), family size (Z10), and the education level of the meal planner (Zll) are expected to affect the opportunity cost of the meal planner's time. Following Gronau's (1977) premise, an increase in the education level of the meal planner should result in an increase in the opportunity cost of his/her time, ceteris paribus. Similarly, meal planners employed outside the home are expected to have a higher opportunity cost of time than their counterparts. The planning of household meals has traditionally been associated with the female members of the household. Increases in the family size are expected to be associated with increases in the opportunity cost of the meal planner's time, ceteris paribus. Thus, an increase in family size is expected to result in an increased consumption of the highly processed seafood product forms such as canned seafood products relative to the nonprocessed seafood products such as fresh seafood, ceteris paribus. Though increases in the education level of the meal planner and family size are expected to increase the opportunity cost of time of the meal planner and hence result in a movement of household consumption patterns towards heavily processed seafood products, results supporting this contention are likely to be masked by offsetting factors. For example, increases in the education level of the meal price representing the market price for a similar product in a given region and season was assigned in those cases where the seafood product was not purchased in the market.

PAGE 46

36 planner are likely to be associated with an increased awareness of the nutritional value associated with consumption of seafood products. This increased nutritional awareness and resultant increased consumption of seafood products are likely to offset the expected decline in consumption associated with the increase in the opportunity cost of time resulting from additional education. Similarly, though increases in family size are expected to result in an increase in the opportunity cost of time of the meal planner and hence a potential decline in at-home consumption of seafood, increases in family size by definition necessitates increased consumption in total. Hence the expected decline in household consumption of seafood resulting from an increase in the opportunity cost of time of the meal planner associated with an increase in family size may be offset by increased consumption necessitated by an increase in family size. The number of guest meals served from home food supplies (Z12), found to be a significant factor by Capps (1982) in explaining expenditure on seafood consumed at home, was introduced into the analysis to account for the expected increase in consumption of total seafood and especially those seafood product forms most likely to be served when entertaining guests. Since the less processed seafood product forms are generally associated with a higher quality product and hence viewed as more preferred items, increased weekly consumption of these product forms is expected to be positively related with increases in the number of guest meals. Although Perry (1981) concluded that the money value of away-fromhome consumption was generally unimportant in explaining seafood expenditures for at-home consumption, a similar variable was included

PAGE 47

37 in the present analysis for two reasons. First, increases in the money value of meals consumed away from home^ (Z13) may imply a lower need to consume meals at home, ceteris paribus. The second reason for including the money value of meals consumed away from home is that a large proportion of total seafood consumption reportedly occurs in the awayfrom-home food market. Furthermore, the away-f rom-home trade as a percentage of the total varies substantially from one seafood product to another. For example, consumption of certain shellfish species such as shrimp and lobster occurs largely in institutional and restaurant outlets, while the proportion of other seafood products, such as canned tuna, consumed in the away-f rom-home market is considerably less (Vondruska, 1985). In general, it is expected that those seafood product forms difficult to prepare at home, such as fresh seafood, are most often consumed in the away-f rom-home market. To the extent that consumption of seafood products away-f rom-home substitutes for consumption of similar products at home, the money value of meals consumed away from home and at-home consumption of seafood are expected to be negatively related. The effect of income on consumption in general and on at-home consumption of seafood in particular has been discussed extensively throughout this chapter. In this study, before tax income (M) was used as a proxy for the resources available to the household for the ^Excludes the value of snacks purchased and consumed away from home.

PAGE 48

38 g purchase of seafood products. No distinction was made with respect to the sources of income and their separate effects on consumption of total seafood and specific product forms. Though increases in household income have been found to be related to increases in expenditures on total seafood consumed at home (e.g., Perry, 1981; Capps, 1982), less evidence exists concerning the relationship between income and at-home consumption of specific seafood product forms. For example, the estimated income parameter associated with a specific seafood product form may be either positive or negative depending upon whether that product form is considered to be a normal or inferior good and may in fact even vary among different segments of the population. Substitutes for at-home consumption of seafood and the specific product forms, denoted as S in equations (2.11a and 2.11b), follow from the concept of the demand function provided in equation (2.4a). In most cross-sectional studies of this nature, substitutes are omitted from the analysis because prices of substitutes encountered by any given household should be the same prices as those encountered by any other household. In the estimation of the total seafood consumption models no substitutes were specified. However, with respect to the models for specific product forms, consumption of the alternative seafood product forms were considered as appropriate substitutes. For example, shellfish consumption by a given household was related to consumption of finfish by that same household. Similarly, consumption of fresh seafood was related to the consumption of the summation of 8^ Though an arguement could be made to use total food expenditures rather than income as an explanatory variable in the analysis, the latter variable was used because it is more directly applicable for answering policy oriented questions.

PAGE 49

39 frozen and canned seafood. In a very strict sense, one might consider that consumption of one seafood product is simultaneously related to consumption of other seafood products by way of the budget constraint (equation 2.1). However, given the very small proportion of the consumer food dollar being allocated to seafood purchases, the problem of simultaneity with respect to these variables is probably negligable. In fact, the Longwood Research Group Limited (1984) concluded that heavy users of one category of seafood tended to be heavy users of other types of seafood. This is in contrast to what one would expect to find if in fact the budget constraint played a major role in determining substitutability among alternative seafood product forms. Econometric Models The conceptual models developed in the previous section (equations 2.11a and 2.11b) were fully specified to include the actual variables included in the estimated relationships. Incorporating these changes and making a similar change in the notation yielded the following weekly expenditure and at-home quantity consumption equations: Econometric Models and Statistical Considerations EXP. a Q + o^Xl + a 2 X2 + + a 33 X33 + 6 0 + g-jXl + B 2 X2 + + B 33 X33 + U 2 (2.12a) (2.12b) where EXP weekly household expenditures on at-home consumption of total seafood and specific product forms weekly household at-home quantity consumed of total seafood and specific product forms

PAGE 50

40 X1-X3 = region of the country in which household resides XI = Northeastern region X2 = North Central region X3 = Southern region Western region (base region) X4-X5 = degree of urbanization in which household resides X4 = central city X5 = suburban (metro) nonmetro (base urbanization) X6-X8 = season during which household was interviewed X6 = spring quarter X7 = summer quarter X8 = fall quarter winter quarter (base season) X9-X17 = household life cycle stage X9 = young single adult without children X10 = young married adults without children XI I = young single adult with children X12 = young married adults with children X13 = middle aged single adult without children X14 = middle aged married adults without children X15 = middle aged single adult with children X16 = middle aged married adults with children X17 = elderly single adult elderly married adult (base life cycle stage) X18-X19 = race of respondent X18 = White X19 = Other than White or Black Black (base race) X20 = household presently receiving food stamps (equals 1 if household is presently receiving food stamps, 0 otherwise) X21 = household caught fish for own use (equals 1 if household caught fish for own use, 0 otherwise) X22 = meal planner employed outside the home (equals 1 if meal planner employed outside the home, 0 otherwise) X23 = sex of meal planner (equals 1 if meal planner is female, 0 otherwise)

PAGE 51

41 X24 = family size (total number living in household) X25 = family size squared X26 = number of years of schooling of meal planner X27 = number of guest meals served from household food supply in previous 7 days X28 = dollar value of meals purchased and consumed away from home (excluding snacks) X29 = before tax income (thousand dollars) X30 = before tax income squared X31 = interaction between before tax income and race (X18 • X29) X32 = interaction between before tax income and family size (X24 • X29) X33 = expenditures on (or quantity consumed) of alternative product forms a. a„ a a 33 = estimated coefficients associated with weekly expenditure equations 2 B 3 ^33 = estimated coefficients associated with weekly quantity equations = normally distributed random disturbance specific to the expenditure equations U2 = normally distributed random disturbance specific to the quantity consumed equations Though most of the independent variables included in equations (2.12a) and (2.12b) enter in a binary manner (X1-X23), family size (X24), education level of the meal planner (X26), guest meals (X27), money value of meals consumed away from home (X28), before tax income (X29), and expenditures (or quantities) of alternative seafood product forms (X33) enter the equations in a continuous manner. Among this latter group of variables, family size and income were specified in a

PAGE 52

42 quadratic form. Income was specified in a quadratic form for three reasons. First, the quadratic specification of the income variable allows for a declining/increasing marginal propensity to consume/ purchase with increased income. Second, the quadratic specification of the income variable allows for a satiety level providing an upper bound on quantity consumed yet at the same time does not restrict expenditures to behave in a similar manner. Third, the quadratic specification of the income variable is easily modelled. Though the quadratic specification of the income variable provides no assurance that the commodity in question is not purchased/consumed given an income below some threshold value, as will be demonstrated shortly, the statistical technique used in the analysis does associate a lower income with a lower probability of purchasing/consuming the commodity. Family size was introduced into the analysis in a quadratic specification to account for possible economies to scale in the purchasing and consumption of seafood and specific product forms associated with increased family size. Two interactions (X31, X32) were introduced into the analysis. The first interaction, that between White households and the linear income term, allows for differences in the marginal propensity to purchase/consume total seafood and specific product forms among households of different races. Similarly, the interaction between the linear family size term and linear income term allows for differences in the marginal propensity to purchase/consume among households of different sizes. An argument could probably be made for the introduction of other interaction terms, in addition to the two specified. However, the use of too many interaction terms would have probably

PAGE 53

43 resulted in severe collinearity problems among the regressors. Therefore, those used were only those considered most appropriate. Statistical Considerations The model developed in the previous section can be expressed in matrix form as follows: y t = X t e+U t ifX t e+U t > 0 (2.13) y t = o if x t s + u t < o t = 1, 2, N where y t = dependent variable X t = vector of independent variables 2 U*. = error term assumed iid N(0, o ) This model specification referred to as the Tobit model after its founder is well known and used extensively in economic studies of a cross-sectional nature (see Amemiya (1984) for a good review of the basic model and its uses). Given the specification in (2.13), an assumption is implicitly made that an underlying stochastic index equal to X t 0 + U t is observed only when strictly positive. In other words, y t will only be positive given a value of X t S + U t greater than zero. Otherwise, y t will equal zero. For example, assume two households with identical attributes with the exception of income. Furthermore, assume that the household with the higher income consumed seafood while the household with the lower income did not consume seafood. This would imply that the first household with the higher income had exceeded that threshold level required to consume seafood (i.e., X t (5 + U t > 0), while

PAGE 54

44 the second household with the lower income had not crossed that threshold level (i.e., X t 6 + U t < 0). Factors such as those in X t probably influence at-home consumption of seafood and thus the Tobit procedure is appropriate for this analysis. As shown by Greene (1981), OLS estimates of (2.13) are both biased and inconsistent due to the non-normality of the expected error terms. Thus, some estimated procedure other than that of OLS must be used if unbiased or at least consistent parameter estimates are to be obtained. Since the original work by Tobin (1958), several methods of estimating equation (2.13) which assure consistent estimates of the true parameters have been developed and used (see Amemiya (1984) for a discussion of the different methods). Since the different methods are widely known and should in all cases provide the same parameter estimates assuming a unique maximum, the different approaches to estimating (2.13) will not be discussed. Though the uses and methods of estimation of the Tobit model are well known and documented, less well known is the amount and types of information that can be obtained from the Tobit estimates. The types of information that the Tobit model provide are discussed here and used extensively in the next two chapters. The uses of the Tobit model, first presented by McDonald and Moffitt (1980), are the basis for the ensuing discussion with some modifications to their work added towards the end of the discussion. Those equations which will be modified are assigned the letter (a) after the numerical numbering. The modified equations are assigned the letter (b). The unconditional expected value of the dependent variable in equation (2.13) was shown by Tobin (1958) to equal

PAGE 55

45 E(y) = XBF(Z) + af(Z) (2.14a) where Z = Xg/a 9 f(Z) = unit normal density function 10 F(Z) = cumulative normal distribution function The unconditional expected value of the dependent variable represents the expected value of the dependent variable associated with all observations. Furthermore, as shown by Amemiya (1973), the conditional expected value of the dependent variable for observations above the limit (i.e., positive observations), y ;: ~ is given by The relationship between the unconditional expected value of the dependent variable (expected value associated with all observations) and that of the conditional expected value of the dependent variable (expected value associated with positive observations) is given as follows : E(y*) = E(y|y > 0) = XB + E(u|y > 0) (2.15a) = XB + of(Z)/F(Z) E(y) = E(y*) F(Z) (2.16a) Defined as 1 -Z 2 /2 B'X _L 10 Defined as a

PAGE 56

46 Thus, the unconditional expected value of the dependent variable is equal to the conditional expected value of the dependent variable adjusted by the probability of observing a positive value of the dependent variable, F(Z). Differentiating this relationship with respect to an exogenous variable, gives the effect of a change in the dependent variable resulting from a change in 3E(y)/3X i = F(2) (SECy*)^) + E(y) (3F(Z)/9X i ) (2.17a) Furthermore, it can be shown that the two partial derivatives on the right-hand side of equation (2.17a) are equal to 3F(Z)/3X. = f(Z)B i /o (2.18a) and 3E(y*)/3Xi p i + (a/F(Z)) 3f(Z)/3X i (2.19a) (of(Z)/F(Z) 2 ) 3F(Z)/3X which upon reduction yields 3E(y*)/3X = Zf(Z)/F(Z) f(Z) 2 /F(Z) 2 J (2.20a) Finally, note that after substitution of (2.20a) and (2.18a) into equation (2.17a) and upon rearrangement of the terms, one arrives at the following expression: 3E(y)/3X i = F(Z)3i (2.21a) a much simplier expression than that of equation (2.13). For purposes of this study, a modification of the above equations was necessary. Notably, the expressions given above are based on the assumption that all of the variables in the vector X t are continuous

PAGE 57

47 in nature. However, several of the variables evaluated in the present study were binary in nature (X1-X23). The main consequence of such a specification pertains to the evaluation of XB given a binary variable. Specifically, the probability of observing a positive value of the dependent variable now becomes conditional on the binary variable being evaluated. Thus, the value of the standard normal, (Xg|X^)/a, and hence the value of the cumulative normal density function F(Z |X^) and the value of the unit normal density function f (Z |X^) all become dependent on the binary variable being evaluated. Hence, equations (2.14a) through (2.21a) need to be adjusted accordingly when discussing binary variables. These adjustments are E(ylX i ) = (XelX^ F(Z i |X i ) + of(Z i |x i ) (2.14b) E(y*|X i ) = (Xe|X ) + af(Z i |X i )/F(Z i |x i ) (2.15b) E(y|X i ) = E(y*|X ) F(Z 1 |X 1 ) (2.16b) 9E(y|X i )/8X i = F(Z i |X.) (3E(y*|x 1 )/3X i ) + E(rHx i ) (3F(Z i |x i )9X i ) (2.17b) 8F(Z i |X.)/8X i = f(Z i |X i )g i /o (2.18b) 3E(y*|X i )/3X 1 = Bi + (a/F(Z i |X i )) 3f (Z i |X i )/3X i (af(Z i |X i )/F(Z i |X i ) 2 ) 9F(Z.|X i )/8X i (2.19b) 3E(y*|X i )/3X i = B.[l (ZjJXj) f (Z |X )/F(Z |x ) f(Z i |X i ) 2 /F(Z i |X i ) 2 J (2.20b) 9E(yfX.)/3X. = F(Z i |X i )$ i (2.21b) Technically, it would be preferable to use the concept of a limit rather than a derivative in evaluating equations (2.17b) through (2.21b) due to the discrete nature of Xi in each of the equations.

PAGE 58

48 For evaluation of the continuous variables in the model, use of equations (2.14a) through (2.21a) is valid. The total change in the dependent variable y given a change is as specified in equation (2.17a) can be broken into two components. The first component [F(Z)( 3E(y*)/3X i ] represents the change in the expected value of the dependent variable if above the limit (positive) weighted by the probability of being above the limit (positive). The second component [ E( y*) ( 3F(Z) /dX) ] represents the change in the probability of being above the limit (positive) weighted by the expected value of the dependent variable if above. The breakdown of the total change of the dependent variable into these two components is probably very consistent with the actions of consumers in the market and is thus useful in studying the consumption patterns of households. It is interesting to note the simularity between the Tobit estimates and OLS estimates. From equation (2.21a) it can be observed that the effect of a change in on the dependent variable y is equal to 8i only when F(Z) is equal to one. This is in fact what one would expect since as F(Z) approaches one, OLS estimates should be obtained. Multiplying equation (2.17a) by X i /E(y) gives the Tobit elasticity, n^, which equals Substituting equation (2.16a) for E(y) and making the appropriate reductions provides the following specification of the elasticity of y with respect to

PAGE 59

49 ax. E(y* ) 3X. 3F(Z) ax. = ( E(y*) 1 ) ( ax. )+ (^) ( i ) (2.23) i The total elasticity is comprised of two components. The first component measure the conditional elasticity associated with the nonlimit observations. The second term measures the elasticity of the probabilitiy of participation associated with a change in X.. Data Source and Considerations The 1977-78 Nationwide Food Consumption Survey (NFCS) provides the data used in the analysis. This is the most recent of the household food consumption surveys periodically conducted under the auspices of the United States Department of Agriculture. The survey encompassed approximately 15,000 households throughout the 48 continguous states and contained detailed information on characteristics for each household. In addition, the survey contained detailed information pertaining to expenditures on and the corresponding quantities of a continum of foods consumed at home (measured at the level at which the foods came into the kitchen) by each of the households surveyed. The survey was conducted over a one year period (April 1977 through March 1978) and was stratified according to a variety of factors including season, region, and urbanization in an attempt to have the sample represent the universe of households in the continental United States as accurately as possible. Though information on 14,930 households was provided on the original NFCS data tapes distributed through the Department of Commerce's National Technical Information Service, only 10,689 observations of the original 14,930 were retained for the current analysis. Of the 4,241 deleted observations, approximately 92

PAGE 60

50 percent were deleted due to the household's refusal to report annual income. The remaining 8 percent were deleted due to the omission of other relevant information. In light of the fact that households reporting incomplete income information often provide poor or incomplete expenditure information (Buse, 1979), it was deemed appropriate to delete these observations. While omission of those observations for which information is missing could potentially lead to sampling bias (Maddala, 1977), a comparison of those households not reporting income with those reporting income indicates that this was not a serious 12 problem in the present study. Among those households not reporting income, 50.2 percent of the households consumed seafood at home during the one week interview period compared to 50.8 percent among those households reporting income. Average weekly consumption among those households consuming seafood and not reporting income equalled 1.94 pounds compared to 1.91 pounds among those households reporting income.

PAGE 61

CHAPTER III TOTAL SEAFOOD ANALYSIS The discussion of the results associated with the total seafood models is given in two sections. First, descriptive statistics comparing/contrasting consumers and nonconsumers of seafood as established by the survey data are presented and briefly discussed. In the second section, the results of the Tobit analysis associated with the total seafood models are presented and discussed. Descriptive Statistics Associated with Total Seafood Analysis Descriptive statistics of the 10,689 households included in the analysis are presented in Table 3-1. Though the information presented in this table is of a descriptive nature without any attempt to separate partial effects, the information does provide an overall comparison of those households which consumed seafood during the one week survey period versus those households which did not. The statistics provided in Table 3-1 are assigned to one of four categories. The first category, labelled total sample, gives the mean values of all variables used in the analysis. For example, as indicated in the table, 24.8 percent of the households in the analysis resided in the Northeast region (XI). The second category, labelled nonlimit observations, provides the mean values of all variables for those households consuming seafood at home. For example, of those 51

PAGE 62

52 >> c s00 c o c H =0 •H 4-1 0 E J_ 4J 3 St C CO CO Pk u C 0 o u 4-1 u a+ 0 w CO M c 0> c E UJ H 3 cn E CO c H > o H 1h u 1 — 1 o c o ss o CO 4-) c t-i -H 0 0) E H CO ,— 1 N 3 H CU w co c > c o o o c o E CJ w to X o o o V4— t CO CD CO CO CU =a t-i r-l rH QJ cco co E E l-i 3 o CU CO CO 1 1 i— i E C — 3 o — i CO CJ CO c c 4J o o o co 4J cu iH XI CO •H CO > 14-4 o CO u •H 4-1 CO •H 4J co CO cu > •H 4-> CU •H l-i u co CU Q i — i 1 CO o 0 00 i— 1 CU XI 4-4 CO CO H CJ> 4-> c QJ U u QJ c CO OJ ,— i sr m m no st st in o" d d d in O CTn O on r-~ rm i— I CN CO i— I o o o o on oo m co O H O CO CN CN CO i — i o d d d 00 i— I CN On , in CO u l-i X CO 4-1 X X OJ 4J s X c •H w C o CJ CO CJ c •H c o CO U 1-1 c 4-1 i-H CO ~QJ QJ 1-1 CO CO X — 1 X r X CJ N Ih S-i o a 4-1 4-1 4J •H 4-1 3 E 3 o In CO C C X c o H O O o QJ CO 0) 3 O CO 00 >c CO X cj cn CO QJ u a 06 r_o C/J X X co i-i C QJ •H E l-i E a. 3 C/J CO CJ co CO X 00 w X 1-4 CU 1-H 4-1 i — I c CO -H

PAGE 63

53 c oo o c 3 -H DO H 4J CO Ih 4-1 fj O CO u o o In u o s Gfl X s— Co 0 E 4J •rl 3 H 4-1 cn cn E CO c H > o H s~ 0 CO c CO 0 o CO DO 4-1 a U •H o CO S H i H 4J CN rH co c > C 0 o c o CJ obs cn o H 0) W E E u 3 CO Co ns cn o i— i cn u ra c 3 4-> o c O u s 4J u c CO 01 4J as U lOcnCOCN^-^CNCNintN — ir-ooocsicNvocor^-c^ o o o o o o o o o o m cn on ctn r-~ c O u M Oj ex in •— i If) CN CN O OCNHO oooooooooo cc CN O o o o c CO OJ mcNcj\cncN -^incnm\O r -ivoor^o OOO'— i O — iOCNOO oooooooooo h m -j cn cn cn COOH o' o' o' incomHinmoHOMn inmcomNHinvooo OOO^hO^-iOCNOO oooooooooo CN O CO 00 O —H o o o ^ CN 1 — 1 rH X X x — c c 3 OJ 3 QJ 01 CJ su T3 X3 rH •n rH H H rH H •H x: •H Xi X! u JS u U 0) u X H o X! 4-1 o o 4-1 •H IS •H 13 u T3 3: a OJ o Co CO Co cn B CO E 1 — 1 0 DO DO 00 00 X! 3 C c c 01 3 3 3 cn c O O O 3 o ^— X vD m >— 1 cn — i — i X i— i X X X c c en c o CO ^4 o U Ih M — ~_o 13 rH rH •H rH H •H XJ H Xi X! u X! u U u o X! th c 4-1 •H • •H CO cn a CO 0) OJ o i — 1 X) ^4 •H rH H X CO l-l 00 u — / c u c u -3 •H co •H 0) O 03 E CO E rH CJ •H X) -a TJ C 'rCO QJ U co H CO 00 CO 'CO cn E a C3 a CO >-, o CO a a rH • — I 1 — 1 i — i rH i— I U U -o •a ~3 O a -o a -a T3 •o a •H •H •H •H rH — i UJ 4-1 c o T3 c o OJ a X ~cn CO i — < — 1 ctj OJ X X kH ^H o u 0 H CJ u •r^ Xi ca 0) SI 4-1 r-1 CO o a

PAGE 64

54 U OO o c CO -H c o •H 41 t-l 4-> 3 E CO •H > O iH U U w 0) C CO O X> 2= O CN CO 4-1 U -H CU E E -H CO G O •H 4-1 D i—l CO CO c > c o u O C 0) O v_/ CO o CO 0) Sh i-H o o o E 3 cu co co E C 3 OH co u co C C 4-1 o o C 4-1 >, u o cc a 4-1 CO u CN O d d O O O 4-1 c CD U OJ C CO o o CN o o r~cn o On d d cn 00 roo d d CN CO o a> cn r~ d d vjD cn cn d d cn o cn d d oo CT i — i -jin d d r~~ cn -jin d d m 4-l a. cn O O CN CD 0 x: H X 03 O CD to I — 1 cu 4-> cn c tn CO cn •H CN 10 C CN co CD X X CO C-M X CO CD X co E CD cd X x E X i— i > — •J CO o •H Si CO O c/1 O E — i -i X 0) CO E CD w u C/]

PAGE 65

55 c sCO c c H DO H *J QJ 1 i 1 1 I i •^"^ in i i i 1 1 i 1 1 1 1 co co i i I | 1 1 1 < 1 — o Q U— | (J H O CO Jh a o co 1 u CD CD CN c CO c t — I r^ Q CO rj CD CO E w Jj c r•H o rH co CO rH m o o a •H CO in o • CN § H 4-1 D i — i 00 CN 25 o m H CO 4-1 • • w CO d > U co CN 1 —J i — i in c 0 SH < i— 1 I— 1 — i t— h m o c 0) co co o >H o GC a 4-> TO CJ / "% cn 0) or pi E E \0 r-CN r-H a; rH 3 CO I - o o o U c 4-) CN X O o c> m 3 w -a c rCN 3 3 o co CN cr O sr-H X ai Cfl sua P. uio G S Cfl eg c — t— 1 CO jC X u Ih •H Cfl CO rH 3 3 3 CJ 3 E 4-1 rH r— Ih u E a O 00 rH CJ Oi E M CN CJ O o J3 — < <4-| X Ih "3 -3 a E E o CN j: . 4h 33 "3 H c 3 c 3 3 w o 3 3 BJ o 0) Ih 3 Ih J3 CO 3 rH r^ H co E CJ 3 cfl Cfl Cfl >-, a CO *J M -Q rH CJ 3 3 hj 4-1 ca 3 E CO H E C C •H o o u O X; 3 rH C C J3 Th E H H 3 CJ z 3 n u CO -3 3 CJ 3 Ex H HH

PAGE 66

56 >1 c *CC o o c •H 00 H 0) -J Cj CO (J 3 o c M 1 o On o CO t/] I. a o i •H 3 •H 4J cn CO E c H > c H u a c CO o SB o C0 4-1 -H 0) S E -H co a o •H 4-1 D H 51 cn e > c o ^ O C CD CJ ^ CO o to 03 st-H QJ to E E (-4 a ^-v CD to to -i e a c rH cn u a a c 4-1 0 0 0 u a 4J >, M O CO CD (J CO u to cu 3 i— I cd > c cfl 0) E OB 3 < I I I I I I in i—i rO O CN O o • cn cn m oo •jCN CN cO I I I I i I c o o CN •H -= o o m o a CN a) CO • tn C X qua 0) N -a t— i •H c cn to co ^— >-. to i— 1 to o •H u E o B 3 U ca CO 4-1 CD IM H *J T3 X! T3 c c C C OJ O CO CO &, •H X 4-J cu (1) u E E CO o o kl u u rH CD c C 4-1 H 1 — 1 O In CD o CC to -a H O JZ V to 3 O £ O V4 o XI E 3 C m o c to a >— i o x: cu to 3 o -C c CU CJ u Dl,

PAGE 67

households consuming seafood at home, 29.9 percent resided in the Northeast region. The third category, labelled limit observations, provides the mean value of all variables for those households reporting a zero level of at-home consumption of seafood during the one week survey period. On this basis, the information contained in Table 3-1 suggests that of those households reporting no at-home seafood consumption 19.5 percent resided in the Northeast region. The final category in Table 3-1, labelled proportion consuming, gives the value of the proportion of households consuming seafood during the survey period associated with each of the binary variables. For example, 61.2 percent of the households residing in the Northeast region of the United States consumed seafood during the interview period. Approximately one half (50.8 percent) of the 10,689 households included in the analysis consumed seafood at home during the survey period. Among households consuming seafood at home, average expenditures and consumption of seafood equalled $2.93 and 1.91 pounds, respectively. For the total sample, average at-home weekly consumption of seafood was 0.97 pounds valued at $1.49 or approximately one-half the volume and value estimated for consumers only. Placed on a yearly ^Ljt basis, at-home consumption of seafood by an average household thus equals just over 50 pounds, or about 18 pounds per capita assuming an average of 2.75 members per household. This value equals about onehalf of the approximately 35 pounds (round weight) annual reported per v capita consumption of seafood in the United States during the 1975-77 period With respect to region, households in the Northeastern region of the United States (XI) had a higher probability of consuming seafood at

PAGE 68

58 home than did those households in either the Northcentral region (X2), Southern region (X3), or Western region (base). Over 61 percent of the Northeastern households consumed seafood at home compared with 44.9 percent, 45.3 percent, and 55 percent of the households residing in the Northcentral, Southern, and Western regions, respectively. Little variation in the proportion of households consuming seafood at home was observed across seasons. Though a slightly larger proportion of households consumed seafood at home during the summer (X7) and winter (base) quarters than in either the spring (X6) or fall (X8) quarters, the differences would probably not be significant if a statistical test were to be conducted. An examination of the different family life cycles (X9-X17) revealed that households with children were more likely to consume seafood at home than those households without children."'" On average (weighted), 55.7 percent of the households with children consumed seafood compared to 45.3 of those househoolds without children. Similarly, at-home seafood consumption tended to increase in probability with increased family size (X24) which generally reflects increased number of children. The race of the household respondent (X18, X19) appears to be an important consideration in determining the probability of at-home seafood consumption. White households (X18) had a significantly lower observed probability of at-home seafood consumption than that of either Black households (base) or households of other ethnic origins (X19). It was assumed (though not verified) that elderly households (X17, base) had no children living at home.

PAGE 69

59 Households who caught fish for their own use (X21) had a much higher probability of consuming seafood at home than did households not reporting fish catches (base). Among those households who caught fish for their own use, 57.3 percent reported consuming fish at home during the interview period. This figure compares to only 48.8 percent among those households who did not catch fish. The employment status of the meal planner (X22) and the sex of the meal planner (X23) had the anticipated effects on consumption of seafood at home. Employment of the meal planner leads to a slight decrease in probability of at-home seafood consumption. A female meal planner (X23), on the other hand, appears to greatly enhance the expectancy of that household consuming seafood at home. With respect to the continuous variables included in the models (X24, X26, X27, X28, X29), differences in the mean values between consumers and nonconsumers for a given variable should indicate increased (decreased) probability of at-home seafood consumption with respect to that variable. For example, the average family size (X24) of seafood consumers (3.153) greatly exceeds that of nonconsumers (2.733). Similarly, the mean values for consumers with respect to education (X26), number of guest meals (X27), and annual income before taxes (X29) exceed the mean values for nonconsumers. Mean values for continuous explanatory variables in the analysis are larger for consumers than nonconsumers with the exception of expenditures on meals away from home. Large differences between the mean values of the nonconsumers and consumers are especially apparent with respect to family size (X24), number of guest meals (X27), and annual income before taxes (X29). Thus, one would expect the probability of at-home seafood

PAGE 70

60 consumption to increase significantly with changes in the value of these variables. Regression Estimates of Total Seafood Analyses Notwithstanding the general usefulness of the descriptive statistics just presented, there are several inherent weaknesses associated with these types of statistics. The foremost weakness associated with these types of statistics is that they do not control for confounded effects among different variables. Thus, one cannot separate the effect of one exogenous variable from that of another when examining changes in the dependent variable. For example, the relatively low probability of seafood consumption among households without children when compared to those with children may be due to some factor such as larger expenditures on meals consumed away from home among the former group of households. The Tobit regression parameters presented in this section can be considered as partial effects in that the confounded effects among exogenous variables have been controlled for. Results of the Tobit analysis relating to total at-home seafood 2 consumption are presented in Tables 3-2 and 3-3. The first column gives a listing of the variables used in the analysis. The second column in each table gives the Tobit parameter estimates associated with each of the exogenous variables. The asymptotic t-values associated with the parameter estimates are presented in the third column, The relatively large sample size employed in this study should assure that the asymptotic t-values are representative of the true values. 2 The Tobit model used for this analysis was developed by the Rand Corporation and is referred to as LIMDEP. Documentation of the model is given by Phelps (1972).

PAGE 71

61 •H oo x c CO. •H c 4-1 4-1 •H 0 i— 1 3 , 4-1 -a •H o i — 1 4J H H U XI N m CO >^ a, xi Pn o W ft u •H 4-1 o o H 4-1 4-1 & co S V>> i 4-1 < M co G a 4J 4J CJ CO E e *r co H B Ih 4J a CO Pu a) •H X >, tn O CO a> 4-1 CO CJ o WON CI H sf mn >n i ... I o o o I I O in o 00 t— t i CO v£) on — i d d I I o h on cn cn o o o I I c o H OO CJ IM =3 o i — I CO I CN o X o d d on o vo on m cn d d oo r-cn a\ cn r-i r- on m -3oroo o CN on oo •ooo oo co o CN 1— 1 X • — X QJ I 1 qj CO c CO on CO xm CO U 1-1 X CO 4-1 X X ^ s X C H w ^ c o u CO o c •H c 0 CO CJ c *J rH CO (-1 QJ o u CO CO x 4-1 X X X 11 N u u o 4-1 1J 4-1 rH 4-1 3 E U 3 CO c C X c O o O 0) co m o o o d d d HIO-J O r-~ r-3in r- co vo r -4 Ol o d o i -300 -vf in cn i — i
PAGE 72

62 •H ^ CD C "H CO 3 CN C ^ — ct) CO fi, J3 c U •H p: ^ — v tr; ^ ta 3 CO C E CO 0 jC 3 u <4H >, HI -o H (U r-1 4J •H T U X N) a CO w a. xi Eh X 0 w 3, u •H 4-> O 0 •H *J — co E ~ >> i •p < 3 n o 3 4-1 JH a ca E E T cfl •h a u 4-1 3 w Oh o 3 o CO cu co -* 00 1— 1 o> o o oj OC x c CO r— o> c vO CT o CM x o c UO CO CO !> CO o i — i lO r— 1 o LO — 1 CN vO Ol o CO l i C^ — 1 CN CO co i — : o o — i — i I I C CN o 1—1 x CO r^X CN 00 rlO CO. , c •H — 1 3 Ol 3 3 3 3 3 o u CO — i 1 r— •H 3 ma 3 •H 3 CO 3 3 3 X T3 X X a x CO CU O CO l-l s 3 r — 1 •H cu CO O CU o Ih Ih 3 "3 CO Ih . >, rH I — 1 i — i l — i i — 1 cu 3 3 X a X — rH O 00 •H CO H CO H CC H — 1 c rH c — i 4J 4J Ih Ih X c J3 c J=. c x: 3 x: a "3 "3 •H -3 -H CU 3 OJ 3 u 3 o 3 u 3 u -3 3 "3 3 -3 ? "3 a "3 ca o O c O •H J •H i — i rH 3 >~ >H s: s s: C3 Ch o CE

PAGE 73

63 co 4-1 •H /-^ > 4-> T3 H f-H ^ I i i to U u CO 3 o 4J i-1 X) X) 3 co r— 1 CN E S -H s* 3 H S CN CN CO 3 4-1 • a en r— 1 o 3. a 1 1 co I o I CO O I r-1 I m m <-< 4J V 3 H X 3 H CO X CO X 4-1 H CU 3 3 CO 3 O U :s o 33 u 5S 3 >H 33 3 cu 3 cm OS CJ 3 3 3 3 3 3 3 E 3-1 c 3 3 3 E >i O 33 ON f-H I O I • I o I co nO m i o I o I in co o o r~oo st vO I ON I • I O I O CN I O l CN CU CN CO X CO w X co
PAGE 74

64 n c 1 — i a .e c u H s ^ — 3 CO CL, V V C X o o w . I (/) 4-1 < S-. cfi 0) 0) 4-1 4-1 0 CO E E h 3 co cn Oh , o CO CD 4-1 co U cn oo in ON i o I o r-sisr CM I st I • I CN CN CN o on CM vO O cn 00 O 3 in r-. sf o> CN 00 rO CN 00 cn rCM O m CC O m rsf CO CC CA o> 0^ CO 00 vO St 1 o i — i go LO 1 st c o — 1 1 o a o 1 o c • • r4 OJ c a r CO CM • • H X u Oh OJ -, CO o CO cr 4-4 u O E H H 4J 4-4 M CO CO 4-1 E CD CO •H o o u CD cn O X fr 2 E H (— 3 >r a z: QJ CO TJ a CO fr. W CJ

PAGE 75

65 to 4-> •H /-n CD C -H M 3 N d w CO OOCj-i £ c U -H E •a d a> co ou c x o w u o o c X CO X c CO H c j-j 4J O — 1 qj a. CO CO X a CJ c u a) u CO ^ 4J x: w U CD CJ CO CJ 00 C E X CO 0 W si u u <*H X CO CO CX CD o CM o o o I rcO c o o i >> 4-1 T3 -H 0) rH ^ ON 4-> -H 'H in O XI N 4-1 4-) V m a. co co E u — co i i CO 4-> , c O CJ CO CO rH -H x; X In CO CJ •H -in CO CO CO O E u E C E 4-1 rH rH CO Sr CO CO o o oo rH rH X 0) s-i UH ^-s u Jh CN QJ O O 4-1 CN <4H X U 13 X) a T3 CO o C c C X T3 T3 QJ o CO co ^ CN CO CO QJ C C M •H 1 >> > U X CO CO CO -> QJ CD CD C~l co co to CD u E E N o rH QJ 3 cr CO O O -H CJ oo <~* CO rH E O O ca M U U CO rH CD r-H rH O O X X QJ a c X 4-1 w CO Q u H H J-j H rH CO CO CD c C H cj> S rH H

PAGE 76

66 CO 4-1 H v o c •" DO 3 tX) C — CO £ n cj •H O CO S-, 0 4-1 c X •H u o W X cu o ^ CD o, w X oo <" on oox 1 c 3 •H u -H U XI NI CU CO a.x> X o ex u •H 4-4 o O -H 4-> 4-) O. CO e u 1 CO 4-1 < u co (1) QJ 4-> 4-) cu ca E E •H CO -H U 4-) co co >> kl o DO CJ 4J o U < CM CJ CN o o> m i cc o co CN CO I cn co C X O 14-1 CO cn QJ 1h ca c rH (0 i — 1 4J a o w o c 4^ o o o 3 n CO II CJ M — I O c c •< 4^ u a 4-> CO c o u cu x: u o X) o 4J c 3 O u u 3 c 140 m CU -3 CN CU C > 1 ca -3 o 4J 0i E rH ^ CO CU 3 01 ~ c 0) — — 1 a X! 00 u ca •rl ca H 3 01 cr 3 (0 > •H X s~ CU c w 1 — i o rH 33 X! CO C 4-1 CO <+j x: 0 c 0 •H H c CD u -a CJ ca CU E Ih 4-1 o 3 CU J H .n c U 4-) •H o cn 4J CU ca co -3 3 4-i ca Cfl X 4H E 0 O 1* •-I CJ-J cu Xl O CO 4J 3 4J U CJ U H •H CJ a r1 o. eg 4-1 u Oh > a CO CO o cu CJ c 4-> x: H O as i o

PAGE 77

67 o •H o cu 4H CO CO c •H CU CO co tsi CO x: a -~ o •H Ch T— 1 CO 13 1 1 c 4-1 O o o0) O w MH ft H o X CO ro w C c O o £ •H CO 4-) ft H •iH cn 00 X c rH C o CO -H C o 4-> 4- •H O rH 4-) 3 CUX1 cn CO-^v H -a cu c o* 4_> 01 Sh CO ^ c 4H x: w CO U CJ 3 QJ CO o" ft c E X CO C "0 W x: u rH o •~i o — cu CO >-, 3 4-1 o -a •H _c 0) rH ^ 4-1 •H -H >, o XI N — 1 cu CO *4 a. xi tu cu X o CD w Sh |3 ft o u cn H H 4-> c cn o •H >, 4-> 4J — i a. to CO E Sh c >-. 1 CO cn 4J < 4J •H X> o u to H co a 4-1 Sh cu CO o e S -r sh CO •H CQ u 4-) cn CO tn u Cu, a) o CO o 4-1 CB CJ lO cn o H CC 0) K co oo on co o o vO rH o Q o o I oo>o in nO CN CO rH cu / — s CO C CO CO tn >, m CO rH U X ca 4-1 X XS CU 4-1 X! c •H w 4-> C o CJ cn cu c H c o CO CJ M C 4J rH CO rH CD OJ Sh CO CO XI 4-1 x: x: XJ aj N Sh rH GJ 4-1 4-1 4-1 4-1 •H 4-1 3 E U rH 3 CO C C Xi C o o O QJ cd CU 3 C on 3 X) CJ CO z r>NO rH rH CO cn CO CO rH r-* cn O rH d O -1 ON Q O O l m co co h CO CN CO NO m no co co -o -h E Sh E ft 3 co co Sh OJ

PAGE 78

68 o CC a. C E X CO o x 1-1 U X T3 -H CD • — I — J-l -r-l -H O -a tsi ai co w a,x fan X o W S-l u •H 4-> O O -H O. CO e u CO 4-J l-i CO CD 01 4J 4-1 CD CO e e CO -H l-i 4-> CO CO cu ai •H OQ >-> 1-1 o CO CU IJ 3 a in co u u CJ IH H o Xi CJ en o 33 c ri — i CC cn o o c i c o O o o o o O O o c c o r~ cn CO 00 o — i CC m CTi in CN \0 cn co CN o r oo i — i cn o CM cn m — i o o rH CN 0 m o a CN i oo sO rH r ro oo rH oo t-i m vQ m o cn co rCN m m on 4 f— ( o cn •H — 1 3 CN 3 3 3 U 3 -3 o> 1 — | > — 1 — t H c CO C -H 3 3 a> 3 3 x -3 X X -3 CO CJ E 3 CO In E 1-1 — i H CJ V — CU — CJ O -3 ~ CO' l-i rH H •H -3 -3 -3 — rH a r— c Ih SO c 1h c 00 H l-l c 1 — 1 a rH 01 H a> H H CO C cj a c O 1-1 CJ 00 H CO •H oox 00 X to E H 1-1 CO 3 •H CO 3 X a X 3 u CO U M a E -a w -3 G — u 3 >^ 1 — i CO 1 — 1 rH rH cu CJ CJ -3 0) J3 rH H 00 H H C-0 H CO H ^3 0 — '. O i-i u tH u H C X c X! C X C -a X) H -3 •H 0 CD 3 u 3 O 3 u 3 U — a -3 "3 3 -3 O O 0 O •H •H H rH 1 — 1 >H >^ >H >* g Ed CJ

PAGE 79

69 to *j •H Q) C CO 3 ^ C i CO OON1 SL\ C ^ B •a 3 CD to — 4-1 C o o cy O rH 4-1 3 cu cdxi co oo'-l CO O CD O 0) OO ac S x to o W £ u u 4H >, 4-1 -a -h CD T-H ^"N 4J -H U ^1 N CD CO CU.O X o PJ u cx u •H 4-1 O O -H •M 4-4 a. co E IH ^ I CO 4-1 < 1-1 CO CD CD 4-> 4-1 CD CO EE-'-' CO *H CQ U 4-1 CO CO PCD >, l-i o CO aj U ra CJ m m CN .-H o" o I I CO CD co oo oo >— I r — f CO -H co o in HCNO IT) vO 1 CN 1 CO r-- 1 CN i CN 1 o 1 rH 1 o I 1 O l r— 1 1 o 1 1 • • Sh CD • • c CU c M CO • 3 1 — 1 0] a, • • C E rH C co 0 a ai 4-4 c; (0 Ih E a f — > O o • X 0) 13 4-1 14-1 a. CO [fl O O W i— H i — i CO o X! cu X c CD 4-i >> H CO •H co X! IT 0 CO CU -C — i CU CU o CO CD o H QJ o u o cc u z 3 >H CX >< CO CU C £ DC U

PAGE 80

70 co 4-1 •H (1) C 00 3 C *H 03 CON -C C w U -H [i< -O 3 cu co 4-1 c u o c l) ex W W C o s ia X T3 4-1 •H 00 X C •H C 4-1 -H rH 3 CDX CO CO-— CU C O* x w U B U f qj oo x co o W -C u O <+H 4-1 •O -H 0) H ^ 4-1 T-4 T U JO N P.X [h X o W U cx 4-1 O O -H 4-> 4-1 CO E in I CO 4-1 < U CO 0) cu 4-1 4-1 X CO ON m i O vO i-H ON r-~ oo rsl O I m rcn O I ON g o CN o CN o ON CN o CN] X) o o c c 3 o 55 in c c ON ON o o o CN rH u — V ft cu , CO o 3 cr u Ih o E rH rH 4-1 -C 4J CO 3 3 u r CU cu 53 •3 3 on En w O I s CN X 3 CU s o Ih CJ XI e 3

PAGE 81

71 co u •H CD C CO 3 ^ C -H co co M j= c ^ u -h b c T3 3 O O -H 4-> 4-> Q. CO e u CO 4-> < 1CO CD 0) 4-1 4-1 CJ CO E E i-l eo -h oa (-1 0-1 CO CO QU >. u o 00 CJ 4-1 CO u co o CM C c d i 0> m c 3 C I 0> o rco qo st i on rH o d i en I -J I co CO o On CN 1 CN o Q O O l ON — I CO rH oo . O CJ CO CO rH JS X r4 M co o •H CO CO eo O E -. 14-1 XJ o o CO CO to CO a C c H eg CO ca u o a V CO co CO CO CJ E E N rH 01 cr CO O o H CO t — i E O o Cfl >m u u CO rH O o JC si CJ c c 00 Q u E-i H 4-1 n rH CJ c c £ KH rH

PAGE 82

72 CD c 00 3 co OON X! c w u •H -a vO o co /-->. v — • 4-1 c U o o•H o CJ w X a w CO X oo co c^ c amo H OOX rH c CO H C 4-> 4-J -H O ( — i 4-1 ~ CDX co 00 '"n / — s TJ a c c •H in O XI CU CO X o w ^ o. 4J O O -H 4-1 4-> QCO E U >> I CO 4-i < l-i CO CD CD 4-1 4-1 CD CO s e CO -H (-1 4-1 CO CO — 01 U o DO CU 4-J a CJ CN u < Z • co -a H O X! cn c ii O cj D O w I— 1 c -JCD 1 CD 1 CN X) 1 CO • CD o > 1 CO _C o 4-1 CO E i—l CO CD CN 4-J CO a* CD rCD -a — i a X m CO u d i •H CO •H a >>4 co a 1 co CO > •H X a o CO iH o rH X cn CO C 4-> 00 a m >W X o c 4-1 CN o •H 1 CO •H ts 4-J co u a CD td CD E u 4-J CD ca CO CD o CD CD C 4-1 00. X! XI H o H rH z CO Xi u

PAGE 83

73 In column 4 of each table, the vlue of the cumulative normal distribution function associated with each variable is presented. The value of this function varies with each of the discrete variables due to variations in E(Xg|X-j). For discrete variables in the models, the values provided for the cumulative normal distribution function are interpreted as the expected probability of observing a positive level of the dependent variable given the occurrence of 3^ holding all nonmutually exclusive variables at their mean levels. Mutually exclusive variables are set equal to zero. "'For example, to determine the expected probability of seafood consumption by average Northeastern households, the value of XI is set equal to one, the values for X2 and X3 are set equal to zero, and the values for all remaining variables are set at their respective means. For continuous variables, the value provided for the cumulative normal distribution function represents the probability of observing a positive level of the dependent variable given the mean values for all exogenous variables. Of course, the value of the cumulative normal distribution function varies with changes in the level of exogenous variables^ Multiplication of the appropriate parameter estimates, given by those values in column 2 by their respective expected probabilities of occurrence (column 4) provides the unconditional or total expected change in the dependent variable due to a change in X^. These estimates are given in column 5 of Tables 3-2 and 3-3. The unconditional or total effect of a change in expenditures or quantity consumed with respect to a change in the independent variable X^ can be decomposed into two parts. The first part represents the change in the value of expenditures or quantity consumed among existing consumers weighted by

PAGE 84

74 the probability of being a consumer. The second part represents the change in the probability of being a consumer weighted by the expected value of expenditure or quantity consumed among consuming households. The values for the first component, the conditional expected change in expenditures or quantity consumed resulting from a change in as derived in equations (2.20a) and (2.20b), are provided in column 6. Thus, by definition the values for the second component of the total or unconditional change can be obtained by subtracting the values in column 6 from those in column 5. The parameter estimates associated with the at-home total seafood expenditure and consumption equations appear satisfactory and reasonable based on several criteria. First, the estimates, with few exceptions, conform to either theoretical expectations and/or results 1 of previous research. Second, the relative magnitudes of the estimated probabilities of observing a positive level of seafood consumption associated with each of the categories of binary variables are for the most part in agreement with the observed probabilities given in column 5 of Table 3-1. Finally, a high proportion of the parameter estimates, for a cross-sectional study of this nature, were statistically significant (at a 10 percent significant level) in explaining weekly household consumption of seafood at home. More detail is given to these factors in the discussion of the individual explanatory variables. Before providing an in depth discussion of the results, a few general findings are discussed here. First, with few exceptions, the results associated with the expenditure model were found to be consistent with the results pertaining to the quantity consumed model. Second, as indicated by the results of both models, the change among

PAGE 85

75 consuming households resulting from a change in consistently averaged from 30-40 percent of the total change. This implies that approximately 35 percent of the change in at-home consumption of seafood with respect to a change in is due to increased/decreased consumption of seafood by those households currently consuming seafood as opposed to entry or exit among households. Thus, approximately 65 percent of the total change is attributable to entry/exit into (from) the at-home seafood market by households. This finding has significant implications to the seafood industry and its support groups who would like to know the relative merits and associated costs of increasing ( at-home seafood consumption by enticing new consumers into the at-home seafood market as opposed to increasing consumption among currently consuming households. Region Total weekly expenditures on seafood consumed at home were estimated to be highest among the households residing in the Northeastern region of the United States (XI) (Table 3-2) which is consistent with the results presented by Perry (1981) and Capps (1982). Similarly, total weekly quantity of seafood consumed by households residing in the Northeastern region was estimated to be higher than that for other regions (Table 3-3). Based on the information provided in column 5 of Table 3-2, the total expected expenditures on seafood consumed at home for a household in the Northeastern region were estimated to exceed those for a household in the base region (West) by $0.45. Similarly, expenditures among Northeastern households were estimated to exceed expenditures among North Central households (X2) and Southern

PAGE 86

76 households (X3) by $0.83 and $0.61, respectively. The information provided in column 5 of Table 3-3 suggests the total expected weekly consumption of seafood by households in the Northeastern region exceeded that of households in the North Central region, Southern region, and Western region by 0.52 pounds, 0.30 pounds, and 0.30 pounds, respectively, ceteris paribus. The relatively high estimates associated with at-home seafood expenditures for a household in the Northeastern region compared to households in other regions of the United States are the result of two factors. First, the estimated probabilities of a household having positive weekly consumption expenditures and quantities, given in column 4 of Tables 3-2 and 3-3, exceed those associated with any other region. Second, the expected expenditures among consuming households residing in the Northeast exceeded those of households in other regions, as noted in the last column of Tables 3-2 and 3-3. Given the estimated differences in at-home seafood expenditures among households residing in the different regions, establishing probable causes for these differences may be beneficial to the seafood industry and its support groups. Traditionally, the Northeast region has had an extensive fishing industry. This factor, in conjunction with the close proximity of most of the Northeastern States to the ocean, has resulted in a relatively steady supply of fresh seafood to the households in this region. Transporting fresh seafood products from coastal states to the inland states, such as many of those located in the North Central region, is risky and expensive. 0'Rourke (1977) categorizes the U.S. seafood marketing system into two distinct segments. The first segment, defined by 0'Rourke (p. 239) as the

PAGE 87

77 coastal fisheries "... are exploited by small, ill-equipped or parttime fishermen, delivering their product to dockside warehouses for fresh distribution to communities within a 50-mile radius." O'Rourke further claims that the New England fisheries deliver much of their specialty catches in this manner. Because of this factor "... large areas of the continental U.S. have below average consumption of many fish and shellfish species." The second segment which may account for two-thirds of the U.S. consumption of seafood is comprised of large, highly capitalized canners or prepackers selling standardized, breaded, and heavily promoted products through nationwide retail chains or institutional outlets." The relative unavailability and expense of certain fresher seafood products in the North Central region probably explains to some extent the relatively low estimates of at-home seafood expenditures and consumption in this region compared to the other regions of the United States. Furthermore, one would expect regional differences to be relatively large for fresh seafood products and somewhat less for the more processed seafood products sold either frozen or canned. The validity of this hypothesis is examined in the next chapter when results pertaining to the seafood product forms are analyzed. Given the differences and probable causes for these differences as relating to at-home consumption of seafood, what are their implications to the seafood industry and its support group? Most obvious and probably the most important from a policy standpoint lies in the realization that the North Central region provides a relatively large and untapped source by which to increase national at-home seafood consumption. To this extent, it may be beneficial to the seafood /

PAGE 88

78 industry and the support groups to look for ways of reducing costs and preserving the freshness of seafood when transporting fresh seafood to the inland regions of the country. Unfamiliarity with many seafood products among North Central households may also help to explain the differences in the at-home seafood consumption patterns across regions. Thus, promotion aimed at familiarizing these households with the different available products may prove useful in the short run. In the long run, as the regional structure of the population shifts due to increased population mobility and the transportation of fresh seafood products becomes more economically feasible due to improved methods of preserving the freshness of seafood, regional differences in seafood consumption will probably decline naturally. For example, between 1950 and 1980, the proportion of the U.S. population living in the Northeast and Midwest sections of the United States declined by 17 percent and 12 percent, respectively, while the proportion of the U.S. population residing in the South and West increased by 7 percent and 44 percent, respectively (United States Department of Commerce, Bureau of the Census, 1984). Additional declines in the Northeast and Midwest sections of the country are expected until at least the year 2000. With these demographic shifts in population should come an exchange of knowledge among households concerning different types of seafood and methods of preparation which may eventually lead to the disappearance of regional differences in at-home seafood consumption patterns. Urbanization Households in the central city (X4) were estimated to have higher weekly consumption of seafood at home than those households in either

PAGE 89

79 suburban areas (X5) or nonmetro areas (base), ceteris paribus. Similarly, households in suburban areas were estimated to have higher weekly at-home consumption of seafood than those households in nonmetro areas, ceteris paribus. Central city households had total expected weekly expenditures (quantity consumed) equal to $0,564 (0.355 pounds) in excess of those households in nonmetro areas. Households in suburban areas had total expected weekly expenditures (quantities) equal to only $0,231 (0.190 pounds) greater than households in nonmetro areas Since proximity to the coast was a major factor leading to the development of many of the larger cities in the United States, households in the larger cities may have greater access to a larger variety and quality of seafood than those households in either suburban or nonmetro areas. For example, New York City, Boston, and New Orleans have major fish markets acting as central locations from which distribution of seafood products to other localities is coordinated. Due to a decline in the accessibility of moderate cost quality seafood as one moves away from the distribution centers, one would expect that the probability of a household purchasing and hence consuming seafood to decline in relation to distance from the distribution center. As noted in column 4 of Tables 3-2 and 3-3, the estimated probabilities of observing a positive level of expenditures and consumption of seafood for a household residing in a central city area exceed those of a household residing in a suburban or nonmetro area. These estimated probabilities are consistent with the observed probabilities of observing a consuming household given in column 4 of Table 3-1.

PAGE 90

80 Season At-home quantity of seafood consumed was estimated to be greatest in the summer quarter (X7), followed by the spring quarter (X6), fall quarter (X8), and winter quarter (base). Weekly expenditures, however, were found to be much more constant across quarters with no statistical differences noted for any season, as judged by the nonsignif icance of the asymptotic t-values associated with the parameter estimates (column 3 of Table 3-2). A quantity change not reflected by a corresponding expenditure change suggests that price must also be changing. During the one year period in which the survey data used in this analysis were collected (April 1977-March 1978), the price for seafood consumed at home as measured by the Consumer Price Index increased by 7.6 percent (United States Department of Agriculture, 1983). Given the 7.6 percent increase in the price of seafood consumed at home, the question arises as to why the estimated weekly expenditures on seafood consumed at home did not show a similar increase. In fact, weekly expenditures were estimated to be lower (though only marginally) in the later half of the survey year than in the first half. A possible explanation put forward to answer this question is that households may have reacted to the increase in the price of seafood consumed at home by reducing quantities purchased leaving weekly household expenditures on seafood unchanged. This would imply the quantity elasticity with respect to price must equal approximately unity.

PAGE 91

81 Household Life Cycle The household life cycle category (X9-X17) appears to be very useful in explaining weekly household at-home seafood consumption as judged by the number of statistically significant parameter estimates. Household life cycle estimates suggest that the composition of the household, independent of size, explains both expenditures and quantities of seafood consumed at home in a rather systematic and logical manner. Furthermore, the manner in which consumption of seafood can b explained via the household life cycle is as expected, given the current understanding of the at-home seafood market. For example, there appears to be a general tendency for increased consumption of seafood associated with the maturing of the household. Households wit the household head less than 35 years of age (X9-X12) consistently consumed less seafood than households in more mature life cycle categories. Households with the household head from 35 through 64 years of age (X13-X16) generally consumed less seafood than households comprised of an elderly married couple (base). Given that the difference in household size between that of an elderly individual (X17) and that of an elderly married couple (base) has been accounted for by the variables representing household size (X24 and X25), the estimated difference in at-home seafood consumption between these two groups of households may represent differences in eating habits. For example, elderly individuals may not wish to cook only for themselves, especially those items requiring any amount of preparation time. This would preclude them from consuming all but canned seafood which takes minimal amount of preparation before being suitable for consumption.

PAGE 92

82 Overall, households comprised of young adults with children (XI 1X12) had the lowest at-home seafood consumption of any of the life cycle categories, ceteris paribus. A representative household comprised of a young single adult with children (Xll) was estimated to consume on average 0.144 fewer pounds of seafood at home than that of a household comprised of a young single adult without children (X9) with lower corresponding expenditures equal to about $0.21, ceteris paribus. Analogously, a household comprised of a young married couple with children (X12) was estimated to consume 0.0753 fewer pounds of seafood with corresponding expenditures of about $0.12 less than that of a household comprised of a young married couple without children, ceteris paribus. Though these differences may appear small it should be kept in mind that the analysis is based on a one week period. Extrapolating to a one year period, a household comprised of a young single adult with children is expected to consume almost 7.5 fewer pounds of seafood at home than a young single adult without children, ceteris paribus. With respect to the life cycle categories pertaining to middle aged heads of households (X13-X16) the results do not appear to provide any systematic trends. Households comprised of a single middle aged adult without children (X13) had lower consumption and corresponding expenditures than did those households comprised of a single middle aged adult with children (X15). On the other hand, households consisting of a middle aged couple without children (X14) tended to have higher consumption of seafood than did those households consisting of a middle aged couple with children. The results pertaining to the household life cycle can be used to design and implement a seafood promotion/marketing strategy.

PAGE 93

83 For example, the relatively low estimates of at home seafood consumption among those households categorized in the younger life cycle categories establish the premise that seafood promotion/marketing targeted towards this segment of the population may provide the seafood industry and its support groups with greater net returns than that of targeting households categorized in more mature stages of their life cycle. Of course, the validity of this premise depends on the relative costs associated with promoting seafood to households in the different life cycles relative to the returns per dollar expended. However, before targeting this group of households the reasons why this group of households exhibits a relatively low level of seafood consumption needs to be addressed. A couple of reasons can be offered to help explain these results. First, the meal planner in the "younger" households probably tends to be somewhat less experienced at preparing meals than the meal planner in more mature households. Due to this factor, these meal planners are more likely to avoid cooking a meal which involves any much of preparation. Seafood, especially fresh seafood, has a reputation for being difficult to properly prepare. Thus, seafood may not be prepared and consumed as often by "younger" households as would be expected among more mature households. A second explanation is specific only to those younger households with children (X11-X12). This segment of the population exhibited the lowest weekly seafood consumption among the different categories in the household life cycle. The children in this household group will generally be of a lower average age than in households categorized in the more mature life cycles. Hence, these children will place a larger burden on the meal planner's time than would older children in

PAGE 94

84 the more mature household life cycles. A related factor pertains to the hesitency among parents of serving seafood to younger children out of fear that the bones in the fish may injure the children or that the children will have problems eating certain types of seafood (such as shellfish). All of these reasons suggest a marketing strategy aimed at promoting highly processed/ready-to-eat types of seafood products to this segment of the population. Additionally, one would expect to find fresh seafood consumption, which requires the most preparation time and with which bones are most frequently associated, to be lower among younger households, especially those with children, than among other households. However, for frozen and canned seafood for which preparation time is minimal and bones are generally not a problem, one would expect to observe little differences in consumption patterns among the households categorized in the different life cycles. This hypothesis will be examined in greater detail in the following chapter. The characteristics and composition of the American household is in a continual state of transition. Knowledge of this transition in conjunction with the information provided by the results pertaining to the household life cycle category can further aid the seafood industry and its support groups in the planning stage of a long term marketing strategy The first factor the seafood industry and its support groups may wish to consider when planning a long term seafood marketing strategy is the changing age structure of the American household. Table 3-4 provides some statistics on the age distribution of household heads for selected years. The statistics suggest that younger households (those with household heads less than 35 years of

PAGE 95

85 CN I CO r-. 00 | • • • • ON 1 o 00 O 1— 1 1 > o 1 O in m 3 00 j • • • O T3 ON [ i — i CO o H CU l—t I ci 0 j — r-H | CU 1 CO CO 1 3 I (0 U m j CN i— l c O I > • cu ON j o o O — ( CN cn CN CU 4-1 n> oo c X CO Ih cu 4-1 CO u >^ Qi u y-i XI >> cu o Oh -a o CO CN m 3 CO r~I • • • CO CU ON 1 m m ON CU JS r-H 1 CN m i— 4 u j 3 -o CQ rH I o r| (1) CU o co I 3 m 1 m m o cu O vO I • • • x ON 1 o g i— 1 1 CN o r-H O I u o 1 1 UH c 1 O o 1 •H 1 4-1 4-> 1 e 3 o 1 On cu XI vO j • • • S •H ON j m o cn 4-> Ih i-H 1 CN o i — i u 4-> CO CO Q. •H cu -o Q — co oo cu CO 4-1 4-1 co c 4-1 CD CO u -3 cu CU Dl, 1 4-1 CD •H CO — CO c CO CO Ih CO o CJ Ih eg u 1 m X! XI CO 0) CO CD >i cu >4-l T3 w CL) o rH o i—i o m m OS XI cu X ro 1 CO 00 m A| o H < V (0 c/)

PAGE 96

86 age) and older households (those with household heads greater than 65 years of age) have become increasingly important segments of the population over the past two decades at the expense of the middle aged households. The results of the Tobit analysis suggest, however, that at-home seafood consumption among younger households (X9-X12) and households comprised of an elderly individual (X17) tends to be among the lowest of any household life cycle category. The seafood industry and its support groups thus may want to consider these age structure changes when planning a seafood marketing strategy. A second factor the seafood industry and its support groups may want to consider when planning a long term seafood marketing strategy is the growing proportion of households in the United States that have no children. In 1960, 43.1 percent of all households in the United States had no children of their own under 18 years of age. By 1982, this proportion had increased to 49.2 percent (United States Department of Commerce, Bureau of the Census, 1984). The results of the Tobit analysis suggest that, at least among certain age groups, weekly at home seafood consumption differs depending upon whether children are present in the household, certis paribus. A final factor the seafood industry and its support groups may wish to consider when planning a long term promotional strategy is the growing proportion of single adult with children households in the United States. In 1970, 84.9 percent of all children under 18 years of age were living with both parents. By 1982, the proportion had fallen to 75 percent (United States Department of Commerce, Bureau of the Census, 1984). A single parent with children can be expected to have more constraints on his/her time than would be the case if both

PAGE 97

87 parents were present in the household. Thus, households in this group should have a higher demand for seafood products for which little preparation time is required relative to households in which both parents are present. Race Black households (base) had significantly higher at-home seafood expenditures and quantities consumed than did White households (X18). Similarly, Black households were estimated to consume greater quantities of seafood than households of "Other" ethnic origins (X19) but their weekly expenditures were not significantly different given the insignificant t-value in column 3 of Table 3-2. Total weekly at-home seafood consumption by a typical White household was estimated to be 0.78 pounds less than that of a similar Black household while expenditures by that same White household were estimated to be $0.86 less than that of a similar Black household. Among households of some "Other" ethnic origin, weekly at-home consumption of seafood was estimated to be 0.42 pounds less than that of a similar Black household while expenditures by this group were only $0.21 less than that of a similar Black household, certis paribus. The estimated probability of consuming seafood at home was substantially lower among White households than among either Black households or households of "Other" ethnic origins. Among White households, the estimated probability of consuming seafood at home was 0.42, compared to 0.61 among Black households, and 0.53 among households of "Other" ethnic origins (column 4, Table 3—3). The large differences associated with the probability of consumption among the different

PAGE 98

88 races probably reflect cultural factors leading to differences in tastes and preferences. The proportion of Black households to that of White households has been gradually trending upwards. In 1970, Black families represented 9.5 percent of all families in the United States. By 1982, the proportion of Black families to that of the total had increased to 10.5 percent (United States Department of Commerce, Bureau of the Census, various issues). This represents more than a 10 percent increase in the proportion of Black families in just over a decade. Households of Spanish origin, though representing a relatively small proportion of the total number of households in the United States, also represent an increasingly important component of the population. Representing approximately 3.9 percent of the total number of families in the Unite? States in 1970, the proportion of households in the United States of Spanish origin has increased almost 40 percent in just over a decade and represented 5.4 percent of the total number of families in 1982 (United States Department of Commerce, Bureau of the Census, various issues). As the statistics tend to highlight, though the proportion of Black households and those of "Other" ethnic origins still represents small proportion of the total number of households in the United States, they do represent an increasingly important component of the population. As such, their needs and wants in terms of seafood for at-home consumption may want to be considered by the seafood industry and its support groups when developing a long term marketing strategy. 3 Persons of Spanish origin may be of any race.

PAGE 99

89 Household Receives Food Stamps Households receiving food stamps (X20) were estimated to have total at-home consumption of seafood equal to 0.11 pounds valued at $0.10 in excess of those households not receiving food stamps. Caution, however, needs to be exercised when discussing the importance of food stamps to at-home seafood consumption since receiving food stamps did not have a statistically significant effect on at-home consumption of seafood in either the expenditure model or the quantity model Though not documented, it is a commonly held belief among seafood dealers that food stamps are an important factor in a households 's decision to purchase and consume seafood. The results of the Tobit analysis, however, tend to refute this idea. Since eligibility to collect food stamps is related to household income, it is likely that Black and elderly households are major recipients of food stamps. In fact, about 30 percent of all food stamp recipients were Black in 1982 (United States Department of Commerce, Bureau of the Census, 1985). As discussed earlier, Black households and households consisting of an elderly couple tend to consume more seafood at home than that of the "average" household. The combination of these two events may have resulted in the association of food stamps with seafood purchases. Furthermore, receiving food stamps may influence the time of month that households collecting food stamps purchase seafood. If receiving food stamps tends to result in these households purchasing seafood during a narrow span of time, seafood dealers may incorrectly associate

PAGE 100

90 this with an overall increase in seafood purchases among this group of consumers. Household Caught Fish for Own Use As expected, households which caught fish (X21) consumed a greater quantity of seafood than did their counterparts. Similarly, weekly expenditures on seafood consumed at home were higher among households 4 who caught fish than among those who did not. As evidenced by the estimated probabilities given in column 4 of Table 3-3, having caught fish was second only to race among the discrete variable in determining whether a household consumed seafood. Employment of the Meal Planner Employment of the meal planner (X22) was not statistically significant in explaining weekly expenditures or quantities of seafood consumed at home. The estimated signs of the respective coefficients, however, were negative as expected given the increased opportunity of the meal planner's time when employed. In the next chapter, consideration will be given to the effect of employment by the meal planner on consumption of those products which require the most preparation time. Sex of the Meal Planner The sex of the meal planner was significantly related to at-home consumption of seafood, ceteris paribus. Households with female meal 4 If the fish consumed during the one week interview period consisted of that which had been caught rather than purchased, the price assigned to that product was based on the average retail price of a comparable product in that region and season.

PAGE 101

91 planners (X23) consumed greater quantities of seafood and had greater expenditures than households with male meal planners. This may reflect a female's familiarity with cooking procedures, etc. Household Size The household size was found to be an important determinant of at-home seafood consumption which is consistent with related studies (e.g., Capps, 1982; Salathe, 1979). The positive linear household size coefficient (X24) and the estimated negative coefficient associated with household size squared (X25) imply that weekly expenditures and quantity of seafood consumed increase with increases in household size but at a declining rate. Because of the nonlinear specification of the household size component in the models it is useful to evaluate the effect of household size on seafood consumption at various levels of household size. The more important effects are presented in Table 3-5. As the information in Table 3-5 suggests, within the relevant range very little economies to size are exhibited in either seafood expenditures or quantities consumed. This is not surprising given the small magnitude of the estimated parameter associated with the squared term of household size (X25) compared to the linear term (X2A) in the two models. The information provided in the table indicates that total weekly seafood expenditures decline with the addition of a fourth member while the total quantity consumed declines with the addition of a fifth household member. Within household sizes generally encountered in the data each additional household member resulted in increased consumption of 0.158 pounds per week with related expenditures increasing about $0.26.

PAGE 102

92 • 00 > CD < e O x: i 4-1 CO -a i— i c o in CO X! CD CO to CD 3 l-l O 3 x: 4-) •H c •H C 01 ao -cr ex C X •H cd T3 •H CO rH CD S-i S) cd CD i— 1 a. 00 (-• t-i Q O CD a, cu N <4-l •H o co i-i -o CD rH XI o E CN J— X< 3 cu co 3 O X! c •H CO *— 1 (11 ao c CO x: o m to o c o co -h 4-1 4-1 o o. a) E 4-1 3 <4-i in CD c o •a u CD 4-1 T3 CO o E O •H <4-( 4-1 CO CO CD w co c o c o D. E O U c CN CN vO in (X) r m r co m lO CO c^ in in i — i cn o m O • a lo O r- en o vO 00 St o cn m in co O m o rH • c , > 4J H CD H — s CD •H < — 1 rH •H co r-i iH 32 H X xj •H X •H CO XI CD rH CO •J) XI CD c 4-1 a CO C 4-1 -3 CO H to rH <4-l XI •H H 3 H IH XI X) O --v O c •H a X XI O o o H H X! •H N a CJ ro •H x: lN (A oo cu CJ r>j >. a E CO U QJ •H >, 3 — G X 4-> m > — 4_> in 3 c HI CO Nl 4-1 CO Ed (J 3 tn -H •HT3 CD CO CO O" 4J 3 w -H T+T3 CD C CO O rH N Qj C XI CO O CH rH Cu QJ U W X! 1 1 -H 4-1 C O u CJ X! •H 3 u ro QJ •HX) Cu O o u CD CD i — i XI Ch (J C 3 — CD oo x co CJ •H ~ CC cc •H CO CD — u CU C 3 c CD XI c ex 4-1 >, Cj C c 3 X XI 3 cx H 4-> O CO •H \ O o 3. 4-1 4J o ro H CD O O X CM. -a U H X! E 1h •H QJ •H CJ H X! E \ v.. H CD E c 01 4-1 u 3 # O. 4-1 Hi Qj 4-1 u 3 /-V CU 4-1 9 CD c_ a, 7) -a ft 3 CO c a. CX CO o. en a, X T3 C •H 3 ca X — 3 0-T3 •H 3 X CD CJ o W CD O 3 CD CD o CD O CD ca 4-1 u w 4-1 m CU u cr CO HI (J W 4-1 CO CD u H c U w u c cc H 3 u CD (J 3 CC >> CO o QJ o CD CD o 3 CH ca o QJ c 1 QJ 0 3 — 1 rH 4-1 (J Oh t-i — aU CO o rH 4-1 u a. 4J a. o ca C O X X X. -Y O X X X CD E— w w C CJ H W W u CJ _2 <3

PAGE 103

93 Expected seafood expenditures among consuming households consistently increased though at a declining rate with each additional household member (Table 3-5) This suggests that among consuming households each additional household member resulted in a small decline in both expenditures and quantities consumed for household members. These declines may relate to price discounts associated with larger purchases and less waste per household member with increases in household size. ) The expected probability of observing a positive level of both expenditures and quantity consumed increased with each additional household member. The observed probabilities, presented in Table 3-1, bear out the fact that households consuming seafood on a weekly basis were in fact approximately 13 percent larger than those households not consuming seafood. Though the expected probability of consuming seafood increased with each additional household member, it did so at a declining rate. It is useful to address why the expected probability of consuming seafood at home increased with household size, ceteris paribus. One hypothesis put foreward to answer this question relates to the expected increase in variety of foods consumed associated with increases in household size. Assuming each member of any given household has an individual preference function which contributes to a household consumption function, increases in family size suggests an increased probability that at least one member prefers seafood. This preference will then translate to an increased probability that the household will consume seafood.

PAGE 104

94 As presented in equation (2.23) an elasticity associated with Tobit analysis can be broken into two components where the first term reflects the percentage change in consumption among consuming units due to a change in X.^ while the second component reflects the elasticity of the change in the probability of consuming associated with the changes in X Approximately 60 percent of the estimated household size expenditure and quantity elasticities reflect an increased probability of consuming seafood associated with increases in household size while the remaining portion reflects increased expenditures and consumption among those households already consuming seafood as noted in the following two elasticity estimates:"' N EXp = 0.1635 + 0.2450 = 0.4085 N Q = 0.1957 + 0.2834 = 0.4791 Subtracting the quantity elasticity with respect to household size from the expenditure elasticity with respect to household size provides a measure of the quality elasticity which in this example was estimated to equal -0.0706. The estimated negative quality elasticity indicates that households tend to purchase relatively less expensive seafood items with increases in household size. Household size remained relatively stable from 1950 to 1960, declining from an average of 3.37 to 3.33. However, since 1960, the average household size has declined significantly reflecting both a decline in birth rates and an increase in the number of single person households. Between 1960 and 1982, the average household size in the ^Evaluated at the mean values of all variables.

PAGE 105

95 United States declined more than 18 percent to 2.72 individuals (United States Department of Commerce, Bureau of the Census, various issues). Thus, holding all other factors constant, the decline in household size between 1960 and 1982 would have resulted in about a 7 percent decline in expenditures and approximately an 8 percent decline in weekly quantity of seafood consumed at home. Of course, other factors have not remained constant. The declining family size, for instance, reflects the gradual change in household composition. Increases in the proportion of younger and older households, as discussed earlier, at the expense of middle aged households will yield a smaller household size. All of these factors must be viewed simultaneously when considering expected changes in at-home seafood consumption throughout the nation. The declining household size has another important implication that the seafood marketing sector may wish to consider when conducting! and planning a long term marketing program. Primarily, the seafood industry and its support groups should recognize and react accordingly to the fact that consumers of seafood products are going to desire smaller portions of seafood due to the decline in family size. J Education of Meal Planner Seafood consumption was estimated to be positively related to the education level of the meal planner (X26). Each additional year of education of the meal planner increased the total household expenditures on seafood consumed at home by just over $0.04 per week, ceteris paribus. Similarly, the total household at-home consumption of seafood was estimated to increase by almost 0.03 pounds per week with each

PAGE 106

96 additional year of education of the meal planner, ceteris paribus. Though increased expenditures and quantities of seafood consumed associated with additional schooling may appear relatively small, it should be kept in mind that the analysis is conducted on a weekly basis. Extrapolating to a yearly basis, each additional year of education is expected to increase household consumption of seafood by 1.56 pounds or about 0.53 pounds per household member. This figure is relatively large when compared to the annual U.S. per capita consumption of seafood equalling 12.8 pounds (edible weight) as discussed in Chapter I. The estimated positive relationship between education and at-home seafood consumption may reflect increased awareness of the meal planner for a balanced and nutritious diet. This relationship is economically important given the increase in the level of education for the population over the past two decades. In 1960, the median number of school years completed by all individuals 25 years or older equalled 10.6. In 1982, the median equalled 12.6 years, or almost 20 percent more than in 1960 (United States Department of Commerce, Bureau of the Census, 1984). Number of Guest Meals Based on the positive relationship of both quantities consumed and expenditures to the number of guest meals served (X27), it must be concluded that households entertaining guests apparently often serve seafood. Two hypotheses, or a combination of the two, can be forwarded in an effort to explain the statistically significant positive relation between seafood consumption and the number of guest meals. First, increasing the number of guest meals by definition

PAGE 107

97 increases the total amount of food served from home supplies which consists of a variety of items including seafood. Thus, increasing the number of meals served should effectively increase the amount of seafood consumed. Second, at least certain types of seafood are considered delicacy items (e.g., shrimp, lobster) to be served on special occasions such as when entertaining guests. Expenditures on Meals Consumed Away from Home Given the propensity by the typical American household to consume seafood in the away-f rom-home market, it is especially important to examine what is happening in the away-f rom-home consumption market. Haidacher et al. (1982) provide a good synopsis of the changing patterns in the away-f rom-home market based upon a comparison of the Spring 1965 United States Department of Agriculture Household Food Consumption Survey with the Spring 1977 Nationwide Food Consumption Survey. Some of their results are provided in Table 3-6. As reported by Haidacher et al (1982), away-f rom-home consumption has expanded among every income group, each household size, and among both Black and nonblack households. Overall, the percentage of meals consumed away from home increased approximately 57 percent between the spring of 1965 and the spring of 1977 from 9.8 to 15.4 (Table 3-6). Among lunch and dinner meals when seafood is most likely to be ordered and consumed, the percentage number of meals consumed away from home increased 41 percent and 74 percent, respectively. The data provided in Table 3-6 also suggest that increases in income are positively related to the percentage of meals consumed away from home. However, with respect to household size, no apparent pattern exists between the

PAGE 108

98 CJ H e Ih 0 G. C CO E T) o C M ca <4H m a o> 3 — i CO c c CJ •H *J (h a D. c Cfi CO rH CO CO U CU H E — 1 W IH •H O >-. ai i-> i-> c U u ra u Ih Ih ca CD J3 a. U rH CO < H CO jC u 0> m o ON in 0> m o ON I s On LO O o oo a-. c CN CN CO CN co r-r- x
PAGE 109

99 percentage of meals consumed away from home and the size of the household. Perhaps the most important information contained in Table 3-6 is that associated with the percentage of meals consumed away from home according to race. Among Black households, the percentage of lunch and dinner meals consumed away from home increased 70 and 204 percent, respectively, compared to only 37 and 66 percent, respectively, among nonblack households. Given the propensity for away-f rom-home seafood consumption compared to at-home consumption of seafood, it is important to recognize the emerging patterns. Just as important, however, is the recognition that the away-f rom-home market is likely to expand even more in the long run, largely at the expense of the at-home market. A growing body of research suggests that the total expenditure/income elasticity for food consumed away from home is approximately twice that of the at-home consumption market (e.g., Eastwood and Craven, 1981; Haidacher et al., 1982). Second, though the estimated price elasticity associated with food consumed at home is apparently more inelastic than that associated with food consumed away from home, the importance of this factor in stabilizing at-home food consumption will probably be negated by changes in the socioeconomic structure of the population. For example, increases in the education level and the proportion of females entering the work force, and a decline in household size are all believed to lead to increased expenditures on and the number of meals consumed away from home (see Prochaska and Schrimper (1973) and Redman (1980) for a discussion of those factors which determine away-f rom-home consumption)

PAGE 110

100 The results of the analysis indicate a rather strong negative relationship between expenditures on meals away from home and weekly at home consumption of seafood which is in contrast to the results reported by Perry (1981). A $10 increase in expenditures on meals consumed away from home was estimated to reduce at-home consumption of seafood by $0,057 and 0.059 pounds, respectively. This relationship is especially important considering the growing away-f rom-home consumption market and one which must be considered in any attempt to increase the at-home demand for seafood. Income before Taxes Income was found to be an important determinant of seafood consumption as has been the case in most related studies (e.g., Capps, 1982; Salathe, 1979; Haidacher et al., 1982; Perry, 1981). The statistically significant positive linear effect (X29) and the negative coefficient estimated for income squared (X30) imply a positive but declining marginal propensity to purchase and consume seafood at home with increasing income. The statistically significant parameter estimate associated with the interaction term between income and race (X31) suggests a different marginal propensity of at-home seafood consumption among households of different races. The negative estimate of this term indicates that white households have a lower marginal propensity to purchase and consume seafood at home at all levels of income than nonwhite households, ceteris paribus. The positive estimate of the parameter associated with the interaction between family size and income (X32), though statistically insignificant,

PAGE 111

101 indicates propensity to increase at-home seafood consumption with increases in family size, ceteris paribus. The nonlinearity specification of the income component in the expenditure and quantity models makes it important to evaluate the effect of income on weekly expenditures and quantities of seafood consumed at various levels of income. The more important effects are presented in Table 3-7. The information given in Table 3-7 indicates a declining, albeit small, marginal propensity to consume seafood at home with increases in household income. For example, at an income level of $5,000, the total expected change in weekly expenditures with respect to a $1,000 change in income equalled $0,026. At an annual household income level equal to $25,000, a $1,000 change in income was predicted to change the total expected weekly expenditures by $0.0242, or about 93 percent of the estimated change at an annual income level of $5,000. About 25 to 35 percent of the change in at-home seafood consumption was estimated to reflect changes among those households already in the market in terms of either increases or decreases in weekly expenditures and quantities consumed. The remaining 65 to 75 percent of the change in at-home consumption of seafood, therefore, reflects changes in the probability of market participation, either entry or exit weighted by expected expenditures or quantities consumed. The positive estimates of the total change in seafood consumption and change in seafood consumption among participating households at various levels of income as specified in Table 3-7 indicates that the level of income required to reach a saturation level of seafood consumption was far in excess of that reported by most households in

PAGE 112

102 I4H CN oo o> CO O • LO CO uO o o ST o> O CO T3 CN Q CN C C Csl c O m o CO • • • CO 0) kl a w CD •H E "O O st CN m rH C o a> o CD CD >-. a O M-l 00 c CO CD LO cx> eji st c PQ m OJ o r~C o o c C • • • • 0) E o u c •H X Q 00 00 o CO co r- 1 o 4-1 CJ H COccj — i rH •H 4j -a -a H X u a) *H CO X fO CD E C X u a CO m p •H co 3 H <4-l w • • XI C O o H CD C CD •H X •H ln o -H 0) N >, CU > — JJ CO 4J CD CO w 4-1 a •H •HT3 4-1 -a x ^ CO o rH 0) CO o CD u w H 4-1 C E O U CO cu cu u O H 4-1 a T3 CO OCX CO CD •H 4-) CO 4J CD C C C CO c Cu 4J CO 0 •H 4-1 O C3 •H O o X a, w co -a U -H X | s•H a) e c CD 4-> U •Jo 0, 4-1 3 CD a. a. tn a. c CO a. x E o a -a H c 4-1 X cd 3 QJ o w QJ o I en c CD co ^ u 4-1 cn 0 u CD rH C U w u c 00 CD c >, CO O CD o OJ o c o rH 4-1 O Q, 4-1 G. u ca o rH a. ^ o X X x X E '— w H U O CD u 3 C7> CN CO 0> -i C 4J CO O" 4-1 x: v — co o w CO CD •O 00 CD C C 4-i O CO (J *H £ CD 4-1 O a. a. x e T3 CD CM IT' CO St m CN Q c n r-H r> St Q g O st o sj rH st o CN o UO C O st o CO O a o> CN st st CO c i — i c o st C o >^ CO QJ QJ 3 0 H O J3 CD -H >< CO CN 4-1 D WH OfcH x; -h i — i XI OO -H CO C X XI •H co O E ^s u 3 ^— Q, 1/1 C 0-T3 O w CD U W 4-1 CO CJ O • — i CD 4-> a, X •H X X co ca \ x ^-n O -H Sh CN •HXJ CO CN CD v>4J C fa-, u o CD *H c a, 4-i O X o, •HUE 4-1 3 a c to E -H C 3 O CO CD U C 00 O C4 u co o X u

PAGE 113

103 the 1977-78 survey. The information in Table 3-7 provides two reasons why the saturation level of at-home consumption of seafood is expected to occur only at very high levels of income. First, the change in consumption among participating households declines very slowly with increases in income at least within the relevant range. Second, the expected probability of consuming seafood increases with income, though at a declining rate, throughout the range of income reported by most households in the 1977-78 survey. Following the specification of the Tobit elasticity, given in equation (2.23), the elasticity of expenditures on seafood consumed at home with respect to annual before tax income equals N EXP = 0.0953 + 0.1436 = 0.2389 Similarly, the quantity elasticity of seafood consumed at home with respect to before tax income equals^ N Q = 0.0496 + 0.1318 = 0.1814 The estimate of the weekly at-home seafood expenditure elasticity with respect to income, 0.2389, is well within the range of estimates given in previous studies, the results of which are summarized in Table 3-8. The current estimate of the at-home seafood expenditure elasticity thus adds to the growing amount of research that indicates that at-home consumption of seafood is very unresponsive to changes in income. ^ Evaluated at the means of all variables. ^Evaluated at the means of all variables.

PAGE 114

104 Table 3-8. Estimates of at-home seafood expenditure elasticities with respect to income 3 Study Expenditure elasticity Method of estimation Capps (1982) Salathe (1979 0.1651 0.3568 1 0.2407 OLS nonlimit observations OLS all observations Perry (1981) 0.0609 0.2040 Tobit Haidacher et al. (1982) 0.16 OLS all observations All studies with the exception of Haidacher et al. were based on the 1973-74 household consumption survey. The Haidacher et al. study was based on the 1977-78 household consumption survey. The first estimate given for Salathe 's study was based on data from June 1972 to June 1973 while the second estimate was based on data from July 1973 to July 1974. The two estimates associated with Perry's study gives the range among the different regions.

PAGE 115

105 Assuming that the quality elasticity of at-home consumption of seafood is positive, one would expect the quantity elasticity to be somewhat less than the expenditure elasticity. The estimate of the quantity elasticity, 0.1814, was in fact less than the estimated expenditure elasticity which equalled 0.2389. This translates to a quality elasticity for seafood consumed at home equal to 0.0515. Thus, an increase in consumption resulting from an increase in income is expected to result in a greater increase in expenditures than quantity consumed; the difference measuring a demand for quality and/or services Household income, measured in 1982 dollars, for selected years and among different races is given in Table 3-9. As indicated from the data in the table, after growing steadily throughout the 1950s and 1960s, the real household income stagnated during the 1970s and even decreased in the early 1980s. The decline in growth in real household income during the 1970s compared to the previous ten year period and the actual decline in real household income during the early 1980s has probably affected the growth in per capita consumption of commercial fish and shellfish (Figure 1-1) and especially at-home seafood consumption. Black households, estimated to have a higher propensity to consume seafood at home with respect to income than did White households, experienced a 13 percent decline in real income between 1970 and 1982 compared to only a 3 percent decline among White households. Given the growing proportion of Black households in the United States, the decline in real income among this group poses an obstacle in any attempt to increase at-home consumption of seafood.

PAGE 116

106 Table 3-9. Median family income in constant (1982) dollars for selected years Year Race of Household All White Black 1950 13,308 13,813 7,494 1955 15,926 16,629 9,170 1960 18,317 19,018 10,528 1965 21,283 22,183 12,216 1970 24 528 25,445 15,608 1975 24,664 25,589 15,744 1980 24,626 25,658 14,846 1982 23,433 24,603 13,598 SOURCE: United States Department of Commerce, Bureau of the Census (1984).

PAGE 117

107 Given the relatively low estimate of the at-home seafood consumption elasticity with respect to income, even large increases in real household income such as that of the 34 percent increase experienced during the decade of the 1960s, will not have an overbearing effect on at-home consumption of seafood. For instance, an increase in real household income during the decade of the 1980s equal to that experienced during the decade of the 1960s was estimated to result in only a 6.2 percent increase in the quantity of seafood consumed at home with related expenditures increasing about 8.1 percent. Outlook for Increasing At-Home Demand for Seafood and Implications The outlook for increased at-home seafood consumption over the next several years does not appear promising without significant advances by the seafood industry and its support groups in terms of more effective marketing and promotional efforts. For example, generic advertising on seafood is extremely small when compared to most other o food sectors. Furthermore, nongeneric advertising on seafood in major media outlets (excluding newspapers) declined 10 percent between 1977 and 1982 compared to a 55 percent increase in nongeneric advertising associated with meat and a 167 percent increase in nongeneric advertising associated with poultry (Anonymous, 1984). These factors alone would tend to indicate a disadvantage to the seafood industry vis-a-vis other food sectors; all competing for a limited household food budget. Furthermore, as discussed throughout the analysis, movement in the g Generic advertising on seafood averaged $85,000 annually during 1981-82 compared to $4.2 million on red meats and almost $27 million on milk and other dairy products (Morrison and Armbruster, 1983).

PAGE 118

108 level of many of those factors explaining at-home consumption of seafood is in a direction not compatible with that of maintaining long run growth in the at-home seafood market. Summarizing some of the discussion presented throughout this chapter, any or all of the following conditions may be expected to <^^\^ result in declining, at-home seafood consumption, ceteris paribus: (1) an increase in the proportion of households with a younger head, (2) an increase in the proportion of households in the United States comprised of an elderly individual, (3) an increase in the proportion of households in the United States in which the meal planner is employed, (4) male s becoming more involved in the planning and preparation of meals, (5) a decline in the average family size, and (6) an increase in expenditures on meals consumed away from home. As indicated throughout this chapter, most if not all the above conditions are currently taking place in the United States. Offsetting those factors which are expected to result in declining consumption of seafood at home, seafood consumption at home can be a? expected to increase given any or all of the following conditions, certis paribus: (1) an increase in the proportion of households in the United States comprised of an elderly couple, (2) an increase in the proportion of nonwhite households in the United States, (3) an increase in the education level of the meal planner, and (4) an increase in household income. As previously indicated all these conditions have been occurring in the United States over the past two decades. Though several factors suggest that at-home consumption of seafood will increase very slowly if not acutally decline while other factors suggest that at-home consumption of seafood will increase, the one

PAGE 119

109 overriding factor that will probably determine the future status of demand for seafood for at-home consumption is that of the change in the market for food consumed away from home. Given the propensity to consume seafood in the away-f rom-home market, increases in away-f romhome consumption can be expected to have a rather strong negative influence on at-home consumption of seafood. Evidence suggests that away-f rom-home consumption will continue to take a larger portion of the consumers income in future years. In fact, many of those factors found in this study to increase at-home consumption of seafood also increase total away-f rom-home consumption of food. Redman (1980) investigated those factors determining expenditure on meals away from home and concluded that increased family income and a college education of the woman head of household, both of which were estimated to increase at-home consumption of seafood also contribute to increased expenditures on meals away from home. Additionally, Redman concluded that a decline in family size, which the U.S. population is currently undergoing, also contributes to increased expenditures on meals away from home. Decreases in expenditure on meals away from home were found among Black households and with increased age of the woman. As concluded in the current study, at-home seafood consumption was relatively high among Black households and elderly couples, though consumption was much lower among elderly individuals many of which are women, certis paribus. As expenditures on meals consumed away from home increases, seafood consumed away from home as a percentage of total consumption will increase proportionately. Thus, the cyclical nature associated with the demand for seafood as discussed in Chapter I will likely

PAGE 120

110 continue as the general economy of the United States and more particularly restaurant trade oscillates with changes in real income. An effort at promoting increased at-home consumption of seafood is probably in the long run the only effective means of dampening this cyclical demand.

PAGE 121

CHAPTER IV SPECIFIC SEAFOOD PRODUCT FORM ANALYSIS Introduction In this chapter, a discussion of the parameter estimates associated with the disaggregated Tobit seafood models is presented. Given the voluminous amount of information associated with the specific product form models, discussion centers on noting and explaining differences in the signs and significance of the estimated parameters among the seafood product consumption models and with respect to the total consumption models. This is in contrast to the in depth discussion given to each parameter estimate in the previous chapter. All parameter estimates are, however, presented in Appendix B. Some descriptive statistics associated with the data used in estimating the specific seafood product form models (fresh, frozen, canned, finfish, and shellfish) are presented in Table 4-1. Among the alternative types of processing (fresh, frozen, canned), canned seafood is consumed at home by the most households and frozen seafood by the least households (Table 4-1). In fact, canned seafood is served by more households (31.69 percent) than fresh and frozen seafood combined (29.63 percent). However, referring back to Figure 1-1, per capita consumption of canned seafood has equalled only about 60 percent of that of fresh and frozen in recent years (edible basis). Thus, it must be concluded that consumption of fresh and frozen seafood products is 111

PAGE 122

112 CO CN In r— *H V o LP; M-i \ P"i 00 o o rH o rH o CN Pi Pi CD QJ X! 4-) co (u U 4-> U pC U] U] uD CO uu CO i — i o *H >u 1 — < - o o *o CO o CD o 4-J CJ ^ CN i — i 14H CO CD 03 cc X o CO u N CO • un cd O CN CO 4-) Jh O pH rH CN rH u Pu t— H •H o o 'o CI) CO (J^ 1 — 1 LM rn C /-Y"v UU "Y~s - I — 1 CO c 9 H 4-> -C CO N XI CO CJ C oo 00 i— 1 Efl o C w C w CD u o o > O E cn E co *H GO ro -a CO T3 4-> CO ID i— i rH CX DO rH CD O C O •H rH C O ^ jr, O o •H _c 3 QJ •H CD o _c E CO CO 4-1 CO cn CD tfl •H D a. s CD to m 3 -o O E o Q c O C JZ 3 J= o o _c QJ co u CX OO c oo 4-1 X c o c t — i M-f 4H o CD -H U -H 1 CD o O E > H > 2: 7^[X < < CC IT) ao rH o o CO a c E CO CO *o T3 rH c o 3 O CD a, co \ 3 Si O co x: rH rH 00 o c -a -h CD 3 00 CO co c u o CD U > <

PAGE 123

113 preferred in the away-f rom-home food market relative to consumption of canned seafood products. Average consumption of canned seafood among consuming households was found to be approximately 46 percent that of frozen seafood and 24 percent that of fresh seafood (Table 4-1). These differences probably reflect differences in the amount of waste associated with the different product forms at retail. For example, most if not all waste has been removed by the processor before canning seafood. Frozen seafood is generally filleted before sold and hence there is no bone or head waste. Fresh seafood, however, is often sold on a whole weight basis and hence there is often head, bone, shell, etc., waste associated with the product. This waste is generally acknowledged to account from about one-half to two-thirds of the raw product form. Price variation among product forms reflects differences in waste as well as the cost of processing the product. Canned seafood, which requires the most processing and which has the largest round weight to sales weight conversion, sold for an average of $1.78 per pound (Table 4-1). This compares to a sales price of $1.61 per pound for frozen seafood and $1.37 per pound for fresh seafood. The percentage of households consuming finfish (47.39 percent) compared to those consuming shellfish (7.62 percent) suggests at-home consumption of finfish was overwhelmingly preferred to at-home consumption of shellfish. The observed difference in the number of households consuming finfish at home compared to the number consuming shellfish at home is probably related to at least two factors. First, meal planners probably perceive finfish products as easier to prepare

PAGE 124

114 than shellfish products. Second, shellfish is considerably more expensive than finfish (Table 4-1). Comparisons of Consumption Parameters Information concerning the signs and statistical significance of the estimated parameters associated with the specific product form and total seafood expenditure models is presented in Table 4-2. Since the signs and statistical significance associated with the estimated quantity models are generally in agreement with those of the expenditure models they are not presented here. Summarizing the results in this manner allows for an easy and logical comparison of the estimated models. The complete descriptive and summary statistics associated with each of the specific seafood product form models are presented in Appendix B. Region Overall, household expenditures on most seafood product forms were highest among households residing in the Northeastern region of the United States (XI) and lowest among households residing in the North Central region (X2). This difference holds true for expenditures on fresh and canned seafood. However, expenditures on frozen seafood showed a reversal of that of the other two processed product forms with respect to Northeastern and North Central regions. Weekly household expenditures on frozen seafood were estimated to be greatest in the North Central region of the United States and lowest in the Northeastern region.

PAGE 125

115 x: w • H CD *4H u — | 3 f— | 4J CD •H sz TD CO c CD CX X CD jr: CO T3 •H o 4H O x; 4-1 •H rH co 4-1 o cd H XI R] H Ih CO > + I + + + I I + jjt >|; + I I + + I + I I + + I I I I + I + I I + + # + + + + + + I I + + + + I I + I + I + + + + + I I I I "l *l $ # # # CO I — I 1 X CD X X W >H X -o C a C iH CD -H e CD U SZ CD Ih u T3 O Ih — TJ iH T3 rH — i •H O rH H H x: \ •H XI XI u ? XI U u 11 o x: cu , H S c rH z' N u — rl -o CO c CO >1 m CD CD CD CD CD u in :-: 4-1 X 14-1 rH •H — ( •H T3 o 4-1 c H O •H 00 Ih CC Ih CD 4-i C o CJ X >: l — 1 c U c Ih 00 CO CJ C •H c •H CD •H CO CO CO CJ Ih 4J r-l c3 w £ CO E o QJ CO CO £1 DC CD sz sz x: u Ih c 0 DC cc CC CO rH c 4-> 4-1 H 4-1 3 c H B r— -C c C c c -a c u Ih 3 C c X) 0 Ih e — HJ 3 3 3 -a •H o O O (0 a CO a. 3 CO CO 0 0 0 O -H 00 z co XI CJ CO CO CO [fa 3 >>H >* £ cu Ih CD 0 ai CO

PAGE 126

116 -a CJ 3 C a o u I xi ra H n3 CO •H n > X CO cu cy: co •H E PL, i-i O 4-1 4-1 T3 u 0) 3 C -a c o CO In u a. T3 O o c QJ co N 0) O CO U X! CO o Ih 4-> C ai 3 C •H 4J c O u u •H O xs co w 3 o 33 + + + I I I I + + + I I I I I I I I -H •H -H x: x: X! U I + 1 I I + + I I I I u (J X! • • o x: 4-> 4-4 -H • • c c •h 3 s 6 b n •a • • 3 i — i o cu cu i — i CO Oi H iH -H X • a. c M 00 Ui U s 3 rH IH c u C CO C eg ca H (0 D CD (J cu E co E H CO n i r m or E -o -O T3 c o s TJ 4-1 , u ^ — CN •H CN c CN QJ co cu 1 — 1 CO X X CJ X r—\ -H i — i CJ CJ S T3 X) o o 4J OJ > tU >, -a T3 -a -a H x; rl CO XI CO o CO •H cu X! 4-1 CJ CJ CO o CJ Rac a o Rec >1 Cau Emp >H

PAGE 127

117 E so J] CU xi in cn rH <— C I + + + + + I u u 3 — c 1-1 T3 CU C a co U + I # + I I o O Cfl o c x 0 + + + I w u ft. I + + i co c + i + i O 3 C C O U CM I X3 03 H M CO > V sCJ c CO w T3 c CN 3 3 QJ a CN X cr o kl — X O GO M jC CO a E cn O 3 C o CO CQ C c cr i — i cn si X >-i c H 03 ca t— i CO CO CO a CJ CO = 4-1 -1 H i — i u l-i E o c CO r-H rH a. , m -a -3 •H c c c CO CO M o e c £ CU Cfi o o> l-l -a CO CO rH 1 — 1 •rH K E CJ CO CO cn CO >-i ca CO 4-1 -o H cu 3 3 C E [— 1 4-1 4J eg CO 4J E W H E O o o •H O o CJ cu ca 3 O O £, — X E H H 3 0) Z CO Q CJ E01 CO -a 3 CJ C CO w u r-H

PAGE 128

118 E U o u 3 T) o o. TJ o o '4-1 co 0 T3 QJ 3 C •H 4J C o u cn I 0 H xs CO •H '-4 a j= c/) co 3 H a u a a a u c a N 0 Ih JS 'ft o u CO *J 0 H I + I + + I + I I I + XI to •H Ih > CO E I* CD J-> c o •H 4-1 o CO Ih (U 4J c :— i CN cn X CD N — 1 •H CO (0 H 0) •H u E CO co Ih 14-1 -a -a C c ca a 0) CD E E O o o u c c M + •z. CO co O X O ca ca CD 4-> tn ca c i-H CO >-< 4J o cx o u ca c c •H 4-1 CO E in O -a CD 4J H U o w ca re c CO o 3 rH ca > I 4-1 cn CD 4-1 CO E H 44 co CD Ih CD 4-1 CD E CO Ih CO CX CO CD > CU •h a 4-> iH CO CO 00 > 0) c 4-> CO CO i—i S CU > cn cu CD i— I 4J co jj § c •H CD •U U CO ^ cu cu o. Ih cu o 4J ON cu E CU ca x: Ih 4-1 co Q. 4-1 co 4J CO 4-1 x: c 4-> CO u cn -h CU M-) 4-1 -H CO c > rH 1 rH CO QJ CU -H > 4-) •h cn 4-) -H •H 4-> cn co O 4-1 c cn co cn ca co 15 3 cn cn cu cu 4-) 4-1 ca co e e •H -H 4-1 4-1 cn cn cu cu Ih Ih CU CU 4-1 4-1 CU CU E E CO CO Ih U CO CO o. cu 4J 4J CO CO x: x: 4-1 4-1 cn co cu cu 4J 4-1 CO CO CJ CJ • •H -H T3 -a
PAGE 129

119 The relatively high expenditures on fresh seafood by households living in the Northeastern region of the United States is consistent with the steady supply of fresh edible seafood products harvested and sold in the Northeastern region. Similarly, the relatively high expenditures on frozen seafood by households residing in the North Central region may be explained by the relative unavailability of fresh seafood in that region and hence the need for substitute frozen products. Apparently, households in the Northeastern region can substitute fresh seafood for frozen seafood which explains the relatively low expenditures on frozen seafood in that region. The relatively high expenditures on canned seafood in the Northeastern region compared to the other regions of the United States was, however, unanticipated given the availability of this product form in all regions of the country at expected comparable prices. A possible higher demand for at-home consumption of seafood in total in the Northeastern region due to differences in tastes and preferences may help to explain the higher expenditures on both fresh and canned seafood products in that region compared to other regions of the United States. The results of the analysis indicated that expenditures on both finfish and shellfish were highest in the Northeastern region of the United States and lowest in the North Central region. These differences in expenditures between the two regions probably reflect differences in availability of fresh quality seafood as well as differences in consumer preference between the two regions. The preceding discussion should be considered by the seafood industry and its support groups when considering what products and product forms to provide in different regions of the country.

PAGE 130

120 For example, the Gulf and South Atlantic Fisheries Development Foundation has coordinated the effort of several groups aimed at expanding demand for many of the underutilized species landed in the Southeastern United States which enjoy only regional acceptance. Much of the effort has gone to developing markets for these products in the Midwestern section of the United States. (See Cato and Prochaska (1981) for a discussion of the program.) Given the familiarity with frozen seafood products in this section of the country, promoting these underutilized species in frozen form appears logical. Urbanization Households residing in central city (X4) and suburban areas (X5) had higher estimated expenditures on all seafood product types and forms than did households residing in nonmetro areas (base), controlling for regional differences, income differences, etc. Furthermore, with the exception of frozen seafood purchases, households in central city areas consistently had higher expenditures on all seafood product forms than did households residing in suburban areas, ceteris paribus. The proximity of many of the larger cities in the United States to major fishing ports would explain, in part, the higher expenditures on fresh seafood consumed at home among households residing in central city areas in the United States. However, proximity to the coast does not explain the relatively high expenditures on canned seafood products by central city households since canned seafood products can probably be transported to the inland areas of the country at a minimal cost. Therefore, some other factor must account for the estimated differences in weekly household expenditures on canned seafood products and to some

PAGE 131

121 extent, possibly the other product forms as well. Households residing in nonmetro areas of the country are often farming families or live in farming communities. As such, generations of households have been raised on meat products and introduction of seafood products may not have been initiated to an extent necessary to establish a strong at-home demand for these products among these households. Season Generally, season of the year was estimated to be statistically insignificant in explaining weekly household expenditure on at-home consumption of seafood. Weekly expenditures on fresh seafood products consumed at home were found to be highest in the summer months (July, August, September, 1977), while expenditures on frozen seafood products were highest in the spring months (April, May, June, 1977), and expenditures on canned seafood products were estimated to be highest in the winter months (January, February, March, 1978), ceteris paribus. Household Life Cycle The household life cycle category (X9-X17, base) was very useful in explaining weekly expenditures on at-home consumption of the specific seafood product forms. For example, expenditures on fresh seafood tended to increase with the maturing of the household, with elderly households generally having the greatest expenditures on fresh seafood products (Table 4-2) Expenditures on canned seafood consumed at home, however, were lowest among elderly households (X17, base), ceteris paribus. Expenditures on frozen seafood products attributable to differences in the stages of the household life cycle were

PAGE 132

122 comparable to fresh seafood expenditures with the exception that households consisting of an elderly individual (X17) had lower weekly expenditures than several of the middle-aged life cycle categories, ceteris paribus. Finfish and shellfish expenditure patterns differed significantly with life cycle stage, ceteris paribus. Households in the younger stages of their life cycles (X9-X12) generally had lower finfish expenditures than either middle aged (X10-X16) or elderly households (X17, base). Elderly couples (base) had the highest weekly expenditures on finfish. However, weekly expenditures on shellfish products were relatively low among elderly households (X17, base) when compared to most of the middle-aged households (X13-X16) and even some of the younger household life cycle categories (X9-X12). Overall, the estimated parameters relating household life cycle category to specific seafood product forms conform to theoretical expectations. Expenditures on fresh seafoods by younger and middleaged households were relatively low when compared to elderly households. The fear of bones associated with nonprocessed seafood, especially fresh, is hypothesized to explain the relatively low expenditures among households with young children. A lack of cooking and/or preparation expertise may be another reason for the low consumption. Expenditures on canned seafood products which generally have no bones and are typically recognized as requiring little preparation and cooking skills were higher among younger and middle-aged households with and without children than among elderly households, ceteris paribus. Similarly, weekly expenditures on finfish for at-home consumption among households with young children (XI 1, X12) tended to

PAGE 133

123 be relatively low when compared to other households, ceteris paribus. Though expenditures on finfish products by households with young children tended to be relatively low, a similar trend was not noticed with shellfish expenditures. It is likely that the specification of the life cycle category was overly refined to account for differences in shellfish consumption among households in different life cycles given the relatively few positive observations for this product form. Race Substantial differences in preferences for the alternative seafood product forms exist among households of different races. Black households (base) made significantly higher expenditures on fresh seafood consumed at home than either White households (X18) or households of other ethnic origins (X19), ceteris paribus. Alternatively, expenditures on canned seafood were significantly higher among White households and households of "Other" ethnic origins than among Black households. The difference in expenditures on frozen seafood consumed at home was not statistically significant among White, Black, and "Other" households. Statistically significant differences in expenditure patterns were estimated among households of different races in the purchasing of finfish products but not shellfish products. Estimated expenditures on finfish products consumed at home were highest among Black households and lowest among White households, ceteris paribus.

PAGE 134

124 Food Stamps In no instance did the fact that a household purchased or received food stamps (X20) significantly affect expenditures on the different categories of seafood consumed at home. Expenditures on fresh seafood products, finfish, and shellfish were positively related with food stamps while expenditures on frozen and canned seafood products were negatively related to food stamps. Fish Caught for Own Use Households who caught fish for their own use (X21) consistently made greater expenditures on the specific seafood product forms than did those households who did not catch fish. Furthermore, with the exception of expenditures on canned seafood consumed at home, having caught fish for home use was estimated to be statistically significant in explaining expenditures on the specific seafood product forms. The statistical insignificance of having caught fish for home use associated with weekly expenditure on canned seafood consumed at home is not unexpected given the nature of the product. Canning fish is not generally attempted by individual households. Hence, households who catch fish are likely to consume it fresh or freeze it for later consumption Employment of the Meal Planner Weekly expenditures on the various seafood products consumed at home were consistently lower among those households in which the meal planner was employed (X22) than among those where the meal planner

PAGE 135

125 was unemployed. However, only fresh seafood expenditures were related to employment of the meal planner at a statistically significant level. This observation is consistent with the rationale previously discussed for including this variable. The cost to the meal planner in terms of the amount of time required to purchase and prepare fresh seafood is expected to be greater than that required for other seafood product forms and thus the employment of the meal planner was expected to have its greatest effect on consumption of fresh seafood. Sex of the Meal Planner A female meal planner (X23) (as opposed to a male planner) was a statistically significant factor in explaining weekly expenditures on canned seafood products, finfish products, and total seafood consumed at home. Expenditures on fresh seafood and frozen seafood were also higher among households with a female meal planner, though not at a statistically significant level. Family Size Weekly expenditures on the various seafood products consumed at home were very responsive to changes in family size (X2A, X25 X32). Both the linear and quadratic parameters describing expenditures on the specific product forms were statistically significant for all but fresh seafood products. The interaction between family size and income (X32) was statistically significant only for canned seafood products. The estimated positive linear term (X24) together with the estimated negative quadratic term (X25) indicates initial increasing (at a declining rate) expenditures on frozen seafood products, canned seafood

PAGE 136

126 products, and finfish products with increases in family size. On the other hand, a negative linear term combined with a positive quadratic term was estimated for expenditures on both fresh seafood products and shellfish products. These results suggested that expenditures on fresh seafood products and shellfish products consumed at home declined, at least within initial ranges, with increases in family size. Given the increase in the opportunity cost of time of the meal planner associated with increases in family size, these results are not totally unexpected. Convenience in seafood products is most often associated with certain frozen products, such as fish sticks, and canned products. Furthermore, the majority of these products are made from finfish rather than shellfish. As family size increaes, it is logical to assume that the meal planner becomes more dependent on these convenience products. The estimated expenditure elasticities with respect to family size for the various seafood products are presented in Table 4-3. Expenditures on at-home consumption of frozen seafood products, canned seafood products, finfish seafood products, and total seafood products have positive elasticities with respect to family size while expenditures on at-home consumption of fresh seafood products and shellfish seafood products exhibit negative elasticities with respect to family size. Furthermore, the total elasticities (comprised of the elasticity of participation and the conditional elasticity) ranged from a low of -0.73 associated with that of expenditures on shellfish to a high of 0.59 associated with that of expenditures on canned seafood products. The above discussion and analysis suggest convenience in terms of purchasing, preparation, and consumption of the various seafood product

PAGE 137

127 >i 4-1 u •H CN ^ co 4-1 CU •H >> 4-1 u-i o ^ i — i CO O -H C H •H 4-1 O o 4-i CO -H CO o •H CO 4-J m CO X i — 1 CO co CO o co cu cu ci, in — 1 m o CO co 1 — 1 co > U CO -H 4-1 4-1 4-1 "O 4-1 o o o o o u -H C rH (_i 1 1 cu O CU O CO a. •H x: a. co 4-> sz cu cu co 4-1 CO U-4 u CO H 3 O rH IS o Si W x; 4-1 H CO CU H 4J •H U 4-1 •H H *J oo u co C -1-1 co •H 4-1 i—i >> E CO CO CU 4-1 3 T3 ffl •H CO < — 1 i — 1 rH o lO — i 00 LOl • 0 U C O CU o CO CM in CN TO CO u •H O JG CN rH r i — i a vO CU 3 4-1 U CU rH o rH CN rH rH rH 4-1 CO CO CO XI •H CO 00 3 C o o o o o O co "O rH COO 1 1 •H C t td O -C -H S-i (D 10 E 4-1 CO a e CO -H > X S-i a a; o c rH o rH u co rH 4-1 — u 4-1 CU 3 o cu -o 3 O co >-< e t3 a. co CU 0) 4-1 T3 E CO O E O a> 'H L — r— — 4-1 CO 4-1 co cu W CO 4-1 B CO u o T) CO r4 rH CO H jC w H CJ CO

PAGE 138

128 categories influence, at least in part, expenditure elasticities with respect to family size. In a study conducted by Gillespie and Houston (1975) consumers rated shellfish more difficult to prepare than finfish. Hence, as would be expected, the expenditure elasticity with respect to family size for shellfish products (-0.734) was much lower then that for finfish products consumed at home (0.562). In the same light, the estimated weekly expenditure elasticity associated with fresh seafood (-0.107) was substantially lower than that associated with either frozen seafood products (0.564) or canned seafood products (0.589); the latter two categories requiring much less expertise and time in preparation than that required for fresh seafood products. A breakdown of the total elasticities into the change in consumption among consuming households and the change in consumption resulting from entry/exit among households is also presented in Table 4-3. As indicated by the information contained in the table, most of the total elasticities reflects changes in the number of consuming households (participation effect) as opposed to increased (decreased) expenditures among consuming households (conditional effect). Since the conditional elasticity measures increased (decreased) consumption among consuming households, one would expect to find it to be a larger percentage of the total, the greater the percentage of the households consuming the product. For example, if all households consumed a given product then the conditional elasticity would by definition equal the total since there could be no entry from new consumers. As a percentage of the total elasticities, the conditional elasticities tended to be higher among those products most often consumed at home. For example, the conditional elasticities associated with fresh and frozen seafood

PAGE 139

129 consumption equalled about one-fifth of their respective total elasticities compared to slightly less than one-third for that associated with canned seafood consumption (Table 4-3). Education of Meal Planner The relationship estimated between education (X26) and expenditures on at-home consumption of the various seafood products examined in this study was positive in all instances and statistically significant with the exception of expenditures on fresh seafood products (Table 4-2) The elasticities of expenditures on at-home consumption of the various seafood product types and forms with respect to education of the meal planner were estimated to be 0.156 for fresh seafood products, 0.606 for frozen seafood products, 0.405 for canned seafood products, 0.205 for finfish products, and 1.249 for shellfish products; compared to 0.552 for total seafood. Thus, increases in the level of education of the meal planner is expected to have pronounced effects on expenditures on at-home consumption of the seafood product forms. The statistically insignificant education parameter estimate and relatively low expenditures elasticity estimate associated with fresh seafood may be the result of a confounded effect between an increase in the opportunity cost of time of the meal planner which prevents him/her from preparing fresh seafood and an increased desire for a nutritional meal associated with additional education. Number of Guest Meals Increases in the number of guest meals (X27) were estimated to be positively related to weekly expenditures on at-home consumption of

PAGE 140

130 all seafood product categories examined in this study (Table 4-2). The estimated positive relationship between the number of guest meals and weekly expenditures on at-home consumption of the various seafood product categories may be for one or two reasons. First, some seafood products are generally considered as specialty products to be served on special occasions, such as when entertaining guests. Second, increasing the number of guest meals by definition increases the total amount of all food consumed, some of which is likely to be seafood. Expenditures on Meals Away from Home Expenditures on at-home consumption of all individual seafood products with the exception of shellfish products, were estimated to be negatively related to expenditures on meals away from home. Thus, as expenditures on food consumed away from home continues to take a larger portion of the household's food dollar, expenditures on the various seafood products consumed at home are expected to decline, ceteris paribus. This is consistant with the fact that a larger proportion of total seafood consumption takes place in the away-f rom-home market. However, it must be considered that all other factors determining at-home seafood consumption are not remaining constant. Income before Taxes Consistent with estimates provided in the total seafood expenditure analysis, the estimated parameters associated with the linear income term (X29) and squared income (X30) were positive and negative, respectively, for all but one of the specific seafood product form expenditure models (Table 4-2). Weekly expenditures on canned seafood,

PAGE 141

131 the one exception, were negatively related to income in both the linear and squared income terms. A look at the parameter estimates associated with the interaction of income and race (X31) reveals that though statistical significance is noted only in the total, all parameter estimates were negative in sign. The expenditure and quantity income elasticities for the specific product forms and in total are provided in Table 4-4. Among specific product forms distinguished by level/type of processing, consumption of fresh seafood had the highest income elasticity estimates (Ng X p = 0.467, Nq = 0.413) while at-home consumption of canned seafood exhibited the lowest income elasticity estimates (N EX p = 0.192, Nq = 0.098). Consumption of shellfish had a significantly higher income elasticity estimates (N EXp = 0.543, Nq = 0.929) than did finfish (N EXP = 0.148, N Q = 0.1163). Table 4-4 also provides information on the decomposition of the total elasticities into their respective components; increased (decreased) consumption among consuming households due to an increase (decrease) in income and increases (decreases) in consumption reflecting increases (decreases) in the number of participating households. For those products in which the number of consuming households was relatively small (e.g., shellfish) the proportion of the total elasticity reflecting changes in the number of participating households was relatively large compared to those products in which the number of consuming households was relatively large (e.g., finfish). For example, about 80 percent of the total estimated income elasticity associated with shellfish products consumed at home reflected changes in the number of participating households compared to 60 percent

PAGE 142

132 n> •H rH U CO -H 4-1 4-1 O CO H CO rH 01 *J •H X) 4J 0) <4-H •H 4-1 O U co to co •H e cu CO •H 4-> n> o •H E U -H CO •rH 4-1 u CJ tn u >, 4-1 H d 3 4-1 C — 1 rH •a •H 0) — S(U o u o •H JZ Q, >, 4-1 4-1 O 4-1 CO -H CO ^H •r-i T3 CO 5 3 0 rH O rH o CO o — IH cr ea T3 CO C CO al jj 4-1 o >1 •rH 4-1 4J 4J H c T3 CJ a C oo H CO >~> c 4J o" 4-1 -H CO X u •H 4-1 — co •H X jC 3 U CO CD o — 4-1 0 CO CO cn •H ft i-H OO 3 3 •a CO w c 0 0 c o •H o rH E 4J D, o CO •H X , E o i-H o u ai c a •H X a CO cu 4-1 4J CO CD e u H o 4J IW CO cu w u o CO CD 4J CO u O cn cn m -j N cn cn co co \D o o r-~ O o cn r-.
PAGE 143

133 u •H O H 5^ -a 4-> CJ H 4-1 o (J CO H c H 4-> -J 0 u •H CO H o X CO 4J CO a r-H CO to cu D. ca >, H U >, *J CO H ij c — 4J H •-3 H a c o u h cs m cs n sj CN sf (N in lti ct\ n O O O O CO un c o o o o o I I I I I I I I I I I I I I I I I I I I I I I I I CO H rH CO c -a CO i CO 01 X! CD ai H H 4-J u H CO N c 4-1 rH O o 4-1 tU o c e 0) 00 -H u 3 •r-l -C a> rH U 4-1 CO CO a U o-

PAGE 144

134 for finfish. The estimated differences in the proportion of the total elasticities accounted for by changes in the number of participating households among the specific product categories is of importance for marketing strategies. The number of consuming households is inversely proportional to the number of potential consuming households. Hence, for seafood products such as shellfish, one would expect to find a relatively large market expansion through entry of new consumers as income increases. The estimated income elasticities presented in Table 4-4 should be of considerable value to the seafood industry and its support groups. As discussed in Chapter I, growth in per capita consumption of canned seafood from 1960 through 1983 averaged only about a quarter of the growth in per capita consumption of fresh and frozen seafood even though about 56 percent of the seafood advertising in major media outlets is specific to canned tuna and canned salmon (Anonymous, 1984). At least part of the reason for the relatively small growth in per capita consumption of canned seafood may be the result of a low income elasticity associated with this product. A comparison of the quantity elasticities with respect to income for the specific seafood product forms suggests the income elasticity for canned seafood consumed at home is only about a third of that for frozen seafood consumed at home and a fourth of that for fresh seafood consumed at home. Thus, continued promotion of canned seafood products may be important to assure a stable though not necessarily increasing market for these products. This is especially true given the fact that demand of canned seafood in the expanding away-f rom-home market appears to be somewhat limited.

PAGE 145

135 Though the estimated income elasticities for at-home consumption of canned seafood were the lowest among the specific product forms, the quality elasticity for canned seafood consumed at home (0.0945) was the largest among the specific product forms (Table 4-4) This may reflect the possibility that the higher priced and better quality fresh and frozen seafood products tend to be predominately associated with the away-f rom-home consumption market while most types of canned seafood are generally considered as products for at-home consumption. Other Seafood Expenditures Expenditures on shellfish were significantly related to finfish expenditures (X33) in a positive manner and vice versa^ (Table 4-2). This relationship parallels the work conducted by The Longwoods Research Group Limited (1984) who concluded that heavy (frequent) users of one category of seafood were also heavy (frequent) users of other categories of seafood. Similarly, purchases of canned seafood were positively related to expenditures on other seafood, though not at a statistically significant level. Expenditures on fresh and frozen seafood were both negatively related to expenditures on other seafood which is in contrast with the results offered by The Longwoods Research Group Limited. Caution should' be exercised when examining the parameter estimate associated with other seafood expenditures (quantities) in the different models due to a possible simultanous bias associated with these variables as discussed in Chapter II.

PAGE 146

136 Outlook for Increasing At -Home Demand for Specific Product Forms and Implications Much of what was discussed in the last section of the previous chapter extends to the discussion in this section. As mentioned, generic seafood advertising of seafood is minimal when compared to generic advertising of either meat or dairy products. As also mentioned, the majority of nongeneric seafood advertising is directed at increasing sales of only canned salmon and canned tuna. These factors alone indicate a disadvantage in long run growth in at-home consumption of most seafood products compared to meat, poultry, and dairy products. Though it is difficult to predict future at-home consumption due to the simultaneous shifting of all factors that determine consumption of the specific product forms, a couple of general comments concerning expected future demand for these products can be made based on the preceding analysis. In terms of at-home consumption of the specific product forms, the potential for long run growth appears greater in the fresh and frozen seafood product markets than in the canned seafood market for several reasons. First, the estimated income elasticities of at-home quantity consumption of fresh seafood (0.413) and frozen seafood (0.251) were about three to four times the size of the estimated income elasticity of at-home consumption of canned seafood (0.098), even though the majority of the nongeneric advertising dollars goes towards the promotion of canned seafood products. Thus, increasing household real income should promote more long term growth in fresh and frozen seafood consumption than canned seafood consumption. Second, the growing proportion of Black households in the United States should benefit growth in consumption of fresh seafood products since as the

PAGE 147

137 results of this chapter suggest, Black households prefer consumption of fresh seafood to that of canned seafood. Third, the analysis indicates that the current decline in family size will be especially conducive to growth in the market for at-home consumption of fresh seafood and equally disadvantageous to long run growth in the market for at-home consumption of canned seafood. Finally, the increasing mobility of the U.S. population together with expected improvements in terms of packaging and transporting fresh and frozen seafood products to the inland portions of the country should prove conducive to increasing sales of these products. Though the above mentioned factors suggest long run growth in at-home consumption of fresh and frozen seafood products and a possible stagnation in the already modest growth of canned seafood consumption, the seafood industry and its support groups need to be aware of certain factors that may impede growth of at-home consumption of fresh and frozen seafood. The main factor which is likely to impede long run growth in at-home consumption of especially fresh seafood but also frozen seafood relates to an increasing cost of the meal planner's time. An increase in the education level of the meal planner, an increase in the number of meal planners employed outside the home, and a decrease in average family size are all believed to be related to an increase in away-f rom-home food consumption. Thus, though increases in education and decreases in family size were shown to be positively related to at-home consumption of fresh seafood in this analysis, it must be realized that they will also result in an increase in the number of meals consumed away from home. Given the substantial proportion of seafood consumed away from home, especially fresh and

PAGE 148

138 frozen, the seafood industry and its support groups must develop new fresh and frozen seafood products that require minimal preparation and cooking time.

PAGE 149

CHAPTER V CONCLUSIONS AND IMPLICATIONS FOR FURTHER RESEARCH Conclusions This study was conducted to enhance the current understanding of socioeconomic and demographic factors believed to influence at-home consumption of total seafood and specific seafood products (fresh, frozen, canned, finfish, and shellfish). A better understanding of the factors determining at-home consumption of seafood also provides information which can be used in examining the away-f rom-home seafood market, seafood import demand, and ultimately the total demand for seafood in the United States. In addition to the "traditional" estimates associated with regression analysis, the Tobit procedure used in this analysis provided a method for decomposing the change in the level of at-home seafood consumption resulting from a change in any exogenous variable into two components. The first component measured the change in consumption resulting from increased (decreased) consumption among existing (consuming) households. The second component measured the change in total consumption related to an increase (decrease) in the number of participating households. For the total seafood consumption models, it was estimated that approximately 35 percent of the change in at-home seafood consumption resulting from a change in an exogenous variable results from increased/decreased consumption among 139

PAGE 150

140 consuming households. The remaining 65 percent of the total change is attributable to entry into or exit from the at-home seafood market by households. Among product forms distinguished by type of processing, the proportion of total change in consumption (resulting from a change in an exogenous variable) attributable to increased/decreased consumption among consuming households ranged from about 20 percent for either fresh or frozen seafood to about 30 percent for canned seafood. This breakdown of the total change into its two components should assist the seafood industry and its support groups in making a comparison of the benefits of attempting to increase at-home seafood consumption by encouraging increased consumption among consuming households as opposed to increasing at-home seafood consumption by increasing the number of consuming households. The variables used in the analysis can be loosely grouped into four categories. The first category includes variables consisting of demographical and seasonal demand shifters. The second category includes those variables determining the family structure. The third category includes those variables that constitute the social and ethnic structure of the household. The final category includes those variables relating to economic considerations of the household. Within the first category, variables denoting region, urbanization, and season were used to explain at-home seafood consumption. The two demographic variables, region and urbanization, were of considerable value in explaining at-home seafood consumption patterns. Northeastern households consumed significantly higher amounts of total, fresh, canned, and finfish seafood at-home than did households residing in other regions of the United States. Similarly, households residing

PAGE 151

141 in center city areas consumed more seafood in total and more of all product forms with the exception of frozen. Season was of significance in explaining seafood consumption only in isolated instances. Within the second category, variables denoting family size and the composition of the household were used to explain at-home seafood consumption. At-home consumption of seafood in total and for frozen, canned, and finfish were found to be positively related to initial increases in family size while at-home consumption of fresh seafood and shellfish were negatively related to increases in family size. To measure household composition, a set of variables reflecting the life cycle of the 'typical' household was included in the analysis. The results pertaining to this set of variables indicate that the composition of the household, independent of size, is very important in explaining at-home seafood consumption patterns in total and for specific product forms. For example, households in the younger stages of their life cycle, especially those with children, avoided consumption of fresh and finfish seafood products and total seafood relative to households in more mature stages of their life cycle. Elderly households, on the other hand, consumed the least canned seafood of any life cycle stage, ceteris parabus. Within the third category, variables denoting the race of the household, the employment status of the meal planner, the sex of the meal planner, the education of the meal planner, expenditures on meals consumed away-f rom-home the number of guest meals, and the household catching fish were used to explain seafood consumption. Race of household was used in the analysis to account for variations in tastes and preferences among households of different races which in turn would

PAGE 152

142 lead to differences in at-home seafood consumption. The results suggested that Black households had higher at-home consumption of total seafood and fresh seafood while White households had higher consumption of canned seafood, ceteris paribus. Similarly, Black households had considerably higher consumption of finfish than did White households, while little difference in shellfish consumption between Black and White households was evident. The employment status of the meal planner, the sex of the meal planner and the education of the meal planner loosely describe some of the social attributes of the meal planner. Employment of the meal planner was hypothesized to increase his/her cost of time and hence result in a decline in at-home consumption of those products which require relatively more preparation time. Though the results supported this hypothesis, statistical significance was noted only for fresh seafood consumption. Households with female meal planners compared to those with male meal planners consumed more seafood in total and of each product form with the exception of shellfish. With respect to education of the meal planner, the results of the analysis indicated that total seafood consumption and consumption of all product forms, with the exception of fresh, were positively related to increased education at a statistically significant level. Though consumption of fresh seafood was positively related to increased education, the relationship was not statistically significant. Expenditure on meals consumed away from home, the number of guest meals, and catching fish often represent a form of social entertainment for members of the household. The results of the analysis indicated that total seafood consumption and consumption of all product forms,

PAGE 153

143 with the exception of shellfish, were negatively related to increased expenditures on meals away from home at a statistically significant level. Total seafood consumption and consumption of all product forms were positively related to an increased number of guest meals at a statistically significant level. Finally, households who caught fish for their own use had higher at-home consumption of total seafood and all specific product forms than did households who did not catch fish. Within the final category, variables denoting whether the household received food stamps and household income were used to explain at-home seafood consumption. Receiving food stamps was found to have little effect on at-home seafood consumption. Income, on the other hand, was an important factor in determining at-home seafood consumption. The estimated elasticities of at-home seafood consumption with respect to income were positive and inelastic for all specific seafood product categories and in total. Among product forms differentiated by level of processing (fresh, frozen, and canned), fresh seafood exhibited the highest income elasticity while at-home consumption of canned seafood exhibited the smallest estimated income elasticity. Similarly, at-home consumption of shellfish exhibited a much higher income elasticity estimate than that found for finfish. Implications for Further Research A natural and needed extension of this study would be a parallel study of the away-f rom-home seafood consumption market. Data collected recently by the Market Research Corporation under contract with the National Marine Fisheries Service could provide the basis for the needed research. A comparison of the at-home and away-f rom-home

PAGE 154

144 seafood markets would provide information which could be used to study the interrelationships between the two markets and help answer questions which continually arise regarding various policy issues. A second extension of this study would be a parallel study of at-home and away-f rom-home consumption of specific species such as shrimp, groundfish, etc. An analysis of this nature was omitted from this study because of the relatively few observations in the data base used pertaining to any specific species within socioeconomic and demographic classes. However, since the data collected by the Marketing Research Corporation includes away-f rom-home seafood consumption, this data base is probably adequate for an analysis of this type. With respect to the present study, there are at least two research options available to improve the understanding of at-home seafood consumption. First, additional variables and/or a change in the specification of those variables used in the current analysis may prove beneficial. For example, religion has often been hypothesized to influence seafood consumption but was not noted in the data base in the current study. Similarly, additional interaction terms might be tested. Estimation and inclusion of a wage rate of the meal planner similar to that discussed by Prochaska and Schrimper (1973) may avoid some of the confounding effects associated with family size and education that were observed in this study. A second option that may prove beneficial when investigating at-home seafood consumption relates to the statistical technique used in the present study. With the Tobit procedure the question of whether to consume and the level of consumption are considered simultaneously. In reality, however, these questions may be addressed independently

PAGE 155

145 by the consumer. In the first step, the consumer may decide whether to consume a given good and in the second step decides the level of consumption. A two step statistical procedure that addresses these two questions independently may, therefore, have some advantages over the Tobit procedure. A final area of study which might prove to be extremely fruitful to the seafood industry and its support groups involves examining the relative merits of increasing at-home seafood consumption by encouraging entrance into the at-home seafood market by nonconsuming households as opposed to encouraging increased consumption among consuming households. Such a study would have to examine the relative costs associated with these two alternatives.

PAGE 156

APPENDICES

PAGE 157

APPENDIX A DEFINITIONS OF SELECTED VARIABLES

PAGE 158

Region (X1-X3) Those areas of the 48 conterminous states as defined by the United States Department of Commerce for the 1970 Census of Population. The four census regions and their states are Northeast (XI) — Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont; North Central (X2) — Illinois, Indiana, Iowa, Kansas, Michhigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, Wisconsin; South (X3) — Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, West Virginia; West (base) — Arizona, Arizona, Colorado, California, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, Wyoming Urbanization (X4-X5) Distinction of households in central city, suburban, and nonmetropolitan areas was based on the standard metropolitan statistical area (SMSA) as defined by the United States Department of Commerce in the 1970 Census of Population. All urbanizations — Composite of central city, suburban, and nonmetropolitan households appropriately weighted. Central city (X4) — Population of 50,000 or more and main or core city within SMSA. Suburban (X5) — Generally within the boundaries of SMSA but not within legal limits of central city SMSA. Nonmetro (base) — All U.S. areas not within SMSA. 148

PAGE 159

149 Season (X6-X8) Surveyed seasons of the year are Spring (X6) — Months of April, May and June, 1977. Summer (X7) — Months of July, August, and September, 1977. Fall (X8) — Months of October, November, and December, 1977. Winter (base) — Months of January, February, and March, 1978. Household Life Cycle Stage Young households with or without children (X9-X12) — head of household is less than 35 years old. Middle aged households with or without children (X13-X16) — head of household is greater than or equal to 35 years of age but less than 65 years of age. Older households (X17-base) — head of households is 65 years of age or greater. Seafood Expenditures and Quantity Consumed The seafood expenditure and quantity data used in this analysis includes food commodities with the first four digit codes equalling 4521 or 4522 in the 1977-78 National Food Consumption Survey data tapes. As such, food items which included a seafood product as only one component of the total item were excluded from the analysis. Seafood quantities were reported in the form brought into the kitche

PAGE 160

APPENDIX B DISAGGREGATED SEAFOOD STATISTICS

PAGE 161

>, c s~ 00 o 0 c H DO H 4J 0J s >h 4-1 3 o 05 s — / a u 3 o o u 0o 05 > CO c cj c i H •H 4J CO to B CO 3 •H > o rH u 0) e m o XI 2 0 0) 4-1 5-i -H o) s E -H D rH cn c c o o c CJ -/ s c U 'f; c o H 4J CO > u o 05 ,a o o) a) U rH 0) a, E E 3 CO CD 05 0) c O rH 05 U CO C 4-J O O C 4-1 30 QJ 4-1 n u c o o M o C I! c o •H CO QJ OS co oo co cn O O r~ CN < — i i — — I O O o o roo roo co m co \D CN CN cn i — I • • • • o o o o cn o m O in o r^CO H Cl H o o o o 00 — i CN CT> CN X co cn u x QJ 05 co XI 01 c O w 01 x: x c CU 4-1 05 01 vD M h in co CNl i — I i — I o" o o" O --,m CO 4-1 X XI c H w o u •H c 0 4J rH CO H CO CO Xi 4-1 N l-l ^-1 CD •H 4-1 3 E 3 C C X) c O ^ 0) 3 0 05 XJ C_) C/5 V; CO U QJ ZD C/3 XI X DO c >-i E O. 3 C/5 00 o 05 CO ^-N XI 00 ^ X — s-l QJ rH 4-1 rH C CO -H fa 3 151

PAGE 162

152 >, c DO o o c H DC H 4J cu ^ 0 CO a. u C o O u UH CJ Oh 0 — s co CO Ih c 0) 0 E *J •H 3 •H co E co = •H > o rH u u o c CO c XJ so CO en a Ih H 0 CD E •H H 4-> CN i — 1 CO co c > c c Ih c a 0) u CO o CO CJ Ih 1 — i cu p, CO E E Ih 3 3 ^ 0) CO CO rH S c w D o 1 — 1 CO u CO c c 4-1 c o o CJ c *-) >, 3 o CO aj 4-1 CO CJ a cu u Ih CJ c tu o >> u cu rl 3 3C QJ CO 3 O ONCNvDHOONOMNvO -— I < — I I — I < — li — fNl i — I 1 — li — li — I oooooooooo o o oiDt>H OOOrHOrHOCNOrH oooooooooo' lOOOiOHinirivDHOMri mincninr-^.— iinooNON OOOrHOrHOrsioo oooooooooo' L.O — i CO — 1 i — 1 X r— 1 X — \ X a CN C c CJ o i i — l e d) cu M 1 — 1 1 1 X OJ Ih Ih X) X) X 3 — n r-H X -3 — i rH •H c H H •H X c c OJ H x X! u c CL) cu Ih X u u c Ih c ^H ~ > •H ca H CO CU CU XI XI CO E CO B rH H 01 CJ cu CJ DO 3 rH •H rH •H -a -a — XI C Ih 00 Ih CO Ih o CU cu cu •H ca c Ih c Ih 00 DO CO CO CO E H CO H co CO CO ca co CO CO s >, cu 0) CJ cu rH rH (30 CO 00 OOr-H rH rH i— i (h Ih 3 c c e -3 -a -a CJ OJ =3 3 p '3 xl "3 "3' o O o o H •H Mi H rH >H >H >H W W a\ CN cn oj aj co — i ca CJ X X 3C 3 CJ Ih O 4n cu o -H X 3 a X 4H -H u o cc ca OS

PAGE 163

153 4-> u •H 3 •H 4-1 CO CO E CO v~C H > o i — 1 u v O a CO 0 £> 0 CO 4-1 1h -H CD E 3 i — i co c c o o c eg c c •H 4-1 > u 0] 0 o CO a -i a) u c CD u rH a a, c a a S CO CD i-H o O. co E E u co CD Cfl co c O o rH CO u CO O c c 4-1 o c o e 4-1 CO Oh E as o o M O > CD u a CN CO r-H U~l CN -H d d I — I CLT* cn d d 00 CN cn • • a o cni r~ o d co m r— • o o CN CO cn <}• m d d ON > — I cn r-~ d d o CO • • o o o CO 3 c O u O rH CD c c CO CO -, CO J3 co o 10 CJ o co CD o rH CD o >H z 3 >^ a CO E CJ S-i 1) c G co CJ E 0 X CU 01 O O rcn o o O 3N o> o d d CO CM cn r-~ o o d 00 CM O CN a> o d d cn ^ CM CD x; co w co CD 1—1 CO CD E rH CD CO tn 2

PAGE 164

154 i. o 3 C H 00 H P OJ E ^ y> 4J 3 3 CO CO w c o o a c CO c o •H 4-> c > so co x o co CO 4-> c U -H O 0) B -H cd > a Cfl XI c c c CJ CD er Pi Cfl B E u 3 CO CN 1 — 1 o CO co cn c > Sh O DC 01 U a L5 CO OJ 3 I — I CO > c CO CJ E 3 o I I I I I I cn o CN 00 CO CN OA IT) lO cn cn cn cc in N •H E cd CN T3 LO rH CN O X x: ^> CD co -a 3 cj o >H x; co C CT •H CO 1-1 In 4J o o E' H C c CO cfl cu E 4-1 O c o a u 3 o C I Cfl in CO CU >1 CO a E cn CD •J CJ I-CN X CO cu E o u CD E o X! E O Ih CO CO CU 00 CN X CO IH a o Q I I I I I I r~; — i T— i o O o o m 0 ) cn O cn X •a CD lr re CN ^ C CO CO CJ CO >: Ih V4 CO CO CO 4-> i— 1 p— 1 rH 1 — CO O o Ih Q — 0 m T3 T3 CJ d C X a co co cn CU 3 3 E O O 0 x X! u H —

PAGE 165

155 ^ S-i O C O O Dh o U CO O C 00 -H QJ 4-> D co m c o QJ Ih Qj cn co w c o u c 0 CO B O 4-> -H 4-i CO > u o X o tfl u c sH 0 GJ E •l-i H OJ 1 tfl w c > c 0 o c 0) u co — o [/! OJ 3 S' a, tfl E E Ih CO OJ CO tfl 1—1 C o i-l EG u CO C c 4J o o o u c J In O co OJ CO U 03 OJ 3 — CO > c ra 0! E CO 3 u o < I I I I I I CO 00 a. on CO CN r m X X 2 Q o x >* O 0> X: d c c x X X' X r-^ o X > 4-> H c j X c X a nb CJ N x t— 1 •H rn cn (0 x: >, co H QJ to a> •H U E o E 3 u a a 4-1 Qj 14-1 H 4-> cn X ~3 T3 -a C -< en c c C C o X QJ 0 CO CO C •H CO to X 4-1 QJ OJ QJ >H tn QJ u E E (/I CO O 0 — 1 C U u Ih 1 — 1 i — OJ c QJ o c u I-H M X X X QJ c i-X, QJ Q X rn tfl ~ U i QJ rH GJ X3 x: CO 4-1 ca Ih X c r H c •H Jo TJ CD 0 •o •H > O CO c a. u CO c (U co tn -a QJ U a £X x: QJ E-h M

PAGE 166

156 ca p •H 0) c -h on c — oo tu x c u •H o co cu 4J U o w 1) CJ w ft w X OO <^ W c o E CO X oo x c 4J C •ri£ O rH 4-1 P 3 CD Cu •H CO OO X T3 TJ CD C H C CD P CO -~y P X PJ U ^ 3 P o -a -H X CD rH ^ •p •H -r>> u X N rH CD CO ^ ^: ax fu CD X O 01 P ft uh o u CO •H •H p o co o •H >^ p P i— i Oh CO E p c I CO co < p •H XI o u to Eh CD CJ P 4J P CD CO O E E -H CO •H 3 P P) CO CO W u Cu a X >, P o DO CD CO U co c o -H CO CD W 00 O IT) O O o o X o o o> co o N C CO CD CO CJ CD X X CO CO P X CD co CO p 0 d o •H P) (0 N •H c CO X! P — i oo O CM | rH O I • • I o o oo m r~cn co co i rrH O X CD /-n CO 1 m co x x c co X p X 3 U CO 2 c w CO cu 00
PAGE 167

157 4-> Z) C "H 00 a N c cc 00 fc, ~ u .5 • — CD CO Oh C X H u O W C! U w CO PL, CIC X 00 CJ 4-> 4-4 cu cO E E •H CO •H a '4J CO CO o CO cu CO u CO CM CO 0> c m m CO CN ST CC -J 00 co i — 1 1 — 1 1 — 1 i — i i — i CN i — 1 — CN CO o i — i 1 — 1 \ m m CO co CN CO co O i— i m o o CN cc CO CN CN o i — i CN i— i CN co o cn o O OJ CO Q c CN CC m CN i — i CO CN r— 1 i — I i — 1 o u >, u o X QJ 01 3 O ~ CD 00 c C CD •H S-i CO X) ^ s o CT> i — I X -o X H CO cc: 4J 00 -H CO -H C X C X D U P U O O i— I S CN X c CD u X3 X 00 -H 00 -H c -c c x: 3 O D u o o >-> >H CO CD i-H t-H X 00^ c H C CO CD l-i -O T3 CD t-H 00 -H CO X U CD -I o -o -d 2 •H T3 CD •H U U CO E T3 T3 CD t-l CC -H CO X CD -H O -a \ T3 S in i — i CD X i-— I w oo c c •H CD CO S-, X) X) rH CD -H 00 X CO o CD X — 4-1 X) -H X) 5 CD ^ CO co —l x X w XI CD CD rH -H 00 C Sh •H CO CO E >> >, CD X i — I i — I U U CD CD XI XI ~ — CD X •H w 1h >H C CO CD E x) X) i — I CD -H O0X CO u X! -H X) S

PAGE 168

158 CO 4-1 •H '-n CD C -H cc 3 CN c — j a cot, c u •H E ^ a 3 0 co ex. C X •H u o w X CD o. w 00 00 CN ON CO 00 CO I I O vO I in I m i to CN I cn m CN CN CN C7> cn co O '/; i — i — i CO O CD X X 314H o CD M CN M GJ Cfl X 3 4-1 3 X. 0J rQ C 4-1 o (J > H CO •H Cfl cu H GJ CJ O u c u >* M CJ C3 CD Cfl 3 c jC Cfl •H CO 3 CO CJ rH Q) CN CO X CO w jO CO CD O >-z. u o 3 3 3 rH 3. 3 CD E UH c 3 o E >, c — E CN CD CN CO X CO CO cu o >2

PAGE 169

159 G o U cn i (JQ Xi ra fa Cfl 4-1 •H '-v 01 d -H 00 3 N C w co oo fa JC c 0 -H -a 3 0) W Oh •u c x u o w 01 u Cu w X OOfTj w d o e co MX X re H c ra •H C 4J 4J HX> o iH 4-> 3 CD CU co 00 X T3 0 C fa u CO w 4J x: fa CJ V (J , 4J — •H ^> +J •H -r u Xi N o CO ^ o-xi fa X o w '-. CX o •H 4-> O C •H J AJ ra £ u i to < u m QJ a 4-1 ~j QJ ra E E T Cfl •h a M ra 'X Ph >, o CO r-~ rI -J I c r-H 00 o co ^ CN cu x co w CO XI CD w rH CO CD 6 — l CD CO fa £ vD m c c d i 00 m CN c 00 — i o o I CO o so o o CN Q o o> c o erst X 0> CO X St o 0 N CO fa -> •h x m w CN s-i u x CD X) CU ^> XI rH XI G O S x) 3 X 3 CD d cd d *h CO CO i — I 3 i — 13 CO O CO cr 44 X 44 0) O O S-4 CJ B c ra C ra CD o c 0 co u 3 -a fa CN X Cfl t-l co Cj CO CJ CO cj 3 CJ c I— I o d ST O sr CN st O o m CN CN ra o E IW O Sh CD X) E 3 25 C o o d i o CO Q O d i CN st O I n o i 0) E O jC E o U ra ra ra x ON X Cfl ra o Q

PAGE 170

160 CO u H CD c DC 3 N c re 00 fci X c u •H o 71 a. 4J C >^ u o w CD u o, X DO re w C c 1 a oo x r — i c a H C j 4-1 -HX> c 1-1 4J 3 a> a. co oox •a CD C W o 4-1 U h o m >, -a *h CD ^ 4-1 -H -H U 43 N CD CO w x o U •H 4-> O O -H 4-1 -U a co E u >. I CO 4-1 < u co CD CD 4-1 4-1 01 CO E E CO -H U 4-1 CO CO Pu CD >> o CO CJ 4-1 CO u in o o on CN o cy oo co CO r-H CN o 5 O o o i & CN co Cj CO X (-1 co co re CO c G 4J rH ro • — i rH X cj a o O 4-1 >-i a o a c U-l T3 ^> a o C c •H XI CO re CO 4-1 M u cu a 3 rr CH G o o CO Ih O X! X! 01 u H 4-1 c C H m Q m O o i 00 X O Q c O I m CO O c I IT! o I c^ oo O o I m rm m i CO c^ I CD •H U G CO CO U rn CM CO -a — CO 0 X c c X o CO C3 -{ CO 01 o CJ CD CU u 4-1 G G N co CO c o 0 •H I — t CO u u ro M H *J c e CD c to M h X p c 4-1 0 o u o a u sO *— > X : JJ O CJ 4J — II o u D o CO c o (L) ro * i o 4-1 X E — Ih CO CD J CJ CJ eg £ )1 4-1 CO c X c o •H 1 — 1 o tfl 4-> — CD U CD CD 4-1 4 — 1 co > CJ u o a CJ UJ si .a [fl H re

PAGE 171

161 C CO 3 /-n C -H CO CON x c ^ U -H Cu T3 3 — Cu w x oo<^> H C o E •H 3 •H co oox c: rH C o CO -H c u 4-1 4-1 •H O rH 4-1 3 CDrD 4J CO oo^ •H X> CD C CD4J CD U CO w c 4-1 X W CO O 1 3 4J O T3 •H X CD 4-1 3^ >, u X N rH CD CO w £i CL.X) Cb CD X c CD w s_ c_ 4-1 O o CO •H •H 4-1 o CO o •H >l 4-1 4-1 rH a. CO CO B C >, 1 CO cn 4-1 < 4-1 •H X o u 0] H 0) CD 4-1 4-1 i-i , u c 00 CD 4-1 CO u o CN O o i u V, X co 4J CD 4-1 X c H 4-1 C o U CO cu c •H CO u c rH 01 cu u CO CO Xi X! X a N >H s 4-1 4-1 4-1 H 4-1 c M M 3 cn C c f— t •H O o O o in Reg ;z 2: C/] Urb u o CO CO oo co — I 00 rH CM rH r-~ oo oo rH i— I OS | rH rH O I CN
PAGE 172

162 en 4-1 •H CD c 00 3 ^ C •H CO CON JZ C ^ U -H &H T3 O CO -~v 4-> C o oU GJ ft w X W amon co x rH c 03 H c i-i •ri o H 4J CD X CD T3 QJ c c a M CO w SZ w u CJ o <^> 0) C£ ft C E CO c w si M u 14-1 >. 4J 13 •H a 4-> H -H u .a CN CD CO ftX> tx-i X O w u CJ •H 4-1 o 0 •H 4-1 4-1 ft CO E < l-i CO 4-1 0) CO EE iCO -H CO. U 4-1 CO CO D-, CD >, 0 CO 01 u cC U en CN CO X , u a <4H •H o si o en 3 O I I o OJ CO CN in m co 00 m m o m vC o m CO CN o C vf in CN co m CN a CN CN — i o O c o O o O o o I I OJ CN c^ CO rcn or x CN 0> ~ O — i i — i — i — i i — i CN i — 1 CN O o vO cjo o CO ^r i-H r-. o rX — 1 CN C cn cn CO OJ cn CN C — i CN • — i sr x St CN O rco X , >> CO H CO CO H —i CD o CJ X! CD £ — rH H •H -H CO H r-H c rH H 4-1 rH 4-1 1-1 c -G c X! c -C c J3 T3 or or •H 03 H o CJ 3 u U u 3 U ~ -a T3 TJ u a o o o 0 •H •H — 1 rH Sh rr W

PAGE 173

163 X p H a; d CO 3 c •rO0N £ C w U H t-, a to ^ ; i c *'* u o 61) ft W X 00 CO W d o E CO •H oo x c •r-i c P H — oi,n m oo O c oP CO w x: w a) O CO BO d E cfl O J3 P U <4-i >, J a •H [l, X O w U ft u H p O 0 H p 4-1 ft CO E P >> 1 Cfl 4-1 < P. HI a 0) p o co E E *H CO •h oa P p CO CO u c DC 0) 4-> CO o o m O o I I — i O c co rO LO o vO 00 i— I IT) 00 vO d d i i 00 O I • I -7 ~* co I 00 I • I CN o c I I m CO i — I CN CO o i i X CN CN I O I m o I vO I • I on oo \D vO r-m l co oo 00 I CO I L-: C c co 1—1 g d i • • a to 3 Cfl ft d 4-> 6 d Cfl o CL> p a Cfl p d o c 01 13 p a, 00 CO O co — 1 — i CO O sr. cu x X ja m Q 0) to I ai p r j !fl •H CN CO pco X CO IP X CO CD p ^£ CJ £5 s — 43 o p o u > P H x: CO H CU rH CD 0 O 00 a O U O CQ u z. cfl ss

PAGE 174

164 to A3 c or co CO ISI X! c v-' o H tu 1 • -G o co '-^ 4-1 c O o oU u w P. W X O0 co w C o g 5 •oo x — 1 c •H c 4J J H o H 4-J a CD XI W 00 '-n — c oo Ih CO w 4-> x: td u u (J .ro o 00 c c E X a C W — u u 4-1 •H X ;T3 T3 -H CD rH ^, 4-4 -H *" CD CO w O.X) CJL4 X o 4-> O O -H 4-> J-> O. CO E u >. I CO -u >-i CO CD CD •P 4-> CD CO E E CO -H r* 4-1 CO CO 0* CD H 0Q Ih c DO CD 4-> id m u C c CO CO CD E o CN CN CO I O I • I CN O* cn m CN CO On O CO vO I vO I • I vO CO CN CD X CO w CO XS CD w rH CO CD E rH CD CO PL, g rH o o o i o o I o> a CO X o I CN CN c I CD N H co co -3C CN •H X rH cd -a XI I — I E O d x: C CD CO rH P CO o 4-1 XI o H r J Q O o c c X rH CN c IT) O m in CN u x CD ^ E -a c CD Ih CO 3 cr [fl C CO CO CD E u_ O c c CO u W O CN X V; Ih CO CD CO ai £ in CD CJ 3> 5 cn c cr X m co 00 m CN CN CN X CO ai E HH C Ih CD J3 O Q c o i CO o c c I X CN rCN i CN X I CD E O X E O a CO CO Cj s: X CN cn Ih CO O Q

PAGE 175

165 ID 4-1 •H 01 c 00 D /— v C -H CO DON £ 0 -H tin TD 3 01 co 4-> c u o oa. u w c o E a OO X — i re CO •H a 4-1 •H o rH 4J 3 oix CO OO'-s t3 o c oQJ u CO w 4-1 X U u o O cxi o CD ft C E re 0 W X vu <— ft; >> 4J -a H a —i / N 4J •H •H CJ N O re X c W pr u •H 4-> O O -H 4-> 4-> O* CO E ^ >. I CO 4_> < 1^ CO CU CD 4J 4-1 CU CO § E CO -H U 4-1 CO CO cu •H 0Q >, o DO OJ 4J a U LO CM co Q O in i — i o o s* o o o I QJ sz — st 1 O 1 • o o c 1 0 •H 4J — M J_J E/3 — C [j CD 4J o c vO 1 o • I— 1 o i CO — — o 4-> 03 E — | ^ co CN 0> • — i 00 CO CT* CN 00 CD •a co m CO i i •H CO — CO 03 •H X C r— i m rH rH ~3 CO C \0 o CO CO CN MH o c O 0 1 1 w •H c j-j CO u CD re E 1h a> — QJ r i i — X! X! OT 4-1 ><; E0 4-1 0) CO CO iH CO X u CO 0) H re re re o E u E CQ H — i 1h re re X i— 1 rH X CU u 01 O o 4-> r I rer CO u a -a ~r en 0 CO o o c c re X o X <4-l T3 ~ CD o re re V cu o c C u H re 3 re a CO 4-1 01 01 cu a CO 4-1 i— 1 0) Cfl co 3 u E E N CO 73 C co 3 re cr re 0 0 •H c co > E c o CO u u u Cfl u re o x jC QJ c re QJ 0 to o 0) u H 4J i— i I— ( CU c oo. X! Ln In Ot Co H CO

PAGE 176

166 b u oo o o b H 4-1 00 01 •H E u 4-1 3 o M CO do CJ on U u o X: C ni E 4-1 H 3 H 4-1 CO E CO C H > o iH U u 0) B to o O CO OT 4J c U •H o OJ E •H E H 4-> CN 3 — I CO CO B > B o u c b CD CJ CO -o o cn oi 3 h H 0) c CO E E l-i 3 CO cu co co E C 3 O rH CO CJ CO C C O O O C 0 4-> 4b aj u u 01 ft B co 01 c o •H on 01 cn r- h i — I CO CN CO o o o o cn n m m m C CO CD CO C_) CD x: x: 4-) 4-J U U o o 2. CO CO CJ to to XI m O cn h > m to 4-1 X XI c •H — > o CJ H c c 4-1 rH CO CO CO XI 4-> N '.u CD •H 4J 3 E B c C XI B 0 a CD 3 O to -Q CJ cn CO U Cj !=> X X CD x CO OO IC CD •H E u E a, 3 CO CO • co X 0) rH C CO -H Cjh 3

PAGE 177

167 >-, c in DO o o c — i DC •H u o u 4-1 0 CO X > — / c_ U C 0 0 u 4-1 u o, C CO X c o c •H •H 4-1 £ CO c •H > o rH u 0) a m o X: p0 co to 4J c •H o o E •H H H cs — co X G > c O u o c o v — X X) o CO ai =•3 (-4 rH OJ cx M E E u CO CO [fl rH E c w 3 o — 1 co u CO c e 4-1 C o C CJ c 4-> c CD U u OJ p. c co ai u >> u o -C OJ CO 3 o EE 00HC0\DcO H r— 1 1— ( X 01 !-i -a X (-1 T3 T3 rH X c Id IT i ] hi c c 1) •H XI u c 11 o u X u CJ o u Ih XI u XI u o a rH _o X H -a rH rH H o 4-> H V n •H H XI •H ^ a> •H XI X u u CO u u T3 X) cfl u 33 01 OJ 0 t— t X! c X 4J t— 1 •H rH •H X o 4J H on 1oc >-l •H C H 3 >H — tH CO •H cfl O -a CO E CO E — H CD CJJ a Gj 0 0 U r-H •H i — H T3 TO XI e. U 00 H oo (h QJ D 01 a •H CO c >H c u 00 00 00 CO to E •H r; H a CO CO CO CO CO E CO E til OJ a V rH !— I CO 00 CO 00 rH rH — 1 u H c a c C a COCO CJ 3 3 3 •o -a XI TS cc CO o o o O •H •H r-1 i— 1 X fw W W LP) i — i CT* CO O O rH O O o o" O vo h cn Of CO CM 00 O rH • • • o o o O i—i o CO CN c? OO o o o o" o CM O 00 m CO rH 00 O rH o o' o 4-1 c D — 1 c o OJ ft X ON X Cfl — i i — 1 CO CD X X XI H 14-| a VCxC O *j O u •H x: CO OJ x: 4-> rH CJ o OC CO OS

PAGE 178

168 >^ c Ih CO o 0 c H 00 H 4-* cu — "V Ih o CO w — O, o c 0 0 u 4H u o cn c cu o = 4-> H — s •H n co S CO G •> C H u x — o a M o _o Z O w m J C •H o QJ E •H rH 4J Ol — CO co e > c 0 u o c o u Cfl o /• — s Cfl Oj Ih — QJ Cfl B E 3 cfl -^N QJ CO m E C 0 H CO u CO c c u c o o u c 4J >, Ih o DC CD CO CJ c o u Ih QJ a. c CO I 1 r: 0> CO • • o o X) CN r~ cn O CTv • • o o m o cr\ d d I~ CN o d d 00 00 vO — I o o •<-] vo cn r~ cn r• o o cn o & cn vo • o o cn vo CN r-~ • • o o CJ Cfl • cn ex C E ? ed 0 *J co o r -a o o J3 >4-( o QJ Cfl — i C) CN '/I •H CN Efl Bl CO 14H X ra QJ ja > 4-1 •rt Cfl _C cn CJ OJ o CO CU o U >* 3 CU Cfl as U Ih O c C ra co CJ E 14-1 O HJ C o E >^ o i — i a E w 00 f-i cn cn o o cn • o o o o <• n • • o o m m \o cn m • • o o cn co —I o d d cn r~O ON o> o d d 00 CN cn vo o> o d d 00 CN O cJn CJN O d d • • u QJ c c co —1 a. cn CN a r— : X CO CN QJ ra ra CN cn CJ J3 X co E Cm rH 4H ra tu CO O E i — 1 CJ o QJ Sex bH

PAGE 179

169 >, c M x Q Q C— •H cc H 4J o rH Ih u a) c oc c JO Z o Cfl E m 4-> c (-i •H o rH o £ •H CO X X •H 4-1 3 DO o CN rH CO 4-1 CS cn M C > u c o Ih < m cn CN c c — i CJ obs in o Ih — 1 cj cu CO E E 3 CO (1) urne w C o t-H w u 03 s c 4J o o 0 C -> o M 4-4 a u en 3 — > c o I I I I I I o o> O X) O i— i CN rH h CU CU X> X) E E 3 3 c c CO CO O O CN x o o O o x x cn X in CN CO 00 un d c CO CN rH .V Oh c )in so rH T. .c cO H ca 0) CO E r— \ H e a 0 CO E u c J 'CJ 14_( n-l X Ih O CN r O X H o >4H C ca co O o (-1 Ih H (fl e 11 CO CO 4-1 H X rH a) CO CO 4-1 E ca r-i E u CJ m 3 i— i O 0 3 a CO u -a CD c PJ CJ n I I I I I I O CN r~rH • • rH rH cn in o CN H x o X n !N X o X X -a CJ Ih X CO CN 3 X 3' in !f, en Ih u M CO rH — I rH H O 0 T3 "3 -a -3 c l3 C3 a cn X 3 3 C 0 X X iEh

PAGE 180

170 e u CO o o c H 00 H 4-1 o s ^ i-i -u 3 o i-H iu CD c CO 0 x z o CO CO 4-) C U -H O O E -ri • EJ -H 4J 3 i — i co • co C > c o u O C CD O w CO X o CO Cj u — CD CX to £ B M 3 co 01 ns CO c r-i 10 u CO e c 4-> 0 o 0 u c 4-> o 00 0) 4-1 CO U CO CD 3 H ffl > c CO CD E co 3 4-1 O I I I I I I o o> r-l CM • • cm o o o~ >, 4J /— ^ •H CN 4J ro C X CO nb CD N i — i •H 3 co co CO X >, CO — 1 CJ CO cu H G u E a M CO CO 4-> CD u CO H 4-1 CO X) a -a O X c C c C o a> O CO M Q, H CO CO X 4J GJ CD QJ >H CD O E E co CO TO O o — 1 >, (J u i— 1 t-H 01 a c o o J^ 4-1 i—i 1— 1 jC CD c CD o 3 U3 Qc co -o I-I o X 0) CO 0 4-1 O a E 3 O o CO T3 — I o x CD co 3 O XI 4-1 c CD u u CD a. CO CO — CD 4-1 CD u ft O i-i O CD — i 4J a 5-, •H _3 o C XI 4J CO -3 a — 3 0 r— 1 — a, CO' H ro 3 CN E s< 1 — i W X OJ >. s oo CO -J CD c rH CD X CJ CO u •H CD S-< c^a > C H CO t. CO _Z o •H X o — a 'X j-j • X X Cl) -J CO •H ca 4J c — CD U i, ^ co OJ •H Dh u o c CO co CO J3 CC 4-1 SJ u — 0) X X CO 4-1 ca CO •H CO X a 4-1 0 rH C 4J •H i-i C o ex CD o ID' Ih •H ft > C 00 u c •z•r-i 4-> cfl c 4-1 o CO CO -o CD CD Ci X a H u

PAGE 181

171 oox cn rH c o CO •H C u 4-J -HX> 3 o rH r-v 4-1 4-> 3 CD Cl, •H CO oox -a -o CO c u C CD U CO w CU 4-1 x: H a. u 0) o ™ X CD 00 0) a. C E X CO o a W x: p H O <4-l o x: 0) cn >> 3 4-1 o T3 -H XI co H in 4J •H (JJ C *rl 3 N 00 tin c y •H E -n — 3 # CU CO P-, -J C X H o o w X CJ u ^ co a w 00 > a o E 53 CD i— I '-s 4-i -H TU XI N CD CO w D,Xi [n X O W ra. u •H 4-1 O O -H 4-1 4-1 0. CO E U I CO 4-1 i-i CO CD CD 4-> CD CO E E u CO CO D-. CD H 3 o DO CU 4-1 co u c o •H CO CU Pi CN o , m c CO 4-1 X XJ /J c •H v-^ v — y o u •H c C c 4-1 rH CO Ih rH CO CO XI 4-1 CD N u u CD •H 4-J 3 E cn C C XI C _CD CO CD 3 O XI cj co u c c X CO JJ CO 3 O 3 3 I O ~ I 3 3 d i cjn o m CN CN r-H O O rH | O O O I 3 O i — > i — i vO 00 o -oo o d o i i vO r-X x CU cn co xi CO a • 00 x CD Ih E Q. 3 CO CO

PAGE 182

172 r 0 4-1 •H CD c •H nn \MJ a N r— 1— rrl oo fci J3 = O •H E ~ 3 Oh CO c X u c 01 u DO CO W amon oo x — a •H E 4-1 HX Q — QJ D_ CO oo x -a CD c 01 Ih CO 4-1 — u 01 u SO Q c E ca 0 x: Ih u 4-4 >i 4-1 •a •H a i— 1 — ". 4-1 (J X N CD co Oh X o Ed ft U •H -i O O 4 — 1 4-1 C_ CO E u >i 1 0) -J < >H CO a 0) 4-> -J a co E E CO •H B u 4-1 a CO Pl, o V-. c M c 4-1 CO C~ CN CN ON m c o> on CC X — i — r 1 o o o Q c a o o c c o S o c o C o o o c 3 c I I I I I 111 r l 3 1 — 1 CN vC m CNI in 0> o CM 0> C co 00 ri — i m r-j — c cr CN r~i — I i — i i-H o c O o O O c m 00 CN r , u 01 HH H 0 x 01 W 3 0 3 ON X CD w i — I oo c C h C 00 C U CD C CD CO U -H ^ E T3 10 -O rH rH CO -H 00 -H c j= c x: O U D U o o x: -H ^ S CN i — l -a x 0) ^ •H >-< c S-. CD CO E -a i-H 00 -H C X 3 U O CO CD — r-l X OO w c rl C CO CD kH t3 -a CD r— I O0 -H CO X o CD — I o -a \ X) 3 T5 CD H IH Ih CO E TJ T3 CD i — I 00 -H CO X u CD i — I o X) m i-H CD X i — I ^ oo c c •H CD CO Vh -o -O r-l CD -H 00 X CO U CD X rH 4-1 T3 -H X) 3 vO "O i — I CD X •H w U U C CO CD E Sh -a a i-i CD -rH 00 X CO o 0) X rH 4-1 -O -H a 3 •H CD ^ CO h co — l X X w V — y -a CD CD i-H *H CO Ih C U H CO CO E >, >^ rH rH ^H U CD CD -a -a rH rH CjJ CjJ

PAGE 183

173 to CO CD C M P N C co co Pl, x: c — c11 Cfl J C u o W 0) u ft X M CO Uj UOU1B oo x 1— 1 a CO H 4J C — 40) Oh HI OOX -a OJ c w o M CO 4J x: w O O a ro (D CO ft c s X CO c w X! Ih o 4-1 •H CO -a -h CD rH 4-> -H *" O XI N CD CO ^ CXXl X o ft u •H 4-> O O -H 4-1 4-1 a. co E (-4 >, I CO JJ < CO 4-> 4-> CD CO E co co O-i 0> h a c CC a jj cc U LO co cm en O 1 o o o I a i VD CN oo oo O 00 O O d d i i H H CO co cn O O CN oo oo r-o d i CN c Q 5 d i o o o I o o co o I O I O I • I o I • • m • • ft E c CO OJ 4J -a CO c o CD ~ ft CO On W o K 1 — 1 i— ( a o a X X XI 4-1 o cu CN co CO X CO m 0) t-i o *J Cj u > H x; CO •H K CD 4-1 rH a o o U O pq u >CO CJ OS CO CO O 00 I i — i r~ CO o CO CO 3 c O c o (0 43 DO 3 CO o >— i CD CN CO X CO w XI CO CD O X 52 in 01 C C CO CO CJ E — o 4-1 c CJ E o ft E UC o c a o i C 5 d i CTi > — i r-O co z

PAGE 184

174 CO 4-1 •H ^ o C f DC 3 N c CO CO tu C U •H s ^ 3 CO Oh 4-J C X u O H o U w c w X 00 1X1 W c o E CO •H 00 X H C CO •H C 4J 4C — 1 4^ 3 ai (0 COX "3 o a w CJ CO w 4-1 x: w U o o ^ cj 00 ft c B X CO o w JS u — 4J T3 •H 0 iH 4-1 •H "r CJ Xi N O CO O.X1 IX, X o w u a. u H 4-> o o H 4-1 4-1 a CO E In >i 1 to *J < W ai ai 4-1 4J CJ CO E CO (-1 4-1 CO CO (X o X ft: •H >, Ih O CO a 4-1 C3 CJ u CD C C CO CO a E o X 0 CO. CO O I O I • I o — I o o < — i CJT\ O co CO O co CN O CN CN o o co O co o CO o co ro o o I C CN •H X tT| CN u X CO CJ -a cu CN 0) XI rH X! X co CD E o E *a CO N 3 Xi 3 0) H C a C S-i CJ cn w CO — 1 rH i — 1 3 CO aj >^ CO o to ~ E t— i H 4-1 x: 4-1 CO a eg H O o Ut H H o a co C o o m m co CN o CO Q 5 CN o o c 4 c o a> o CO o o I rCN O c d o CN co i O I c c r-> CO CN 1— 1 X ex om H co X CO H o CO E E o O CO E u CN "4-4 O lH X O CN co MH X — 1 O >. c CO CO CO c CO u IS u •H CO E ai CC to 4J X r— 1 CO CO 4J E CO u CD CO 3 rH o a >* a ca Q T3 CO X u

PAGE 185

175 cd — u w 4-1 •H /-n C "H 01 3 N C w c •H s D $ CD Cfl Cj_ •U C X CJ O W CD U a. W x DO w c o E Ed •h MX X i-i C cfl H c 4J 4-1 H c rH *J P 0) P_ Efl DO X -3 U c w 11 Ih CO ^— 4-) x; w u CU U co a 00 c E X IB 0 W Ih u 4h X) -H 41 H ^ 4-1 -H "H U J N CD CO a, j3 [i, X o a. u •H 4-> O O -H 4-> 4J n, co e u CO 4-1 < S-i CO QJ CD 4-1 4-1 CD CO E E CO -H U 4-1 CO CO Cu CD Ih c DO u *J CO U 3 o CM o I o I o m o o o Q o i o -sr 'CO co CO O m o O I CO co m ON in o >, r-H CO CJ •H E u H CC 0 ^ — \ o >H IH CO CN co -d X) 00 o X c c c X 0 o ca CO •H ca ca 4-1 CD 1) CD CD u U F E M CO ca cc CO O 0' •H i— i CO u u u CO u — i 4-J QJ c c QJ o Cfl *J t—i — 1 sz co CC J n Ot Co o o CN UCJ c m g o 'w LOO O 4J IT] — — fl 0 u CC u CO C c o CU • — i u o G > 1 r o 4J 0] E i — CO CU — cr CD G T J, H 0 CO Cfl — X •H >C u o — — ca CO Co m c c c CO •H C — ca CJ CD CO E QJ Cfl CD o o Co CD CO -CO -TO CO H t— CO

PAGE 186

176 X *J H G C CO c •H CO OCN JS C c o •H Ci, CD E • N -a o a CO M 4J C u O C Cj O w n-< ft Ed o X 00 w c c o o E H 3 ft E •H to MX c iH c o CO •H = u 4-> •H O — 4-1 3 CDXi CO H -a cu c Cr 4J (D Ih CO w S 4-> co U cu 0 00 cr ft c E X CO O H 03 u rH u O C cd 01 >% rj 4-> o — •H G Q iH ^ 4-> •H -H >> CJ iH a CO w Jh :s ft , i CO 4-1 < u H -Q o U CO H o CD 4-1 4-1 M CJ CO O E "H IW •H CQ Ih 4J CO 3 CO o C_ 0) •H H X rH r< o CO a 4J CO u co c o •H CO 0) CO CO r-~. r-r— i O o o 3 O o I o o I o o CO rH | O O I c o I o o I o m ct\ o co co r^i— i in co co in >o o> CM CO CO ON -sT 3 o C/2 ^ C o CO N •H C ra In E3 o Q 5 o CN a o o o cn m CN o Q rH o O I o o rH in o m co , m CO 4-1 X •H r J c o — ( ca u CO u Ih Ih CD 4-1 3 E C XI C CU 3 o u C/] 2: c c CO ca CD in rH O co in o o o o o o i C O I O CO CO CN CO rH CO O O o c o o I o I GiOOiH \D co rO H E ft 3 GO C/J

PAGE 187

177 1 l •H — CL CO CO X! c ^ O •H E 3 OJ cn ^ 4J c o o* 0 G CJw X uc amon •H X CO X rH C B-ri C O 4J 3 0>X> CO BO o C c cj (-) CO 4J Ed u a) O CO a CO G. c E X on 0 w 1u m — H i— 1 — s 4-1 •H T u 43 N O CO — / P-X2 X c W o H 4-4 o O H jJ 4J c CO E -4 >> 1 CO 4-) < CD o 0) 4-1 4J X CO CO >> )-i o CO CJ 4-1 CO u X r- co in CN m o> CN >* o — i Q o c c C z c o o o o c O C c o o o c a o Z u >, o H c x: to c = I i o CN o> co o CN St X m r— r— 1 m m CO CN C CN H o H r^ CN c c c !-4 c — •-> — 1 > I 1 i — •c 3 c H OJ CO Cv 0) X X) X •a X CO Cj E a to E JrH •H > >, rH rH rH i— 1 0) CJ 0) X! a> X! i-H rH DO •H CO H 00 H CO J rH 0 rH O — — i—i U Ih 1-4 cc X! c _~ c -C C _r a T3 •H ~ •H 0) 0) 3 (J 3 O u 3 u — ? 3 — T3 ~ o 0 0 O 3d •H r— 1 — >h 2£ SI ^c

PAGE 188

178 CO j-> •H 0) c 00 3 ^ C -H d ca)N xi c v_x U -H pL< to 4 — c u o cy •H CJ CJ w ft X C0<^> c^ c o e co •H CC X 1— 1 c ri •H G j-i •H o — > 3 cux CO CO '"N ~ EU c oH O 1cO w >^ 4J x w to -H T O (1 w ftX fa X o w ^ Cu u 4-> O O -H 4-1 4-> ft CO e u >> i (0 4-> < in CO CL1 CD •P 4-1 CU CO E E CO u CO CO H Cil u 0 H 0) u 0) a: C o i— i as O i 1Z CN cn CN] c m CN I i — i as t-H in cn rm i ri • I in cn cn i r i • i cu en 3 C o u o l*-4 CO ~ CO CN EL) CO TO XI to CD Ih c c co 1 — I CO CJ E c Cu E 5, O rH ft E W c c o i vC c c a r-as m cn cn o i cn i • i o i 00 o
PAGE 189

179 co u •H CD c DC C •H CO 00 Nl XI C w u H "C CJ CO 4-1 c o oU •H "i 4J T3 -H 0) rH 4-) -H •f CJ J~l N CU CO ft X o W w a. u H 4-) O O -H 4-1 4-1 Cu CO S u >> 1 CO 4-1 < ^ CO cu cu 4-1 4-1 01 CO e e cO in 4-1 CO CO o cc CU 4J CO u m en rc o o I O I • I m 5 o i 0> rcn x cn cn i C5> CO c I • • CU c c r-~ CO CN — X ft ai > (U .a H c oj C u c CO Cfl cn E a 01 CO ~ o (U l-l a i QJ i—i —4 3 H cn E cu CO : ~ UH >, co o CO CP* 4J i— i o G i — 1 —I u .c 4J cn co a 4-1 E en i — i X CU CO •H o O u cu CO 3 o fx* E H H OJ CO Q o cd -a o CO w c^

PAGE 190

180 X u •H J C 3 ^ c •H re 00 N — C u •H [in -a O w ^ — c # o c o •H G u ^ X O. w X oo <^ W c c i CO cox c 10 •H B 4J •H c 4J CO T3 G — CD G CO 4-1 X! U G o g DO E E ca 0 Gu U <•*-{ 4J xs •ri i— 1 4-1 H *r u .£3 Nj -. i to -j < H CO G ai 4-1 4-1 G CO am HIT QQ U 4-1 eo CO — G >> Ih O 30 ai 4-> CO CJ O a o cn cn o in co CN o ON CN CN o Q O c I I — CO r ] O cd 4J co c 3 II c u D (J ca • n CO 3 o o r~ i o co I - CO — o j_> CO E i — i CO G en o XJ , G CO w rH X Ih G CO o H CO CO CO O E G E 4-) t-i rH CO G re re i-l — 1 X G M. MH G 0 O 4-1 r i T3 CO U — -c -a -o CO C m 0 •a C c C X O X <4-4 T3 T3 CD O ca re d) C E (-i •H re ca CO a 4J G G CD G 4-> CO CO 3 U E E N co -a re G 3 3 cr CO c 0 •H c re E o O CO Ih G u co Ih 3 4-1 G _G 1 — i n JZ PL, E C E 4-> o — — O o ca G jB 3 O CD 3 — < re > G re

PAGE 191

181 o Q. (J o Cu o >1 U 00 o c CO -H 0) E u 3 CO CO W CO a O o i •H cn cn E on c •> o — u a c B 0 2 O CO 4-1 QJ E e -h 3 r-l co c C O O C u ^ IB c o H *J Cfl > 9) K O o3 co a) 0) cx 10 E E t-i 3 CO • CU CO CO E C 3 O iH CO U CO C C o o o CJ c ^ >, u o 00 a CO u -J B a CJ a a re a £ c o •H DO OJ OS r-~ r~co oo cn CN CN i — I o o o o 00 — I CN C"\ H S-l 3 cn c o o o QJ CO !5 ;s cn in C3> cn m in cn cn cn o o o oo cn o cn cn \o cn co co o o o CN CN vO cn O r^co *in co 4_> X -O o C -a 0) 3 U C/j vD CN cn O CN O cn cn cn cn cn o o o o o cn o < cn pm CN CN CN CN o o" o o O \0 O 00 cn cn m r- CN CN CN CN • • • • o o o o 00 CN CO CN cn cn o CN CN CN CN • • • • o o o o X o if, cfl c D W QJ If. 00 [3 •H U c C/l oo QJ 3 CO

PAGE 192

182 >1 a u DO o o c •H DO H 4-1 o E 4-1 3 o rH 3 o cj c CO 0 XI ss c / N CO c o a E H H 4-> es H CD co C > E 0 c c cu obs CO cu 3 cu ft M E E u 3 CO a ns CO o H m u ca e c — i 0 0 o u G 4-1 3 o CO OJ 4-1 a U r--OcNu~>c'00\£)Ooo fl N sT CN n 00 O i-H d d d 3 3 cu oo o o d d d in m ci m n i irnoo^ OOOri O •— lOCNOO OOOOOOOOOO £N O CO m fo i COOh d d d o -a m — 1 CO i— i — i : X X 3 CN 3 3 CU o 1 — C GJ CL) 3 r — v 1— ( 1 — 1 X GJ 3 3 •o X X 33 rH X ^> ~3 rH rH •H c rH •H H x: c a CD H x: X! u c cu CO U x: u (J cu 3 u — IH •a -a r-i o x: 3J t — i rH •H o j •H — 1 H H JS jj •H CJ •H x: JZ u CO x: a u T3 -3 ca CU o X! a 'GJ 0) GJ — i 3C rH o x: 4-> rH •H i — 1 •H X o o 4-> •H co CO >, 3 H 3 C 3 c kn -3 u J= CD •H 3 a QJ a 33 CO E co E rH — GJ GJ 01 a GJ CO U 4-1 I — 1 •H r-H H 3J X) T3 -a 3 V-i H 00 M CO 3 a GJ GJ GJ •H ID 1 — 1 c S-H c 3 CO oo CO DO CO E •H eg •H CO CO CO CO CD T3 CO E CO E >1 > 1 — 1 GJ GJ cu 1) rH r— O 00 00 CO CO rH — rH rH U 3 x; c c C C ~ "O — X) GJ CU GJ 3 3 —5 T3 "3 3J 3^ ca o O o O •H ;H —J <— 1 3 m 2u w UG 3 a -3 3 o ft CO Gj U <4H O Gj gj 3 OS OO I — i X 01 -J •H XI o CO CO X XI cu u x: ca 2 o a

PAGE 193

183 re C rj o H u CU c CO o z 0 CN CO 4J >h "H QJ S 3 rH • CO a o CO c 0 •H J — 1 a > u 1) CO XJ o c CO CJ p CO d d r- — co — i ro co d d rco cm r~cn r~ d d CN i— I CO • • o o m m vO CO -3m d d co a. E CO -c O o <4-l CO o > H cu u CU O 0) CN CO X CO CO CU o CU CO c is o o x: to •H 4-| CO 3 CO u 03 c c CO CN CJN O CN O CO CN d d t — I ON O oo t — ( d d CN CX) XJ — I >> IH CO CU O CO O E o H CJ 0 CU Ma z: Emp 2 Sex Jh

PAGE 194

184 jD DC ; c <£ •H CO H J-j CJJ (-1 H> C co co a u c o o u IH u Dm c w CO sc dj o E 4-1 H = •H 4-1 co Efl E co C •H > O rH u CJ c X: o XI z o on Cj 1 c o co 4-> c •H O E -H •H 4-> T— I CO c > o u C 0) w CO X o H o 00 03 4-> co OJ a -1 CO > c CC CJ B 3 u CO CD er iH a, CO E E U CO o K co in c CO CN t— i C c 4-1 i—i 0 o o c_; c 4-1 CD N •H E CO I I I I I I CN CO co m co • • CN O O O O CN CO 00 0") CO — i m 3D C cc m CN O CN • CN cn CN CN co x in o CN X o m H r i o X >X cu . (0 CO DO O n i i i i i i m x o m co m co ^ — o 4-1 J CZ B CO CO si to 0! 3 E O 0 0 x X u H H

PAGE 195

185 u CO o c CO -H 0 H u — / 01 re CO c xs z o CO +J U T CD £ DO B O •H E 'ri U 3 H O CO c > c o u O C CD o l-l a 00 a •i-j re O CO CD i— i re > C ca aj E 9 u CO a; er t— i a. CO E re B CO m 00 0 CO co CN 1 — 1 E B vO O i—i CO u CO CM vO c c r— 1 o o o CJ c 4-1 I I I I I I m cn cn o> in r-l X O ^5 CJ\ CN CC cn r-CN cn i — -i O m re: o >— 1 I— 1 c 1 cn qg c o CN 1 — rH O o O o • • >> 4-1 / — \ •H CN 4J CO re > J re nb GJ — s N -re •H re cn CO re CO X — CJ co GJ H IH e u e re u re re i — CD i-i cn •H 4-1 T3 cn r o -re o X CO re c C re 0 X o o re re iw a. H re CO X 4J 0J o aj >H CO 0) U E E cn CO --re 0 0 H c t-i u u Ih — 1 re r— i U re c CD O o Jji 4-1 M 1— 1 _re Q CJ c 4J CJ M C x CO •a i— I o .re 1) CO re 0 -C lt-1 c l-l CD .re — re CO — I— I c jC CD co re o £ C G re re CD re. J) ca -r a 4-) O u cx o ^1 GJ — — r_o 0) 7Z ™ — Q ~ p, H — — -i CO CN E x I — X QJ DO CO i GJ — I — •H Cj CO > c cC 4J ca JO c o o H o j_j i JJ DO • H CO 1 QJ o ca CJ •H M o — 0] Co 1 i GJ — r* tri CD i. 'J) H CO _c re 4-) 0 •-1 c u •H re o T3 re, C O T3 Ih H re. > O 00 c a. H C CJ ca 01 a re QJ a. x; re re

PAGE 196

186 0] 4-1 •H '-n 0) c • H M 3 N C v_/ to ootu •c c O -H -O 3 0) U) Oh 4-) C X U O W CD U otd W C o c o 00 X rH c CU CO *H C 4-J 4-) *H 4-1 o rH /""^ -•— i 4-J CD Cl. •H 01 CO X lJ "O 1) c w CD i-i CO N — • u> 4-1 pl l U CD U ^ X CD CO CD Oh X CO o PH O Mh o •4— Cj CO 4-J kj r; CD rH /, ~~ N U -H *r U^N rH CD CO ^ CD X o 0 W ^ ^H O u CO H •H 4-> O CO O -H >. 4-J 4-4 rH QCO CO E U c >> 1 co CO 4-J < 4-> H o u co H CD CD 4-> 4-J CD CO o CO -H Jh 4-J co CO CO u CD re (-i o oo CJ i-J CO c o 3 T— CN m co co cn o o o* d CN vD cn cn on m cn O O d d oo -J m — i rO o o o O IT* in cn i-h co m o cn cn cn r-o I x ri • • i m cn m co m co .m co 4-1 x XI c co X) Ch 3 x; 3 O CO 3 o CC; CO 01 00 O a m o —i o o o o o co >— t m i o o I o I o I v£> N O CN -? m cn cn o • • • i — i O CN I I I O on cn O m -i o d d i i O I X x oo u C 01 i 3, 3 oo oo cu cn co •—^ XI CO w x ~ — u 01 iH 4-1 i — f c CO -H

PAGE 197

187 B] 4-> •H > C r CO P N a CO 00 fan £ u •H TJ (0 cL CD 4-J C X U O O (J w D. w X 00 ID UJ on oox rH e co H c j 4J — x^ c H 4-J o a. CO X — a c o u CO 4J u cu o QJ oo D, G E X CO 0 w X! u u t_ >^ ~J nd •H o rH *j H u ,43 N 0) CO P. XI X 0 PJ u o H 4^ 0 O — — a CO s rH ; 1 to 4-1 < u 0 0) CB 4-J 4J Cj CO e E ra •H a >n 4-) cd co PL, a> So u o DO CJ 4-> CO cj m CN o co co 00 m oo CM CN 1 — 1 m co X c i—H o CN i — i m 1 — 1 O o o c O o o c o O o o o 3 3 o c I 1 — 1 0> oo 00 o CO CO i — i CN o — LO CN in CN 3 i — CN 3 in CO cso 3 m c J -4" CO CN r I CJO uO CN O 3 CO o 1 — 1 O i— i r-H i— 1 i — I 3 3 o o 3 o 3 3 ~ I o H u >, u 0) •H o x: CO to o o O r- \ '-v 3 O 3 O rH x tp x 0) w QJ rH -H oo c >-i c C HI *H o •H t-i CO m CO TP E TP rH rH 00 -H 00 -H £ x: e x: p u p o o o >H >-. X QJ — I 00 c C QJ •H S-i W T3 OO -H 00 -H c x: c x: P U P u o o CO CO co rH XI X w TP QJ QJ rH -H 00 Sh £ rH •H CO CO E >> >, QJ XI 0) X: rH rH TP H TP 3 >h Ih CU QJ TP TP i-H rH

PAGE 198

188 CO 4-1 tu c w 00 3 N C ^ to cofa x: c O -H a) to Dh 4-1 c x u o fa cu u ^ w 00 ft) c o X fa •H X CO 00 X c a •r4-1 O rH 4-1 3 CU fa to OO X "G a c fa a g to -— 4-1 x: fa u V U CD o cc c_ C e X o w X! u u — -J •a •H G — 1 3^ O X) NI a CO w ft Xi fa X O w u ft u H 4-1 O c H 4-i 4-1 ft ti E G >, 1 CO 4-4 < IB G O 4-1 4-1 a to E E -H CO •h a 4-1 CO to a, •H X CO > g o CO G 4-1 to CJ m CN o o c o o co co co O t-H O O r^m m m o o 5 I m O l On co O co CO CO m i •H X! CO H CO tu -C _j — G G 0 G c G >H z to 1) CG PS G 01 3 C s 0 G 0 CO CN o c 00 O I O I • I O CO CM CO m to E CJ GJ

PAGE 199

189 03 +J •H -~n C tH 00 3 N c CO 00 JS C o •H E '"N — 3 CO PL, o *-> C X •H u o w X a CD ft w X M on •H oo x c CO •H C u 4-J t-H,0 0 I— 1 ^-N 4-1 3 CD CL, CN (0 00 X •H in T3 0) C W X in — i CD >h co CD 4-1 x: w o u 11 U cu oo O, C E X CO W x: o Ih Ih c CO o ca u o -H -H cn r00 i 00 X X U J3 N m r-i St CN c 4-> 4-> / — \ ON 1 X m c cn CO cn • 1 E V — y cn cn CN -, 1 1 1 (0 4-> < ^ co CD CD 4-1 4-1 a. v y E OJ > OJ H c CD C Ih c re CO CO e 01 cn [fl CO o 0) Ih Ih rH r-H 3 3 •H CO E tu re re <4H] BJ o CO C CO — 4-) Ih XI i— i o E i — < r-H 4-1 x: 4-1 (fl re E CO rH tu .5 •H O C u CD S) 3 r— C X E-h 3 >~ G CO Q
PAGE 200

190 CO 4-) •H ^"n a C -H cc P N) c ra CO tu X! C o rl -3 0) CO S-, 4-J C X u O tj c U — CL, w X 00 (TO W d o E a •H 00 X! rH c CO H 4-1 ^J O — I 4-1 3 01 Ph CO oo x ~3 a c w 01 >-i co ^ *J x; w U a U CO ^ X 0 w sa. o •rl 4-1 O o •H 4-> 4-J (X co E M X 1 CO 4J < M CO CJ 0) 4-J j-i a CO E (0 Ih 4-1 M CO o X X o oo ai 4-1 CO a cn c o m o o cn oo I cn i o i o o o I CO -J in o i o o o o d i — i c o I CM o o c d i r-CN CO Q O o> CN — i X 01 X to X a co W H X u 1-4 to 01 •H co a a o E u E 4J i — i i — i en ra ra h i— i X cu u 4-1 a 0 0 4-J CN u -in -a -3 cn o T3 c c C X 4h X) -o QJ o CO CO CJ C c i-i •H JO ra CO ra 4-1 0) CJ 0) to 3 U E E N a 3 3 a" ra O o •rl E o o co U u X o -C JC a c 3 u rH 4-J 1-4 • — i c c H T3 o c '— ra ai to Ih 01 O o o m CN C 5 cn oo ra E iCJ to 01 4-1 E C 4-J ra CU 4-1 01 co -C 4 — 1 ra ca. ai X 4-1 c O 1-1 4-1 1—1 o CO 4-> ~ 01 u 0) 3 0) 4-1 rH — co CO 4-1 •H > 0) o c a to X! j3 to H ra ^4

PAGE 201

191 i PC X) CO 4-1 H 0) c CD C •H Co 00 Ni C ^ U •H fj_, E -a 3 o co ^ 4-> C o cr u •H cd X cw re X 00 W amon CO CO X c rH C o CO -H C u 4-> 4J -H O rH 4-1 3 CU-Q 4-1 CO 00^ •H 4-J CD t-i CO c 4-J XI PJ CO U OJ u 3 CD 00 cr O. C E X CO o -3 wj: ^ iH o J3 , 3 4-> o CD rH n 4-1 -H T >, O N rH CD CO w Cuxi U-i X O Oh o U CO •H •H 4-) O CO O -H >-> 4-1 4-1 rH Dh CO CO E V, c >, 1 CO CO 4-1 m co in o CO CO CO co m CM oo o o CO r~ CO o CN i— I I • • I O O X in 4-1 x o co rc C rH CO CO -Q u u 4-> 3 C Xi CD 3 O co c o cn ca a co CO CM CM -. E a. 3 co co u a c •H Ph S

PAGE 202

192 •H 0) C C -H CO CON XT C w a -h Uh -3 3 a) tn j-> c =; u o O" 0) u a. w x com fcJ c o •H X co MX H c eg •H C 4J U -H o rH 3 cuxi en oo^ -c cu c cr o v-t co > — *j U C) U to — C E CO o w x: u o mh •H 4-) T3 -H O O H 4-J 4-1 Oh CO e ^ ^ I 10 4-1 >-. en cu cu 4-1 4-1 CD CO E E CO -H U 4-J co cn Dh 0) >1 c DO a 4J CN C cn 1 — i cn — o O o O O — o c o o o o c c a o c vD vO CN c CN C CO h CI m cn CN in LO IT r i cn m cn o \£> cn H E -a rH 00 -H c x: a u o 3 CN CJ — I 00 c C 0) •H tn -o x -o x H CO B c CJ u •a 00 -H 00 -H c x: c x: 3 u 3 u o o cn CU >—i rH X 00v_^ c •H C cn cu u -a T3 CU rH 00 -H co x: o 0) rH O T3 — 2 X rH r—i cu x , >, rH rH U U d)
PAGE 203

193 Cfl •H 11 c DC ~ J •H -3 00 N J3 C w 0 •H Cu a a> CO ^ 4-1 c # 0 0u •H u U X ft CO X 00 cc w on CO X c Cfl H e 4-> 4J H O 1 — 1 4-i a CUX CO 00-— v a 0 c ocu CO — 4-1 x w u 0) U CD 0 DO ft c e X Cfl 0 w x Ih 0 14-1 4-1 T3 -H 0) i-H 4-1 -H O X N 0) CO a. x Pi, X 0 ft u H •P O O -H 4-1 4-1 ft CO e u >, 1 CO 4-1 < Vh CO CD CU 4-1 4-> CU CO E E •H CO -H ca U 4-1 CO CO (X CU >> Ih O CO CJ 4-1 Cfl U CO CM vO O -H I O O I O O I f-o o en co m o o o o CN E s — C Cfl 0 CU 4-1 'J — Cfl >H E c O 0 cu -0 H-t ft 00 0' CO 0 O cn 1 — I 1 — a c X x — s 0 X X X <4-l 0 O Cfl 1 1 a) 4-1 C ] cu Ih c 1 Cfl •H CM co c CM •/. CO X n m X CO X ca 14-1 CD S-i CJ — X X E X C 4-1 0 u > 4-1 >> •H X CO •H CO x: Cfl 0 [fl (1) X tH cu QJ 0 co CJ 0 r-l 0) 0 O O CQ u >" 3 ft >H CO cu Cfl e OS U UJ

PAGE 204

194 01 4-1 •H 0) C a 3 ^ C -H cd ooN JZ C ^ O -H Ph 13 3 QJ CO w c X u -c x u d) CJ ^ a oo C E X CO o W X S-. O <-4— 1 >1 M O DO CD 4J a U 33 m CN I o I • i o CO 00 I O I • I X — I c c 30 X O co cm Q O co rc o >, 4-1 CD i — 1 '"^ r\0 o o 4-1 -H fH — > in CN 1 CN O CN] m r1 Ln CO CN CO 1 CO • • X O W Ih o •H 4-1 O x rH o O -H oo i in o CO 4-) 4-> ON i 00 c i a> a. co rn • i E U CO co CN co >, 1 1 CO 4-1 < Ih CO a) cu 4-> 4-1 (30 CN — 1 CO QJ CO on o> ON CN e E -r vD o M co -h ca CN CM 1 — 1 c C k 4-1 n — • 1 CO CO O o c o a. vj0 c c CM rH CO X c •H X m ca H CO C 1 c CO E Ih Ih X E 11 0 -C co QJ T3 01 E M CM CM 01 rH JO 4 — i X 14H X — X [0 cu E 0 E a o CM K ca CO N X 3 QJ X rH O >> cu x-. •H c a C in c *t — 1* CO eg to E QJ CO CO o QJ Ih Ih rH rH 3 i — I H 7) E 0) CO H-l CO 0) >, CO o cd cr J — 1 Ih -2 iH c E rH rH 4-1 X 4J m co CD 4-1 E to — OJ JO •H O o u QJ to -i O X X X H >QJ X ca Q QJ T3 3 QJ X X X X X

PAGE 205

195 x u •H CD c 00 3 ^ C -H (0 CON -c d w O -H Pt, "O 3 CD CO 4J d >! U O O" CD O H d o O0 X —J c CO H c 4-1 4-1 •H c i — 1 4-1 s 0) X> tn CO '-n — a d cr o CO w 4-J XI w o 0) o oo c_ c E X c c W -3 u T3 -H 0) i-H ^-n •P -rt T U J3 N a,xi X o u •H 4-1 O O -H 4-1 4-> O. CO E >, I co 4-j < >H CO CD 4-1 CD CO E e CO -H U 4-1 co co CU CD O 00 a 4-4 CO u m on •/j CD X Ih o CD XI CD E o CJ c m c Q C C a o CN CN CO O o 5 in C CN C c Q C I o r i O i in o c a i vt in in O o O O o o c o o CN m rn CN c c o C I i — i X (X3 CJ CN o Q <* CJ ^> 1 Q 'J) c CJ 3 cr nj QJ L ,_4 [fl rn *H Q i-H — ~r .— cu 3 CO O 3 O J, r-* <— 'rj ,-— 'j fli CO R *— m JIT i~ u H CO -3 4J SQ X IH O O [fl 4-1 QJ U CD rH 14-| CO >4-( > CD QJ C) x: XI e— • H

PAGE 206

196 M DO 0 c U DC rH H O A-> r~ IU CO w M ,— — t j j rj o i — i r n UJ U a o c a — 1 \ .(_) — -. 3 rH 4J CO CO — u i V— r \ v-J r Sj o •>* — u to (3 u •H 0 QJ e c: E •H •J CN a H CO CQ (— i— o 11 1 r i 1 — < m Q X fl ) n, X £ = u re /-—•. CU X X C o x u a) e c u 0 o o C c H u o CO o *j CO CJ c CU u H CN CN CO i — ( d d d d co o in i — I ON H m < X X rH c s e c CO CO X in CO r-l U X CO 4-1 X XI CO -a c •H 4-> c 0 O X CU c •H C o co o •a 4-1 i—i CO )h CU CU CO co 4-> x -C cu X u r-l CJ 4-) 4-1 4-J H H P E u 3 cn c c c 0 o O cu re cu P 0 J-) Urb U C/3 00 oo co r--vo r-~ in o --r o
PAGE 207

197 >> o o c H 00 -H u oj E /-\ >h 4-1 3 H <4H U Oh O CO CO >H d u c E 4J •H H 4-1 CO co £ to a •H > c — U CJ d CO Q SB o CO 4-> U -H CU E E -H 3 rH CO C C O O C U ^ n c 0 to > Ih a -J3 o to CJ S— c7\r--ir)-covOinH OOOr-iO--iOcOOO oooooooooo incoiriHini/ivOHOMn OOO.— iO^-iOcnOO oooooooo'oo* CO ^-1 i— I X X X ^ w C7* X X w c c cu 0) u U T3 TO) r-H .H -H •h x: x: o u o o ^ X5 CU CU rH -H 00 IH C >H •H CO CO E 00 CO c c 3 3 o o >H >H X CN — i X d cu Ih — — i H Jd U JH CU •H 3 3 CD 00 c U o IS T3 CJ •H Sh Ih CO tu .U "3 r— I H sz CJ X! d a u •a I — I •H jd u tu 00 d H CO •a o DC CO OJ CU H •H CO CO E >> 1—1 I— I u u cu cu — -1—1 rH W til o co ~cr m o m -j un in o' o" o' uO CO CM MN O OOOh o" o' o" vO O CO CM CO CO CO O 1 — 1 o' o" O* CN O CO lO CO i — i CO O 1 — I o o o" d tu T3 d o 0) o. co a> to to 1 — — i CO CJ X jQ Ih V 14H CJ Ih a 4-1 O u •H c Cj si i— i U ^2 O -~: CO OS

PAGE 208

198 c u 00 n Q H DO H a i— re o u re re. — \J (—1 j 1 rj n i — i O m UJ M O o E 4J •H 3 •H 4-1 co E B re H > o H o o C co 0 .re a: o CO 4-1 Sh -H CD E ^ E -H CN 3 rH w CO C c o o c [0 re o H re > CD co .o 0 CO CD Ih rH CD ire. CO a S re CO o ns CO c — u CO a c -J c 0 0 U c >-. u o 00 ai CO CJ 4-1 c CD CJ Ih C re. re re CJ CN r1 i — i r~d o o d o o CO CN O ON o o" r^cn o cr> o" d m m co in d d r-co O o> cn ro d r-co cn d d co O cn r~ d d oo oo m oo d d o o 00 CN d d CT* i — i O d d 3 rH CO D, • • a. re (-1 E r-4 G re O CO re 4-1 a re tn o r E l o i a •a 14H re. ro o O CN CD c rH X 4-4 o 0 CO CD CN CD re CO CO •H CN [fl re CN re: (/, X CO <+-l re 01 ~< re £ CD a ^3 -re -a — i > 4H CO CD •H CO x; CO O o E rn CD GJ O 00 a O r— a O _re u >z re 2: a, >< b re e 0) CJ re. s.

PAGE 209

199 a o H u 3 co ca c a 4-) ox H 4J ro co E S — C H > O — u CU c CO 0 -C z 0 co CO 4-> C h'H O C O h O C CD O w to o c c CO a. E 3 E o co co 6 C 3 O rH co u co C 4-> o o C 4-1 >, u o 00 a 4J CO CJ to QJ 3 i — CD > C CO CJ E a 3 < OJ N •H 10 E CO I I I I I I IT) r-l co co ro • • CN O rH o oo H B o OJ 01 E xi in o • • E E o r-l CO 3 3 X rH O 3 C U 1 1 u — 1 rH H CO E o CO CO 4-1 Ih J 4-1 ro CO 4-1 s o o u CU CO 3 H hi Edu >H Gue CJ E O ja E 0 sco CO CO o CO CN X CO CO o Q CO CD X CO 'OJ o <4H CU XI CU E o u c I I I I I I co r~ CO —t • • co en i—i o-> CN 00 ro co oo ON ON tf co rn in CO O o O rH in • • .-I CN co O X •3 o 0> 3 CN 3 X eg co CO M >H CO CO — 1 — 1 rH O O -3 "3 T3 -o 3 3 CO 3 3) CO 3 3 O O -3 -3 Eh

PAGE 210

200 >1 c 1h CO o o C •H cc H 4-1 CD u 4J <* o co m u 3 c 0 u — — 1 u pL, o s — s CO co S-H c o o § 4-1 H 3 •H 4-> cn CO s eg c H > o H S-l u >H -H CD E E -H c o 4-> 3H B co c > c o l-i odcu u ^ CO O o CO GJ 4-1 CO u in o i — CO > c CO o E TO 3 u to CD er i— 1 D. CO e E u CO in 00 CD CO co i i—l o o O u e 4-> o-> oo CO 00 CN U~> co cn I I I I I I m vo CO -H o o I I I I I I c 3 0> Q o St — i o O o c CN CO CN Cxi 1 — 1 o c CN C-J — i O o c o m L-. in vo co fh • • CN — l • • >! *J H CN HJ CO X x^ ;enb a N r— | •H C co to CO X >^ to rH CD to CJ H M E u E a CO J CD C4H co •H 4H •a CO CO T3 -a -3 o CO C c c C o X GJ o CO CO 4-J ^-^ a, •H CO CO X 4-> GJ GJ GJ rH CO CD u E E to CO 13 CO O o H 3 u u u U 3 i— 1 CD c c o o O i4 H M x: p CD c 4-1 CD H o 3 W CDco — O CJ CO a o J3 CD -3 E 3 co O 00 X: CO 3 3 m r— i 00 3 CO T~H H 00 r( CD J3 3* 3' 1— 1 3 3 r— I 4-1 M -3 rH o .c CD CO 3 O £ 14-1 O 3 CD CJ U CD

PAGE 211

201 ca 4-1 •H ^ a c -H to BO 3 N B v — •o a OOLk o JC C o o •H m E ^ CO 3 :|< 4-> 4-> O rH •H 4-> 0) Cu "O CO MX c T3 0) c w CD 01 Ih CO w c_ 4-1 x: w X o Cj cu cu cx c~ C E •O X co C H W si sO u MH x: cu CO 3 >1 o 4-1 r; XJ •H CU rH '-N >-. 4J •H -H rH u X) N JaS cu CO -— CU a. xi [in CU X O Cn u 14H o CO U •H •H co 4-1 O >-> C rl rH 4-1 4-1 CO (X a C E M CO >, 1 CO 4-1 4-1 < •H XI o H Ih M a a u 4J -> o 0 CO <4H E E "H CO H 3 CO 4-1 u CO CO H &< CJ 4-1 CO •H 4-> CO 4-1 CO Ih CO X o 00 0) 4-1 ca CN c o •H cu OS CN co o i m CN rc d i 00 On CO cn m on in co cn CO CN CN o d o on 3 r~ rH CO CO m r-rH O d d o m i — i \jO CO rH -J CN • • o o O CN O 00 CN ON co m o >m co 4H x XI u ^ c o rH CO Ih CO -O 4-1 Ih Ih CU 4-1 3 E C X) C CU 3 O UW2 C 0 ca a cu CO co cn r~-tf m co HH(<1 o o O I o o o I oo m cn m O
PAGE 212

202 co 1 1 •H > — s o -1 i— t r n "1 rO —J LXJ rrl XJ (j E 3 ft o CO (X, C X u o u O d. K oo > w c c i H oox i-i c CO H c 4J o rH / V in 3 QJ Cn m oo x -3 (U c w QJ H CO w *J X W U i N QJ CO ^ O.XI tn X o w H D, (J -H 4J 0 O >H 4J a. co £ 1 09 4J *S 01 cjj 0 4-J 4-J a CO 5 E r CO iH 3 H 4-> CO w C•rt X >> H O DC 0) 4J CO u C LO m X st X 1 — 0> X OO st St o> IT i— i r-> CN o st — 1 g> rO i — i i-H — i O o c C c o c o o c o o C O i — i i-H CO — 1 rco o i—H r— 1 CN 0> CO st X o O st CN On CM m CN co St CN c — H CJ O o c O o c o o O CN o> co — 1 C o CO CN C o m i — i X O st st Q X m CN 00 o o> X <* o c st CN CN — 1 CO st X CO O st st cocoLO st o> o CO i— i co co st O 3 CN i-H I I 0> o o a> co OJ LO CN co in 00 o a> CO CN v£> CO CN Oo m CN CO in 00 CN LO o CN st c c; st o o i-H 1 1 O o o o O I I I I I I I I I ^ — \ 0) r~ N in O n • • CO "3 st rH "3 — l r3 a QJ — i Cj i — 1 0) X CJ X XI rH -3 4-1 rH H rH •H X o o 4J •H oo-— >h OO M >> o •H 1 — I CN 3 Ih 3 c H 3 03 o CO' — 1 1 — 1 i — i •H 3 CO 3 •H QJ CO 0) CJ a X XI X X -3 X to QJ S CJ LO M E H 1— 1 •H CJ CJ aj d) QJ U (h "3 -3 00 Ih V4 H •H tH H or! T5 •o "3 XI H "3 rH 3 H •H CO c c CO 3 u 3 O — 4 0) — 1) •H O H •H cO rH c a o 3 0J Ih o CO •H 00 •H oox; oo x: m E orH co Ih •H t-l CO u CO x; CO CO U CO o CO -3 E T3 CO —J E a u U >> i — 1 l — 1 rH H —4 CJ CJ a .3 CJ O CO H OO rH CO H CO H rH 0 i-H o H 1 H^ rH H H X c .c c J3 c -3 c -3 "3 X) •H -3 •H CJ CJ CJ 3 U 3 O U 3 u •3 ~ > "3 XI ^> ~3 -3 W C o o O ; H ;H rH H 3 O >-< fSh ^w rr

PAGE 213

203 w c DO CZ 00 Uh jC c •H E ^ -o a # co D> 1 cn 4-> < !-i co CD QJ 4-1 4-1 QJ CO E E H CO -H a t-i 4-1 CO CO CU QJ (c 00 u CO u o on cn m roo I • • I in rH I I ON CN I • I NO c m o o o vO oo cm m rH O I I s CN o CN St oo O O l c — c o CO qj Ih Mh o o u (B K oo on co rH i — I CO X X XI QJ U J* 4-1 QJ O •H XI CO X 4-1 rH 20CQ CO E O O >4H w > H QJ u QJ O QJ CM CO X CO w X CO QJ O a c > o u 0 14-1 -C co H H-l X CO 3 CO o CN 0) CO CO QJ O u a c M tfl OJ B a, E w CN QJ CN CO X CO w X co QJ O

PAGE 214

204 m 4-1 •H ^ CD C tI M 3 N C w x: c a 3 CD CO CU 4-1 C X O O w D, fc) X CO CD H C o E ca x oox rH c •H C 4J 4-1 -HXl O I — 1 4-> 3 CD (X, Cfi OCX •rl CO •a cd c w X m CN CJ i-t co s — CD — s jj x: w 3 o O U m CD oo Oh c e X CO o W x: u U MH >, c 00 a CO CJ oo i o I c o o 0> . 4-) T3 -H 0) rH ^ CN X o w u (X u •H 4-1 O o m c cn m in O -H 1 o 30 •— i m cn 4-> 4-1 ON 1 cn , 1 1 i W 4-1 (-< CO 0 0) 4-1 4-1 CN i£J c in c < — i 0) CO vD m X m o cn E E tI *-H 1 14-1 X i 1 X vi o E o E a o CN X <4-( s — CO CO N 3 x CD X rH 0 >, QJ X •H C o C Ih a CC co M E CJ to tn CO o o Ih rH 3 i — i 3 •H cn E , CO O CO O" 4-1 X rH O E — { 1 — 1 J XI 4-> V] C3 ra 4J E n rH CJ CO •H o o U CJ CO 3 0 X [x s H H 3 o CO Q CJ "3 3 o w CJ

PAGE 215

205 00 3 N C cd oofc J3 C O -H "O 3 $ 0) co cu PCX O O w ex w w c o •H X CO oo x — c jj •H c 4_J 4-1 •H.O Q 1 — 1 4J 3 CU Oh m OOX "3 o C W Co u co — 4J JZ uz U CJ CU 00 fi. C E x CO o w m u 4 — 1 4J -c •H CD tH '-n 4-1 •H 1U N CU CO w Q.X) tt, X o W u c u H 4-1 0 O •H JJ 4 — 1 O, CO E ( i >> 1 CO — < M CO CU CJ 4-> 4J CJ co E B • CO w (X o •H X >. M O 00 CU 4J CO CJ m co cu X CO CD M O 4-4 a jO cu E C u c o o o CN CN r 1 lo o O o CN r-( o o o o I C^ CN X en co CO CO O H H CO i — I i — IX o o x) -a O T3 C C CO CO c/i co 3 3 O O .c x: 0! u a 3 cr s. CO E CU c o H 4-> u CO Ih 0) r-H o o 00 o CN Q O co 00 CN o c x 0) -H U E CO CO U 4-1 --s CN •o -a co C C X CO CO — m cu cu B B N O O -H (J (J CO c c ON o> in c CN CN m o X CI co C 0 l_ cn 01 B] l-H o CO co 01 u CO o m m i CN ON CO r> p i i 4_> c CO 4-1 LO c o u o a o x: of o •H CJ u 4-1 co c 0 o cu SI c H o o — 1 — .A CU CN 4-J c 0 jj u u U CO c cu o CN IT) CU — > — > 1 c _c 1 w s i— H Ih CO CJ 4-1 3" r \D -3 0) c/> l-l 'H CD 3 — CO *H o i— 1 H — 1 CCJ c co C— c c 0 H 4J c u OJ CO E Ih o 4J c 4J •H 4J CU Cfl 4-1 QQ X 4-1 O <4-l 0 IJ cu u p cu 14-1 CO 4-1 > o QJ CJ sz H

PAGE 216

206 > Q — 4_j ] 1 CUX *H CO CO^ j_j 11 C CD— < CU — 1 CO — 4—1 X! W u o o fD cr cu GO c E X3 CO 0 1 -G Ih Q CJ <4-t CD CO >> o w 4J — •H cu r-H '"n >~ •H 'H i r— ( o X tsi o CO ^ w CuX b_, fl] X o Ih ex n CO u •H •H to 4-> o >^ O •H rH 4-> 4J CO Ch CO c E CO >^ 1 CO 4-1 4-> < •H XI O H Ih co o Cu >-t 4-1 4-> o 01 CO n-i E E -r ca •H ca CO Ih 4J u CO CO •H Oh CJ co en >. Ih o 00 a 4-> CO o lO C o •H on QJ ft! h H CO t-^ in ON CO I— I 4-1 -j 4-1 M L, 3 CO O o O CJ 25 Ul ^2 -tf —I CO on on , LO 4J X C H w o u ri C o 4-1 rH CO Ih CO CO x 4-1 N u u u H 4-1 P E c C C X C o cc cu a c CO X O cn z; ca Ih o E3 c/; cr. C H Ih Dl. C/n x r/j CO X • oo x Ih CJ 3 CO CO p_

PAGE 217

207 m jj H CJ C cr •H CON! £ 3 U H E TJ 3 01 en j-> a 5u o u a 00 CO w on 00 X 1 — i c CO c hj J-) •H o 1— 1 JJJ 3 CU.O m oo^ C3 UJ C CDcu CO ^ u cu (j CD O DO cC E y. CO 0 nC ch fr 4-1 -O -H -H •rl O .Q N , 1 CO 4-> < ~ co QJ 0) 4-1 4-1 0) CO E E •H CO -H Sh 4J CO CO 00) u cu 4-4 eg U CO o o r- co o CN o rH cn CN m CN i — i CN vO g> OJ i — i o~ crr~— cn o co r cn -Jc — i vO IT! 1 — 1 co co — i c O CN i-H c o O O C c o c c I I r- ro St c o C C CM 43 CN co r- r*c CN cn r- -cr 00 cn oo c •H X rH •H u O ^ — s JJ r — s •H cr v y u CO u >. — v IS O -H i— 1 2: CN 3 3 3 u 3 "3 u — i — 1 — i C c* 3 H CJ CU CU u -a X! X) X CO o E CJ CO l-l E rH -H CJ CJ Cj o 0 i-l -a oo — i •ri i — 1 •H ~ ~ T5 •o — rH 3 V-i •H CO c Im C CO c Ih 3 o rH Cj a H cu -H H cfl rH 3 cu QJ c CJ CU cn •H CO •H oo j: 00 rj X, E •H cd i-i •H lH CO u CC J3 UJ -3 a u CO U -a ca C3 E TJ CO -o E a u CJ >1 >. — i rH r-l >— rH a CJ a J3 cu — — 1 H 0 cc H GO •H CO •H CO •H o rH O rH 4-) r^ i — 1 Ih t-> c c J3 c c J3 T3 -3 •iH "3 •H CJ CU o 3 u 3 U 3 u 3 u -3) C3 -a T3 — o o o o Mi •H r^ rH 3 ro 2 w rO rr

PAGE 218

208 00 4-i •H cu c DO •H CC 00 N — c w -H O i— 1 Q r i 4-1 3 CDX a. cn co oo^-n / — s vO 1 c a) co C X — X o W (-. a. u •H 4-1 O O -H 4-> 4-1 Cl, CO E U >, 1 CO 4J < U CO 0) ID 4-1 4-1 CD CO E E •H CO -H U 4-1 co co P-, CD >> so M CJ 4-1 C3 u cn CO o O H CN H O CT> 0> s* sj m CN ON i on in i • • i vO CO i i CM O st O vO O CO On o I CN CN I O I o cu u > J >i H X CO •H 00 X CO o CO CJ X 4-1 rH 01 cu o CO 0) 0 X CD O >H z a >-z. co CO E U W

PAGE 219

209 00 3 ^ C -H S) DON x c U -H fa •a § 0) CO 4-1 C O O O" CD O ^ pa oo — 1 r-l o o 4-1 -H -H O CN X i cc N St 00 00 CN i CN a) co n-^ h ex o •H 4-) O co o O -H oo 1 CO 4-> 4-1 m i X i — 1 in Cu CO co • I E U CN CN CN >> 1 1 CO 4-> < u co 0) CD 4-> 4J r- I s — i d) CO o CN X a> EE -rl r — s • i CO CO o o o O Pm H CN CN 0 X rH XI 4H o 4H X H CO CD E O £ — C CN to m CC CO N 3 X 3 CD X rH o >-, cj X •H G CD c u c CO CO to E 01 CO 10 CO o 0 1h M rH rH 3 1 — 1 3 •H CO E CJ CO CO i*-i CC a) >> CO O co cr 4H u X H 0 E i — t i — I *J 43 4-1 w CO CO -in E co rH CJ eg H O o u CD CO 3 rH o X fa H H CJ CO o CD T) X fa UJ. £

PAGE 220

210 en 4J •H U 0) U co CD oo a, C E X CO o w -C Sh O 4-1 u H 4-1 O O -H 4-1 4-1 a co E !-. >1 I CO 4-> < U CO 0) QJ 4-1 4-> QJ CO E E CO ^ 4-1 CO CO CX, QJ h ca u o 00 aj 4U ca U X3 CN Q O o o o 4-> -a -h o 4-1 *H tH 00 O £1 N CN QJ CO w CL.-Q [i, X o W U & r-v£> cn 0 00 CN o o Q o o I 00 m o> CN CN O o I in cn CN CN O o cn in Q O in o cc CN in m cn in oo cn o CN i — i X cn X co >, QJ CO rH X '~ co <3J •H CO a CO O E U E 4-1 i — i rH cn u co ca rH rH X CD u 4-1 Cj O O 4-1 CN T3 -a TJ cn O O T3 C c C O m -a CU O CO CO m CO c c S~ •H CO CO ca n 4-) cu CD o co w CO 3 U E E N 00 a> 3 3 cr CO O o •H E O o cn tH o u '7) u O -C sz cu c c CO u H H 4-1 h1 1— \ X! In In Ot cn cn X CO -a c 3 O Ol, 00 m I o 00 CN CN I C ca jj co c c CJ o ca CN -rj Cj IT) i) CO o || M o (3 ca l— fl\ \J n ^H 1 r> M 1 rC2 o o 1 — 1 L_y 1 1 ca o -i1 irj ~ 1 1 r* Cj CU rr 0 — CO CO i•H ca 3 CO cr w •rl X u 0 rH rH X) CO c R) 14-1 O C 0 CO •H c 4J CO (J cu ca E u o H3 a 4-1 E C (h *J •H CU 4H a d 4J CO ca QJ X IH c o •H 4-1 i— 1 O CO TJ CO u CU 3 cu 4-1 rH IrJ ca ca 4-1 H > CU u o qj QJ CO J3 ca H Eh ca

PAGE 221

211 u oo o c OO -H p in -Jo w G. U O Oh O 4-> 3 CO CO c O cn o CO CO a in — o co o o o o o o o o X > CN X c in O O O • • • o o o H o ss c Ih CD Xi w 3 O C/3 X 0) f — • CO P X Xi c H ^ o CJ H c 0 •P i— 1 CO Ih ca CO XI •w N S~ !h CJ •H 4-> D E C C X) S CO (U 3 o XI CJ co Ih 3 \C ^ CO N o o o o o o o o Ml in I— 1 0) o E H •H CO CN CN vO o co m "H J— J -Hh ^7 H~ I J (a B CO 1 CN CN CO rH co co co CN CN CN a H > o 1 — o O O o o o o O o O u 4 CD to c CO — 4—* I— i M 0 CO CO •H Ih J_ •H o c QJ E H ca co o 0) o o — I rH • c c Ih o O o o o o o d o o o c 0) CJ cn Xi o / N CO CD c<3 Ih rH a Oh CO E e u =3 co CO i-H CN o> m m o 00 CN a) o CO co
PAGE 222

212 4J ^ i-< Ih c O o c ai CJ sqo CO cu u H cj (S, CO E E Ih CO CJ su CO o rH CO o CO c d AH o 0 o a c 4-1 U o CO ai 4J TO U 4-1 C o u Ih a c CG O u >> u CD <4H H O X r^-Ocouo Or— i o i— i O CN rH O oooooooooo mr^cNOoocNCNrsiooo OOO^hOi-hOcOOO oooooooooo lOCOulHiOiOO'-iO^iO in in co in ts ilo^ooc^ O O O r-n O >— iOcmOO oooooooooo •H H H X •H £ cu •H X XI u .£ w X U u -a -a 03 u X! OJ OJ CJ CJ rH X o £1 4J rH H p — 1 •H o H H CO Ih 00 1^ H c S-i c Ih > H •H a m CJ T3 CO E CO E r— 1 H CJ CJ OJ O CO Ih 1 — 1 H rH i — 1 -a Tj -o a c rCO 1h CO Ih CJ OJ CJ QJ H 03 c Ih c Ih CO CO CO CO CO E •H CO H ro crj 03 CO 03 co E CO B >> OJ OJ OJ O rH — I CO CO CO 00 .-H rH rH rH Ih Ih c C c c -a -a — "CJ CJ CJ 3 3 3 T3 T3 -a o O O O H •H l — i r—f >H w hJ r- o m CN N CO O rH o' o' o' CO CO — 1 CO a X X Ih l-H a u 0 HI CD CJ •H X a CJ X! -J —i CJ 3 o 03

PAGE 223

213 c M w Q r* •H DO •H *J CD Ui i— 1 c nj CO O c o (J 1. ~^ U c ca c o c 4-J H •H *J en to E M c •H > 0 rH u G c w o Z o B ca 4-> c H c a) B H H *J CM 3 rH TO c > c o o c a CJ obs in 01 cr rH ft CO E e 3 o CO 1 — 1 c c iH u CO c c *J o 0 0 u c o CO (Li 4-> ca u 4-1 c CD U (h O C a o 2 o in o o o o o O CN 0> o o ON rH - r-^ O O d d o o cn r• • o o •H On 00 — I cn r- • • o o - o d d I* QJ c CD c CO CO • • 3 1— 1 ca o. • • ft c u E rH QJ CO o cfl C *J CD c ca or c J a Td m 4-1 tx CO ^"N o o CN QJ o X! / — N f— 1 X CO o ca i — i o 4J CN a co N— CO CN ca H CN ca C CM ca a 01 CO 4-1 X co QJ X (0 E CD w o XI .a E rH > 4J >i ^— /
PAGE 224

214 u o • CX, o o U U-l Oh O U 00 o c 4-1 3 to CO |H C CD o i 4-1 •H •H 4-1 co CO E CO c •H > o H Ih u C O S-i O C CD C_> w CO o Ih o CO CD 4J CO CO CD 3 H TO > 3 co o E s) 3 4-1 U < I I I I I I CJ\ i— I CO CO CC O m O r CO H CO CN ro CO CT> o CO CD er i— 1 O. 03 E Sh 3 CO I s -. CN — i ID CO CO in CO 3 ON <* I s — i 0 1— 1 co u CO CM i — i — i i-H C c 4-> i — i i— i o o O CJ c 4-> 00 m CD N E C3 CM X T3 m CN • • o X Ih X! CU (1) c CO T3 c CD CO O Ih i — 1 ua cx c cr 1 — 1 •H co CO cu Ih Sh E CO o ja U-J E E O 3 3 C c CJ 1 1 H rH •H M H CO cu E CO CD 3 CJ CN CO CD E H-l O Ih CD XI E 3 S5 CD E O JC E O >> CO CO CO CD s I I I I I I \0 r-rO CO CO o -tf CM i-H O co o> i — I 00 o o LO CM CO o CO X o Ih 0> co CN 3 X — CO CO CD CO co yIh in C0 CO CO 4J H rH 00 H 1— 1 CM CD O o X Ih -o ~c o IH -o 13 CO CD 3 3 Ih -O CO CO' a CO CO i— I
PAGE 225

215 C CO o Q rI— •H Gil H 4-J QJ 4-1 -C 0 — Q u 3 C Li — i o CO e CJ o 4J H •H 4J CO o: E M d •H > o i — 1 u QJ a CD 0 XI o m o M H 01 p. CO E E u 3 CO cu ns CO c co u c c 4-1 o o o U cS JJ u c DO a 4-1 a U I I I I I I CO CD H cfl > c ra QJ — I sill") oo - U in in CO c o CN in 1 — 1 in CN i — i CO O O CO cu rsH sj CN XI go CO 00 o r1 — o m X' o i — 00 m i — i >c — 1 i — i o o d o 1 — 1 >, 4-1 r-% •H 4-1 00 C X CO qu 0 N — — i •H c co CO ro X >, CO o co QJ •H E CJ E 3 l-i CG CO 4-1 01 Sh 4-1 H 4-J -a co CO 13 XI T3 o X or C c C C o X CU OJ QJ CO QJ U E E 1/1 ra -a CO 0 o i— i c >-, Ih u u u rH 3 rH UJ c c QJ o o J* 4-1 M r— X! a Oh O In Ot We cn rH o xi CJ CO 3 o u QJ X) E 3 CO 13 rH O x: o CO 3 C X c QJ O t-i a Oh

PAGE 226

216 4J CO o c DO o Q cd CO li* '— ,c C • ca 0 •H 0) § co "O n 1 UJ Mh 4— C PS co u O W •H CJ U w U-l ft w — 1 X 00 ro ,_j W c CD O i e* l E CO CO d Q •H co OCX cu i — i C rr\ •H c |,> 4-> •HX j i O — i •H [ i D Q) 0-, m 00 X C *o o c w CD 0J SCO w cx X w f 1 (1) u ^ CD (D CO p, C E X od c X S-l Q U el (D co S3 o 4J r; -o •H iH >- 4-J •H -H r—j u X N ^ o CO ^ C-Xl (x, CD X O W u ft t_3 1 Q CO u •H H CO o o •H | 4-1 *J ft CO E In co >i 1 co 4-> 4J < •H £i O H M en QJ c J-i t-l 4-1 O CJ CO MH £ CO CO s4-1 u ro w rx, o •H X •H x U o 00 ro U in ro CN ro o >— I O CN O O d d i 3 c o co O • CN O 1 CN ^11 m d -> m ro CJ 4-1 X in c -H w o ro o O X X ^-v XI C o co i— H CO M X CO CO ^j CO )h N U U CJ c QJ QJ •H 4-1 D E c •H E rH 4-1 r: C XI c o E rH C ro CD D o m ft 3 CO -H c_) co i< CO CO Ih 0 3 cc

PAGE 227

217 CO 4-) H ^ CU c ^ cc 3 N c v — • co 00 b-, X c u •H E ^ CJ CO PL, C X u O W o O w ft H X co <*> w a o E a cox rH C CO H 4-> 4J o • — 1 — u 3 CO 00 X "C Gj c w GJ Ih co ^ 4-> (J o cu cc ex c E X co c w jC Vr o 4-1 X 4-1 X) -H 0) rH 4-1 -H •H U X N CU CO PL, XI &H X o o •H 4-> O O -H 4-1 4-1 Cu CO E u >, I CO 4J h CO CU CD 4) 4-> CU CO E E CO -H 5l 4-> CO CO (X CU So I* o cc G 4-1 CO CJ I/O co u u o Gj CO 3 O ac m o co CN o CN co CN cc o CN uo CJN CN OO i — i o o CN o oj C CN CN • — o o c O c C c C O O o O c O o o o r- o CN o CN in CO o DO — i CO CO — — I in — i O c CO o CM r— < LO en r — i c^ o o i — i LO co CO CO m in c CO CN CN m Cjo m en c cn m r o O o o o o • o c m o rH i — i CO 1— 1 CO o CN o> CN co CO CO c I vO O I— 1 CO 1 1 lO 1 — 1 CO o -. •H X rH •H X 1 — c 4-1 H CC u cn o •H i — 1 CN c M c c a TJ > CO 1 — 1 1— 1 — i H a n 3 •ro CC CJ CU CU X TO X X -a X co 0! E CU co l-i E ki -H H CJ CJ cu Cj U V, ~ T3 CC u r-H H rH H TO TT ~ T3 -r rH T3 rH c u CC c Ih c oo C Ih c cu rH cu H a; rH O H •H cc c CU iCJ c CJ rH o oo •H CO •H cox: 00 X CO E •H M cfl Vh H CC u CC JC CO -C CO u co O CO -a B TO w -a E T3 u o >i >> rH CC rH rH rH 01 cu CJ X QJ X rH rH 00 •H •rH 00 •H CC — 1 i — i o rH o rH HI rH 4-> M Ih c C rC c x c x; T3 •a TC •H T3 •H CU QJ u 3 u 3 u 3 cr T3 T3 T3 O 0 O 0 H H rH r-H >>H > a: w

PAGE 228

218 co w H '~2_i (DC"" oo a si c w co oo tu x: c -a 3 # 0) CO 0-, PCX u o w 0) O w w oo ^ c o E en a. w x CO MX T— 1 a cd •H 4— 1 J o rH ^ — s J— • 3 CD cx, m oo x "3 o c tu CJ r4 CO ^ 4-> x: pj u CU o 00 /— c E CO c x: M u 4-1 !*, 4H a •H o i-l ^ 4-1 H 'H U XI SI a CO ^ a.x> fao W a u •H o o •H 4J 4-> P. E U >, 1 CO 4-1 < CO CU c 4-1 4-1 CU CO E •h a 4-> ro co 1U H X So DO 0) 4-1 ro o CN ro on m o vO On I O I • I o m rH ro o On r-l O m — j CN o ON ON I oo i o rH c c o I g> o c c 1 -tf rH ro in ON ON o o vO I CN I • I o I 3D O o o I • • u CD c CD c Cfl ro 3 rH CO a< a. 3 4J E i — i 3 ro O ro CJ 4-1 CD a 00 u E c o o ^ — N CU TD 4H a, cc ON co O 0 n rH i — i c o r— > X) S / — s o X X 4H o CD co rH CD (J CM CD H — / CN in •H CN 10 c CN Cfl CJ CO X ro 4-1 X (0 1) X ro *4H CJ X w XI E X! O 4-1 o u > 4-> >i > y 01 H XI ro H CO XI CO O CO 4-1 i — i CD 0) O GO CD O rH CD O u o CO u 25 3 D) >z ro CU ro E Pi PS CJ CJ

PAGE 229

219 x 4-1 •rf ^ Q) C •' 00 3 N C ^ co ootu x c U -H E oX) 3 01 CO CU 4-1 C X U O W o, w w c o rH c a •H C 4-> 4-1 -H X O iH x~s 4J 3 , u o CJ *J a u o c c d i 00 CN o I LO o c o r~rO o I CO CN CO c 4-1 13 -H CO r1 co. co 0) CO ^ O O C 1 o < • X o w u u •H 4-1 O CN m co O -H m i i — i 4-1 4-1 .-< i cc o C a. co co • i E o CN r i m So 1 1 1 C/l 4-1 < 1h CO 0) 01 4-1 4-1 co rn iO o a) co CTv CN o CN EE -H O 1 COm co -h a CN 1 LO O t-i 4-1 • 1 co co O .—4 O O 1 1 co in o o co c co 5 CN DO 00 0> ro CO LO o c c o CO o c o cn c> o CO CO CO • • U 0) c a r CO OJ • • rH X o i~ E Cj -, E CJ O cc CO o X) cu E '-. CN CN 0) r— X <4-( X) 14-J X I— 1 X 03 Cu E c E "d 0 CN X IH CO CO N X! 3 CO X H C >. 01 X •H C CO C H c CO a CC E CD CO CO o CO Ih 2 u CU i— 1 i— 1 •rl EQ E CJ CO CO <^ CO CO c CO cr U X — I C E H < — 1 4-1 Xi 4-1 03 CO CO 4-1 E 03 H o CO H o O u o co rH o X Eh CJ CO a CJ T3 3 CO c/3 w CJ

PAGE 230

220 4-J o c H DC 3 N c — y co h c u •H u co o—1 u o w o ft w X 00 -, 1 ca < co c CD — CJ CO E E -H co •h a 4-> CO M Co •H X x re >, (h O DC CJ 4-1 CO CJ ID ro O CN O c o 1 — I rH O o Fieri c c co CN 00 in c r j I Q O o I o CN X Cfl CJ co X l-l (h CO CO cc CO o E 1 — 1 i — i CO rH rH X CJ cu o O 4J t-l TJ o — c 14H a o Cj d c u H JD co CO CO 4J 03 w 3 u 0) 3 3 ~ CO E O o CO l-l C J3 Cj u E— E— CO o o o o C o o I CO CO on O o i X CD -H U E CO CO U UH CN •a "o m c c x co co ^ QJ CD CD N c c ~ o o MH CO o M M O m cn o o CN o o o o CN CI ro CO X 10 0 o I — 1 o O — I I CN co i — i I c CO 4-J co c o U o a 0 3 o •H 4-J u 3 4-i CO c o u Cj -3 J — C •H 0 • o •a lO CJ O -j • 3 3 0 1! u u C3 • 3 CJ CD 0> 1 — I uj rH > 1 re J3 o 4-1 1 CO E rH h CC CJ 3 4-1 — CU ~ o CO rl •H CO 3 X ~ — CO •H X u c rH — H -3 rc C CO UH o 3 O CO •-I c 4-1 ca CJ CJ n E Ih • CJ CO 11 4-1 E -C C u 4-J •H CJ 4-> 4-> CJ CO -C H 4-1 CQ ca Cj X 4H 3 o •H 4H H o ca 4-> T3 a u Cj 3 Cj 4-> rH I4H CC CO 14H H > a u o CJ OJ Cfl -3 -3 ca H ca

PAGE 231

221 CD C 00 3 z-n C -H co ootxi X! C ^ U -H Cl, -o 3 CD CO 4-> C u o cr CD U ^ w 00 (X> c o cx H co •H 00 X o — i G o co H C 4_i 4^ H >, o p-H 4_> 3 CDXS •H Cfl oo--^ 4_> "3 CD c coCD M co s — CO 4J x; w 3 (J QJ O CT> C7* CD C£ CX c E CO 0 — | Ih Q U Mh x: en CO 3 >i O 4-1 X! T3 •H QJ i-( >, 4J •H t rH U X3 N J* 0! CO CD a. xi [j-. CD X O W sQ, O CO u •H •H co 4J 0 >, 0 •H i— i j-i l-> CO c Cfl c E u CO >, i a) 4-1 4-1 < •H XI O H CO o CD u 4J o O CO m am E r •h ^ co U 4-> u CO 0) •H Oh CD h o CC CD 4-1 CO u CN G o •H CO o st O H ON ON CN CN 1 — O O O O • • • o o o H CO r^co — i .IT) CO 4-1 XI H CJ c o rH Cfl l_ CO X) 4-) r4 M o 4J a E G xi c a 0 C GO :< c o CO CO CJ Cfl 00 o c o i cm in O rH O O o' d N vO >D sf rH o o o o" I I c 3 o I 00 ON ON O co m on O rH CO rH r-on cn CN ON \0 I CO CN 00 I • • • I rH O rH I I r-co ~s XI CO w X w Ih CD rH 4J rH C CO iH

PAGE 232

222 m j-> •H 9) C! CO 3 c 00 00 N -3 u •H fa — o CO '-n *J c o o(J •H o u w X ex W X oo fo w on oox i-H c re H c u 4-> ri o H 4J 3 CD- (0 00 ^ -a CJ C CDo u CO 4J fa u a U 3 CO 3 E X a 0 W £, 3 0 CO T3 -H CD T— I '"N 4-1 -H "H U N CD CO v_/ Q. XJ fa X O fa -h Da u •H 4-> O O -H 4-> 4-1 CO E Sh >•, I CO 4-> < U CO CU QJ P 4-1 CD CO E E *H CO -H 0Q. U 4-1 CO CO fa CD O 00 QJ u re u co o X! LO cn O X c o CN o On r-H lOO ro o CN O rH | O O O I o I o I o o 5 o I o o I co cjo X c i — i \0 CO X c i-H vO CN X) o CN I — 1 c 1 — 1 O i — i • — CO C o o c o O o o co O X i-H c — 1 CN LO o co LO a' — 1 r> O in go OJ o CN CN in iC o — l 3 1 — I O T-H i — i rH o c CN X) 1 — X CO oo o CN *J OJ co CO CN co co X co 00 vO co o X) 1 — 1 m CO r- o m X i — i 00 co CO i-H r- o> m o> o o> 00 co i-H o> — i 1 — o o o r-H o o i — i 1 — 1 1 — 1 in •o — CD _C o Cj 1 — 1 CD X H o x: 4J r-H X H H U o 4J •H CO u CO >> J o •H i — i CN 3 u 3 3 u I — 1 r-H 1 — 1 •H c cc c •H a 0) a X ~ X CO a E 0) CO CU 0/ CJ CJ Vh -o 4H r-H H •H ~ xs TJ — TJ i-H •H cc c Ih c 00 c c QJ rH CJ rH CJ •rH 1 — 1 c CU Ih 0) c CD Ih CJ 00 •rH 00 H 00 JTJ H >H CG ,H -rH M RJ 3 -3 re -3 re u CO -a E -a to -a E a u U — 1 r-H i-H i-H rH CD 0) o -3 o CC H oo •H CD •H cc H r— o — i 0 i-H 4-1 c J3 3 XI 3 -3 c x: a T3 TJ •H o 3 U 3 u 3 O 3 o -3 3 T3 -3 Yo Yo Yo Yo rl a O CO — — i r~re CJ X — i 3C! •H ^— Ih 3 3 ~ re CD 0 CJ E 3 rH H T3 CO T3 r-H 3 3 CJ H •H re oox: w E re 3 >-, CD X! H H r-H Ih (h H CD CD T3 "3 rH — i 3J u

PAGE 233

223 co H ai c CO c •H oo N c ^— ^ u H b< e T3 3 a co ^ 4-> c U o ^ 4J •H QJ iH *J •H "H U QJ CO w [in o M X 4-1 O O -H U 4-1 O. CO E U >> I CO 4-1 < M CO CD OJ 4-1 4-> Q) CO E E rl CO -H Vh 4-1 (0 CO (X CJ >. C DC CJ 4-) to CJ' if) cn Q CO O O I o I o I CN O oo r-~ oo m O O o o i i •J O H rH CO O co in a Cfl c c s — \ QJ T3 a. CC o> Cfl O CO *—< p — i CO C 0! X X m O V u OJ CO Cfl X CO 01 u o o 4J 0J CJ > H sz tfl •H CO o £ 4-1 rH QJ QJ 0 u o CQ u s~ CO QJ ai ai CM CN I O I CO o CN I o m co i co 00 00 m i oo I • I QJ Cfl 3 u o a B C cfl CO u E O 3 c i oo CO Q C c I o o CT* t — i rH CN CC I o m a o I <4H of sz Cfl — 1 OJ 4-1 CN AI •H CN Cfl C CN CO 14-1 X 03 a X cfl E J3 >, CO 0 CO CO a> O n QJ o 3 a. fw z CO s u eg

PAGE 234

224 Cfl 4-1 H M 3 ^ C -H tO CON JS c ^ O -H fa T3 § 0) tfl 4-1 C o o O 1 0) U a. w H C o CO 4-1 4-1 o 4-1 3 CO X c H C o 0)^3 CO OO'-v c oco U II U < CD 00 a. c e X CO o U 14-1 •H X CO to 01 01 4-1 4-1 01 CO £ e CO -H (-1 4-1 to CO CU 0) •H oa >> Ih o CO a j ra u 00 CO 1— I o c o m — i i o I • i o i m O o i 4-1 01 1—1 '"N o o O 4-1 -H "H f — N o> U/iN > 1 1 CO 4-> < rm o CN CO I?) 00 in CO o CN in o O o CO O c 0> C J in CN Ih CJ -> ra o ra a< o E i— 1 rH i~> 4-1 co QJ ra H o O X H H CJ CO CJ c c ra co cj E c o n u 3 W 'O CN X tfl CO o ra CJ 3 CJ 0> o o m CO CJ 3 C crco o 01 CN CN X ra CJ E 4-1 O Ih CJ X2 E 3 CO O o 3 d i CO 1 — 1 c o c cri St C-] O I m o o o I o J3 E o Ih ra ra ra a cc CN X (0 Ih ra o Q

PAGE 235

225 CO 4-1 •H 0) c 00 3^ C -H CO OON x: c ^ U -H [X tu a-) c U o oH CD O w X P. X CO m W c amo •H oox — i c co •H C 4-1 *J H o rH 4-1 3 0) on CD C CD•H a S-. CO X u x: w CO u CD u ^ a CO Cc E X CO 0 w J3 U 4H 4J -a -h 0) rH 4-> -H H O XI CD CO x o W 1c o •H 4-> O O -H 4J 4J a. co E in >, I tO 4-1 H 4-1 CO CO CD Ih 0 DO CD 4-) CO OJ cn CO CN O I O I I o o I c I s cn rm CN cn cn CN I o o i X cn Q o O l CN . CD CO [fl rH X CO •H ra CO CO O E u E 4-1 rH t-H OT Ih CO CO r— rH C) u 14-1 o O O 4-1 OJ m Ih -a -c cn O cn o ~D 3 C e X O X 13 T3 CD o CO CO 14H CD c 3 u •H CO X ca CO co 4J CD CD CD CD to HJ cn CO U E E N CO •o c 0J a 3 rr CO O O •H c CO E o o :/: U u [fl u 3 4-1 O x x CD c c CD C 00 u H H 4J 1—1 H X cx c c 3 o 1 — M o c o ca 11 X 4-J M-l O c 0 1 1 4J u u 4-1 ca c 0 u 0) jC 4J c H Ih O <4H o c o oG m 0 > 1 CO s: o 4J ir, E i— 1 CO 4J — o -a ai CO Ih H CO 3 co — H X Ih O rH i-H T3 CO C CO if-l o 3 O CO H c CO U 11 CO E Ih 0) co CD JH E _c 3 4J H OJ -H 4H 11 CO _3 Ih *J CO DQ O IH 3 0 •H OH i— 1 o 01 4-> CJ U a 3 a HJ rH oh CO <4H H >
PAGE 236

REFERENCES Amemiya, T. "Regression Analysis When the Dependent Variable is Truncated Normal," Econometrica 41 ( 1973) : 997-1016 Amemiya, T. "Tobit Models: A Survey," Journal of Econometrics 24(1984): 3-61. Anonymous. "Conference Told to Step Up Advertising, PR," Seafood Business Report 3(Winter 1984): 21. Becker, G. S. Mandatory Federal Inspection System: An Overview Washington, D.C.: Congressional Research Service, The Library of Congress, TX 501 B, Report No. 83-198-ENR, 1983. Blokland, J. Continuous Consumer Equivalence Scales The Hague: Martinus Nighoff, 1976. Bockstael, N. E. "Uncertainty about Consumption and Consumer Uncertainty," Marine Resource Economics 1( 1984) : 67-77 Brown, J. A. C, and A. S. Deaton. "Surveys of Applied Economics: Models of Consumer Behavior," The Economic Journal 82(1972): 1145-1236. Burk, M. C. Influences of Economic and Social Factors on U.S. Consumption Minneapolis: Burgess Publishing Co., 1961. Buse, R. C. Data Problems in the BLS/CES PU-2 Dairy Tape: The Wisconsin 1972-1973 CES Dairy Tape Madison: University of Wisconsin, Agricultural Economics Staff Paper 164, July 1979. Buse, R. C, and L. E. Salathe. "Adult Equivalent Scales: An Alternative Approach," American Journal of Agricult ural Economics 60(1978): 460-468. Capps, 0., Jr. Analysis of Aggregate Fish and Shellfish Expenditure Blacksburg: Virginia Polytechnic Institute and State University, Agricultural Economics Bulletin 82-1, May 1982. Cato, J. C, and F. J. Prochaska. "Market Development in the Midwestern U.S. for Gulf and South Atlantic Seafoods from 1977 to 1980," Proceedings of the Sixth Annual Tropical and Subtropical Conference of the Americas College Station: Texas A&M University, TAMU-SG-82-101 1981. 226

PAGE 237

227 Christy, F. T., Jr., and A. Scott. The Common Wealth in Ocean Fisheries Baltimore: The Johns Hopkins Press, 1965. Cox, T. L. R. F. Ziemer, and J. Chavas "Household Demand for Fresh Potatoes: A Disaggregated Cross-Sectional Analysis," Western Journal of Agricultural Economics 9( 1984) : 41-57 Currie, J. M., A. J. Rayner, and J. Stewart. "Postscript." In W. J. Thomas (editor), The Demand for Food: An Exercise in Household Budget Analysis Manchester, England: Manchester University Press, 1972:122-133. Eastwood, D. B., and J. A. Craven. "Food Demand and Savings in a Complete, Extended, Linear Expenditure System," American Journal of Agricultural Economics 63( 1981) : 544-549. Eckert, R. D. The Enclosure of Ocean Resources: Economics and Law of the Sea Stanford, California: Hoover Institution Press, 1979. George, P. S., and G. A. King. Consumer Demand for Food Commodities in the United States with Projections for 1980 Giannini: University of California Giannini Foundation Mono. 26, March 1971. Gillespie, S. M., and M. J. Houston. An Analysis of Seafood Consumption Patterns and Product Perception in Texas College Station: Texas ASM University, TAMU-SG-75-216 1975. Greene, W. H. "On the Asymptotic Bias of the Ordinary Least Squares Estimator of the Tobit Model," Econometrica 49( 1981 ): 505-513 Gronau, R. "Leisure, Home Production, and Work — The Theory of the Allocation of Time Revisited," Journal of Political Econom y 85(1977): 1099-1123. Haidacher, R. C, J. A. Craven, K. S. Huang, D. M. Smallwood, and J. R. Blaylock. Demand for Red Meats, Poultry, and Fish Washington, D.C.: United States Department of Agriculture, Economics Research Service, National Economics Division, September 1982. Hassan, Z. A., S. R. Johnson, and R. Green. Static and Dynamic Demand Functions: An Application to Canadian Data Ottawa: Information Division Agriculture Canada, No. 77/14, November 1977. Hicks, J. R. Value and Capital London, England: Oxford University Press, 1957. Leser, C. E. V. "Forms of Engel Functions," Econometrica 31(1963): 694-703.

PAGE 238

228 Longwoods Research Group Limited (The). "A Usage Segmentation Analysis of the 1981 U.S. Seafood Consumption Study (final report)," prepared for the Fisheries Council of Canada, October, 1984. Maddala, G. S. Econometrics New York: McGraw-Hill Book Co., 1977. McDonald, J. F., and R. A. Moffitt. "The Uses of Tobit Analysis," Review of Economics and Statistics 62( 1980) : 318-321 Mincer, J. "Market Prices, Opportunity Costs, and Income Effects," In Carl Chrest (editor), Measurement in Economics Stanford: Stanford University Press, 1963:66-82. Morrison, R. W., and W. Armbruster. "Generic Adverstising of Farm Products," National Food Review NFR-23( 1983) : 14-18 Muellbauer, J. "Household Composition, Engel Curves, and Welfare Comparisons between Households: A Duality Approach," European Economic Review 5( 1974) : 102-122 Murphy, P. E., and W. A. Staples. "A Modernized Family Life Cycle," Journal of Consumer Research 6( 1979) : 12-22 O'Rourke, A. D. "Marketing and Distribution Problems with Extended Jurisdiction." In L. G. Anderson (editor), Economic Impacts of Extended Jurisdiction Ann Arbor: Ann Arbor Science Publishers Inc., 1977:237-246. Perry, J. S. "An Econometric Analysis of Socioeconomic and Demographic Determinants of Fish and Shellfish Consumption in the United States." Ph.D. dissertation, Gainesville: University of Florida, 1981. Phelps, C. E. "LIMDEP — A Regression Program for Limited Dependent Variables." International Publication, The Rand Corporation 1972. Prais, S. J., and H. S. Houthakker. An Analysis of Family Budgets Cambridge, England: Cambridge University Press, 1955. Prochaska, F. J., and R. A. Schrimper. "Opportunity Cost of Time and Other Socioeconomic Effects on Away-From-Home Consumption," American Journal of Agricultural Economics 55( 1973) : 595-603 Purcell, J. C, and R. Raunikar Analysis of Demand for Fish and Shellfish Athens: University of Georgia, Department of Agricultural Economics Bulletin 51, December 1968. Redman, B. J. "The Impact of Women's Time Allocation on Expenditure for Meals Away from Home and Prepared Foods," American Journal of Agricultural Economics 62( 1980) : 234-237 ~~

PAGE 239

229 Salathe, L. E. "Household Expenditure Patterns in the U.S.," Washington, D.C.: U.S. Department of Agriculture, Economics, Statistics, and Cooperative Service, Technical Bulletin No. 1603, April 1979. Slavin, J. W. "Review of U.S. Seafood Market," Infofish 2(1984): 22-26. Tobin, J. "Estimation of Relationships for Limited Dependent Variables," Econometrica 26( 1958) : 24-26 Tomek, W. G. "Empirical Analyses of the Demand for Food: A Review." In R. Raunikar (editor), Food Demand & Consumption Behavior Athens: University of Georgia, March 1977. United States Department of Agriculture, Human Nutrition Service. Food Consumption: Households in the United States, Seasons and Year 1977-78, Nationwide Food Consumption Survey 1977-78 Washington, D.C.: United States Department of Agriculture, Human Nutrition Service, Report No. H-6, 1983. United States Department of Commerce, Bureau of the Census. Statistical Abstract of the United States Washington, D.C.: Bureau of the Census, 1970-1985. United States Department of Commerce, NOAA NMFS Fisheries of the United States Washington, D.C.: NOAA, NMFS, 1965-1984. Van Dress, M. G. Dinning Out: Separate Eating Places Keep Customers Happy, Suppliers Busy Washington, D.C.: United States Department of Agriculture, Economics Research Service, Agriculture Information Bulletin No. 459, July 1983. Vondruska, J. "U.S. Consumer Attitudes toward Fish," Infofish 3(1984): 30-34. Vondruska, J. "Trends in U.S. Markets for Processed Shrimp." Unpublished paper presented at the annual meeting of the National Shrimp Processors Association, Lake Buena Vista, Florida, February 20-23, 1985. Wil son, J. R. "Seafood Production, Markets, and Policies: United States." In B. Thaker (editor), S eafood Production, Markets, and Policies: Belgium, Federal Republic of Germany, Ireland, The Netherlands, The United States Corvallis: Oregon State University, 0RESU-X-82-002 1982:108-157.

PAGE 240

BIOGRAPHICAL SKETCH Walter R. Keithly, Jr., was born November 6, 1954. He was raised in Berwyn, Pennsylvania, a small suburb outside Philadelphia. After completing high school, he entered the University of Delaware located in Newark, Delaware. There he received a Bachelor of Science degree in the spring of 1977. Later in that year he enrolled in the Graduate School at the University of Florida to pursue a graduate degree in the Department of Food and Resource Economics. After receiving a Master's degree in 1981 he enrolled in the Doctorate program in the same department. Walter is now employed at the Center for Wetland Resources, Louisiana State University, in beautiful downtown Baton Rouge: home of the 6th Annual Sports Festival. 230

PAGE 241

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. "red J. PrSchAska, Chairman Prof essorSjf-'Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. James C. Cato Professor of Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of J. Scott Shonkwiler Associate Professor of Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Richard L. Kilmer Associate Professor of Food and Resource Economics

PAGE 242

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Associate Professor of Food Science and Human Nutrition This dissertation was submitted to the Graduate Faculty of the College of Agriculture and to the Graduate School, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy August, 1985 Dean, Graduate School


Table 3-2. Continued
Category
Parameter
estimates
a.
i
Asymptotic
t-ratio
Expected
probability
F(Zi)
Expected total
change resulting
from change in
3 E(EXP)b
3X.
Expected change
among consuming units
3E(EXP*) F(Z.)
9Xi
(1)
(2)
(3)
(4)
(5)
(6)
Meals away from home:
Dollars (X28)
-0.0120
-3.598
.4759
-0.0057
-0.0020
Income before taxes:
Thousand dollars (X29)
Thousand dollars
0.0817
3.572
.4759
0.02518
0.0088
squared (X30)
-0.0003
-2.619
Interaction terms:
Income and race (X31)
Income and family
-0.0416
-2.086



size (X32)
0.0047
1.577





Table B-4. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Actual mean
values
Interaction terms:
Income and race (X31)
12.745
14.992
12.410
Income and family size (X32)
46.628
58.023
44.929

Other seafood:
Dollars (X33)
0.763
0.665
0.778

Weekly expenditures and quantity:
EXP
0.335
2.582
0.000
Q
0.208
1.600
0.000

Number of households:
10,689
1,387
9,302

Percent of households:
100
12.98
87.02

Si
The data provided in this table associated with the binary variables (X1-X23) should be interpreted as
representing proportions rather than percentages. To obtain percentages, multiply data by 100.
170


136
Outlook for Increasing At-Home Demand for
Specific Product Forms and Implications
Much of what was discussed in the last section of the previous
chapter extends to the discussion in this section. As mentioned,
generic seafood advertising of seafood is minimal when compared to
generic advertising of either meat or dairy products. As also men
tioned, the majority of nongeneric seafood advertising is directed at
increasing sales of only canned salmon and canned tuna. These factors
alone indicate a disadvantage in long run growth in at-home consumption
of most seafood products compared to meat, poultry, and dairy products.
Though it is difficult to predict future at-home consumption due
to the simultaneous shifting of all factors that determine consumption
of the specific product forms, a couple of general comments concerning
expected future demand for these products can be made based on the
preceding analysis. In terms of at-home consumption of the specific
product forms, the potential for long run growth appears greater in the
fresh and frozen seafood product markets than in the canned seafood
market for several reasons. First, the estimated income elasticities
of at-home quantity consumption of fresh seafood (0.413) and frozen
seafood (0.251) were about three to four times the size of the
estimated income elasticity of at-home consumption of canned seafood
(0.098), even though the majority of the nongeneric advertising dollars
goes towards the promotion of canned seafood products. Thus, increas
ing household real income should promote more long term growth in fresh
and frozen seafood consumption than canned seafood consumption. Second,
the growing proportion of Black households in the United States should
benefit growth in consumption of fresh seafood products since as the


106
Table 3-
-9. Median family income
selected years
in constant (1982) dollars
for
Year
Race of Household
All
White
Black
1950
13,308
13,813
7,494
1955
15,926
16,629
9,170
1960
18,317
19,018
10,528
1965
21,283
22,183
12,216
1970
24,528
25,445
15,608
1975
24,664
25,589
15,744
1980
24,626
25,658
14,846
1982
23,433
24,603
13,598
SOURCE:
United States Department
of Commerce, Bureau of the
Census
(1984).


90
this with an overall increase in seafood purchases among this group of
consumers.
Household Caught Fish for Own Use
As expected, households which caught fish (X21) consumed a greater
quantity of seafood than did their counterparts. Similarly, weekly
expenditures on seafood consumed at home were higher among households
4
who caught fish than among those who did not. As evidenced by the
estimated probabilities given in column 4 of Table 3-3, having caught
fish was second only to race among the discrete variable in determining
whether a household consumed seafood.
Employment of the Meal Planner
Employment of the meal planner (X22) was not statistically
significant in explaining weekly expenditures or quantities of seafood
consumed at home. The estimated signs of the respective coefficients,
however, were negative as expected given the increased opportunity of
the meal planner's time when employed. In the next chapter, considera
tion will be given to the effect of employment by the meal planner on
consumption of those products which require the most preparation time.
Sex of the Meal Planner
The sex of the meal planner was significantly related to at-home
consumption of seafood, ceteris paribus. Households with female meal
4
If the fish consumed during the one week interview period
consisted of that which had been caught rather than purchased, the
price assigned to that product was based on the average retail price of
a comparable product in that region and season.


18
Cross-Sectional Demand Analysis Considerations
Price, Income, and Socioeconomic Variable Considerations
Two features of cross-sectional demand analysis of concern in
doing applied research in this field relate to proper model specifica
tion and method of estimation. In terms of model specification,
concern has traditionally centered around variable selection. In
practice, researchers have estimated demand functions (Equation 2.4a)
when using cross-sectional data as follows:
qiJ qij(mj Zj)
where
(i = 1, 2, ..., n)
(J = 1, 2, ..., k)
(2.10)
= quantity of the ith commodity demanded by the
Jth household
mj = income of Jth household
Zj = a column vector of socioeconomic characteristics
particular to the Jth household
The demand functions specified in (2.10) differ from those given by
(2.4a) in three aspects. First, a vector of socioeconomic characteris
tics (Zj) whose values are specific to the Jth household has been
included in the latter equation. Second, the vector of prices (p) has
been excluded from the latter equation. Third, all variables have been
subscripted denoting the Jth household.
Given the specification of the cross-sectional demand functions
(2.10), estimation entails the use of a sample of households differing
in socioeconomic characteristics and income rather than that of a
representative household as presented in (2.4a). Differences in


CHAPTER IV
SPECIFIC SEAFOOD PRODUCT FORM ANALYSIS
Introduction
In this chapter, a discussion of the parameter estimates
associated with the disaggregated Tobit seafood models is presented.
Given the voluminous amount of information associated with the specific
product form models, discussion centers on noting and explaining
differences in the signs and significance of the estimated parameters
among the seafood product consumption models and with respect to the
total consumption models. This is in contrast to the in depth discus
sion given to each parameter estimate in the previous chapter. All
parameter estimates are, however, presented in Appendix B.
Some descriptive statistics associated with the data used in
estimating the specific seafood product form models (fresh, frozen,
canned, finfish, and shellfish) are presented in Table 4-1. Among the
alternative types of processing (fresh, frozen, canned), canned seafood
is consumed at home by the most households and frozen seafood by the
least households (Table 4-1). In fact, canned seafood is served by
more households (31.69 percent) than fresh and frozen seafood combined
(29.63 percent). However, referring back to Figure 1-1, per capita
consumption of canned seafood has equalled only about 60 percent of
that of fresh and frozen in recent years (edible basis). Thus, it must
be concluded that consumption of fresh and frozen seafood products is
111


57
f
^fa
s'
households consuming seafood at home, 29.9 percent resided in the
Northeast region. The third category, labelled limit observations,
provides the mean value of all variables for those households reporting
* ^
a zero level of at-home consumption of seafood during the one week
survey period. On this basis, the information contained in Table 3-1
suggests that of those households reporting no at-home seafood consump
tion 19.5 percent resided in the Northeast region. The final category
in Table 3-1, labelled proportion consuming, gives the value of the
proportion of households consuming seafood during the survey period
associated with each of the binary variables. For example, 61.2
percent of the households residing in the Northeast region of the
United States consumed seafood during the interview period.
Approximately one half (50.8 percent) of the 10,689 households
included in the analysis consumed seafood at home during the survey
period. Among households consuming seafood at home, average expendi
tures and consumption of seafood equalled $2.93 and 1.91 pounds,
respectively. For the total sample, average at-home weekly consumption
of seafood was 0.97 pounds valued at $1.49 or approximately one-half
the volume and value estimated for consumers only. Placed on a yearly £>JL
basis, at-home consumption of seafood by an average household thus &''
1/7
P
/
equals just over 50 pounds, or about 18 pounds per capita assuming an
( average of 2.75 members per household. This value equals about one-
half of the approximately 35 pounds (round weight) annual reported per
capita consumption of seafood in the United States during the 1975-77
period.
With respect to region, households in the Northeastern region of
the United States (XI) had a higher probability of consuming seafood at


25
As a final example, the authors consider the case in which non
consumers are observed only because the time period represented by the
survey does not cover a sufficient span of time necessary to observe
consumption by most or all households. As in the previous case, there
exists no reason to expect nonconsumers to exhibit a different
behavioral pattern from that of consumers and hence there exists no
economic reason for excluding them from the analysis.
Summarizing these three cases, if the differences between con
sumers and nonconsumers can adequately be accounted for, then there
exists no reason, from a methodological viewpoint, for excluding the
nonconsumers. Though there is no methodological rationale for exclud
ing the nonconsumers, caution must be taken when including nonconsumers
in the analysis because of statistical problems. As discussed later,
use of ordinary least squares when the data includes a large percentage
of nonconsumers will generally be inappropriate because of resulting
biased and inconsistent estimates of the true parameters. An appro
priate statistical technique to be used in conjunction with problems of
this nature will be presented towards the end of this chapter.
Related Seafood Consumption Studies
When compared to cross-sectional demand studies on those food
commodities which comprise a large percentage of the consumers food
budget, cross-sectional demand analyses for seafood products tend to be
somewhat limited. This probably reflects a lack of consideration of
"nontraditional" agricultural commodities in such surveys until recent
years.


Table 3-3. Continued
Category
Parameter
estimates
^i
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
9E(Q)b
3Xi
Expected change
among consuming units
3E(Q*)
3Xi
F(Z.)
(1)
Sex of meal planner:
(2)
(3)
(4)
(5)
(6)
Female (X23)
0.4437
2.344
.4499
0.1996
0.0672
Male (base)
Family size:
Total number in
.4060
household (X24)
Total number
0.3759
2.710
.4469
0.1579
0.0530
squared (X25)
Education of meal planner:
-0.0117
-0.896
Years (X26)
Guest meals:
0.0662
3.809
.4469
0.0269
0.0099
Number of meals (X27)
0.1201
7.069
.4469
0.0037
0.0180


91
planners (X23) consumed greater quantities of seafood and had greater
expenditures than households with male meal planners. This may reflect
a female's familiarity with cooking procedures, etc.
Household Size
The household size was found to be an important determinant of
at-home seafood consumption which is consistent with related studies
(e.g., Capps, 1982; Salathe, 1979). The positive linear household size
coefficient (X24) and the estimated negative coefficient associated
with household size squared (X25) imply that weekly expenditures and
quantity of seafood consumed increase with increases in household size
but at a declining rate. Because of the nonlinear specification of the
household size component in the models it is useful to evaluate the
effect of household size on seafood consumption at various levels of
household size. The more important effects are presented in Table 3-5.
As the information in Table 3-5 suggests, within the relevant
range very little economies to size are exhibited in either seafood
expenditures or quantities consumed. This is not surprising given the
small magnitude of the estimated parameter associated with the squared
term of household size (X25) compared to the linear term (X24) in the
two models. The information provided in the table indicates that total
weekly seafood expenditures decline with the addition of a fourth
member while the total quantity consumed declines with the addition of
a fifth household member. Within household sizes generally encountered
in the data each additional household member resulted in increased
consumption of 0.158 pounds per week with related expenditures
increasing about $0.26.


Table B-8. Continued
Category
Parameter
estimates
i
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
8 E(EXP)b
9Xi
Expected change
among consuming units
3E(EXP*) F(Zi)
9Xi
(1)
Income before taxes:
(2)
(3)
(4)
(5)
(6)
Thousand dollars (X29)
Thousand dollars
squared (X30)
Interaction terms:
-0.0017
-0.0000
-0.156
-0.548
.3483
0.0056
0.0016
Income and race (X31)
Income and family
-0.0002
-0.019



size (X32)
Other seafood:
0.0036
2.749
Dollars (X33)
Constant:
0.0073
0.941
.3483
0.0025
0.0007
OQ
-3.1487
-12.006



3The value of XB at the i
means of all X^'
s is equal
to -0.78992; o
= 2.0310.
The effects of the interaction and/or squared terms have been accounted for in the construction of the
associated linear terms.
190


79
suburban areas (X5) or nonmetro areas (base), ceteris paribus.
Similarly, households in suburban areas were estimated to have higher
weekly at-home consumption of seafood than those households in nonmetro
areas, ceteris paribus. Central city households had total expected
weekly expenditures (quantity consumed) equal to $0,564 (0.355 pounds)
in excess of those households in nonmetro areas. Households in
suburban areas had total expected weekly expenditures (quantities)
equal to only $0,231 (0.190 pounds) greater than households in nonmetro
areas.
Since proximity to the coast was a major factor leading to the
development of many of the larger cities in the United States, house
holds in the larger cities may have greater access to a larger variety
and quality of seafood than those households in either suburban or
nonmetro areas. For example, New York City, Boston, and New Orleans
have major fish markets acting as central locations from which distri
bution of seafood products to other localities is coordinated. Due to
a decline in the accessibility of moderate cost quality seafood as one
moves away from the distribution centers, one would expect that the
probability of a household purchasing and hence consuming seafood to
decline in relation to distance from the distribution center. As noted
in column 4 of Tables 3-2 and 3-3, the estimated probabilities of
observing a positive level of expenditures and consumption of seafood
for a household residing in a central city area exceed those of a
household residing in a suburban or nonmetro area. These estimated
probabilities are consistent with the observed probabilities of
observing a consuming household given in column 4 of Table 3-1.


38
purchase of seafood products. No distinction was made with respect
to the sources of income and their separate effects on consumption of
total seafood and specific product forms. Though increases in house
hold income have been found to be related to increases in expenditures
on total seafood consumed at home (e.g., Perry, 1981; Capps, 1982),
less evidence exists concerning the relationship between income and
at-home consumption of specific seafood product forms. For example,
the estimated income parameter associated with a specific seafood
product form may be either positive or negative depending upon whether
that product form is considered to be a normal or inferior good and may
in fact even vary among different segments of the population.
Substitutes for at-home consumption of seafood and the specific
product forms, denoted as S in equations (2.11a and 2.11b), follow from
the concept of the demand function provided in equation (2.4a). In
most cross-sectional studies of this nature, substitutes are omitted
from the analysis because prices of substitutes encountered by any
given household should be the same prices as those encountered by any
other household. In the estimation of the total seafood consumption
models no substitutes were specified. However, with respect to the
models for specific product forms, consumption of the alternative
seafood product forms were considered as appropriate substitutes.
For example, shellfish consumption by a given household was related to
consumption of finfish by that same household. Similarly, consumption
of fresh seafood was related to the consumption of the summation of
8
Though an arguement could be made to use total food expenditures
rather than income as an explanatory variable in the analysis, the
latter variable was used because it is more directly applicable for
answering policy oriented questions.


45
E(y)
= XBF(Z) + of(Z)
where
Z
= n/o
f(Z)
9
= unit normal density function
F(Z)
= cumulative normal distribution function
(2.14a)
The unconditional expected value of the dependent variable represents
the expected value of the dependent variable associated with all
observations. Furthermore, as shown by Amemiya (1973), the conditional
expected value of the dependent variable for observations above the
limit (i.e., positive observations), y* is given by
E(y*) = E(y|y > 0)
= XS + E(u|y > 0) (2.15a)
= X3 + of(Z)/F(Z)
The relationship between the unconditional expected value of the
dependent variable (expected value associated with all observations)
and that of the conditional expected value of the dependent variable
(expected value associated with positive observations) is given as
follows:
E(y) = E(y*) F(Z)
(2.16a)
Defined as
10
Defined as
_1_ -Z2/2
2 it 8
B'X
L 2
o 1 -Z 12
- e


Pounds (edible weight)
Figure 1-1. U.S. annual per capita consumption of commercial fish and shellfish (edible weight), 1960-83
SOURCE: United States Department of Commerce, NOAA, NMFS (1983).


131
the one exception, were negatively related to income in both the linear
and squared income terms. A look at the parameter estimates associated
with the interaction of income and race (X31) reveals that though
statistical significance is noted only in the total, all parameter
estimates were negative in sign.
The expenditure and quantity income elasticities for the specific
product forms and in total are provided in Table 4-4. Among specific
product forms distinguished by level/type of processing, consumption of
fresh seafood had the highest income elasticity estimates (NgXp =
0.467, Nq = 0.413) while at-home consumption of canned seafood
exhibited the lowest income elasticity estimates (Ng^p = 0.192, Nq =
0.098). Consumption of shellfish had a significantly higher income
elasticity estimates (NgXp = 0.543, Nq = 0.929) than did finfish
(Ngxp = 0.148, Nq = 0.1163).
Table 4-4 also provides information on the decomposition of the
total elasticities into their respective components; increased
(decreased) consumption among consuming households due to an increase
(decrease) in income and increases (decreases) in consumption reflect
ing increases (decreases) in the number of participating households.
For those products in which the number of consuming households was
relatively small (e.g., shellfish) the proportion of the total elas
ticity reflecting changes in the number of participating households
was relatively large compared to those products in which the number
of consuming households was relatively large (e.g., finfish). For
example, about 80 percent of the total estimated income elasticity
associated with shellfish products consumed at home reflected changes
in the number of participating households compared to 60 percent


227
Christy, F. T., Jr., and A. Scott. The Common Wealth in Ocean
Fisheries. Baltimore: The Johns Hopkins Press, 1965.
Cox, T. L., R. F. Ziemer, and J. Chavas. "Household Demand for Fresh
Potatoes: A Disaggregated Cross-Sectional Analysis," Western
Journal of Agricultural Economics 9(1984):41-57.
Currie, J. M., A. J. Rayner, and J. Stewart. "Postscript." Iii
W. J. Thomas (editor), The Demand for Food: An Exercise in
Household Budget Analysis. Manchester, England: Manchester
University Press, 1972:122-133.
Eastwood, D. B., and J. A. Craven. "Food Demand and Savings in a
Complete, Extended, Linear Expenditure System," American Journal
of Agricultural Economics 63(1981):544-549.
Eckert, R. D. The Enclosure of Ocean Resources: Economics and Law of
the Sea, Stanford, California: Hoover Institution Press, 1979.
George, P. S., and G. A. King. Consumer Demand for Food Commodities
in the United States with Projections for 1980. Giannini:
University of California Giannini Foundation Mono. 26, March
1971.
Gillespie, S. M., and M. J. Houston. An Analysis of Seafood
Consumption Patterns and Product Perception in Texas. College
Station: Texas A&M University, TAMU-SG-75-216, 1975.
Greene, W. H. "On the Asymptotic Bias of the Ordinary Least Squares
Estimator of the Tobit Model," Econometrica 49(1981):505-513.
Gronau, R. "Leisure, Home Production, and WorkThe Theory of the
Allocation of Time Revisited," Journal of Political Economy
85(1977):1099-1123.
Haidacher, R. C., J. A. Craven, K. S. Huang, D. M. Smallwood, and
J. R. Blaylock. Demand for Red Meats, Poultry, and Fish.
Washington, D.C.: United States Department of Agriculture,
Economics Research Service, National Economics Division,
September 1982.
Hassan, Z. A., S. R. Johnson, and R. Green. Static and Dynamic
Demand Functions: An Application to Canadian Data. Ottawa:
Information Division Agriculture Canada, No. 77/14, November
1977.
Hicks, J. R. Value and Capital, London, England: Oxford University
Press, 1957.
Leser, C. E. V. "Forms of Engel Functions," Econometrica 31(1963):
694-703.


Table B-5.
Summary statistics for Tobit analysis of weekly household expenditures on frozen seafood
a
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
9E(EXP)b
aXi
Expected change
among consuming units
3E(EXP*) F(Z )
aXi
(1)
(2)
(3)
(A)
(5)
(6)
Region:
Northeastern (XI)
-0.5133
-1.909
.1230
-0.0631
-0.0114
North Central (X2)
0.3275
1.249
.1562
0.0512
0.0102
Southern (X3)
-0.1684
-0.648
.1357
-0.0229
-0.0046
Western (base)


.1423


Urbanization:
Central City (X4)
0.0736
0.325
.1401
0.0103
0.0020
Suburban (X5)
0.0860
0.419
.1401
0.0120
0.0023
Nonmetro (base)


.1379


Season:
Spring (X6)
0.0200
0.085
.1446
0.0029
0.0006
Summer (X7)
-0.4958
-2.047
.1251
-0.0620
-0.0114
Fall (X8)
-0.0822
-0.359
.1401
-0.0115
-0.0022
Winter (base)


.1446


171


129
consumption equalled about one-fifth of their respective total elas
ticities compared to slightly less than one-third for that associated
with canned seafood consumption (Table 4-3).
Education of Meal Planner
The relationship estimated between education (X26) and expendi
tures on at-home consumption of the various seafood products examined
in this study was positive in all instances and statistically signifi
cant with the exception of expenditures on fresh seafood products
(Table 4-2). The elasticities of expenditures on at-home consumption
of the various seafood product types and forms with respect to educa
tion of the meal planner were estimated to be 0.156 for fresh seafood
products, 0.606 for frozen seafood products, 0.405 for canned seafood
products, 0.205 for finfish products, and 1.249 for shellfish products;
compared to 0.552 for total seafood. Thus, increases in the level of
education of the meal planner is expected to have pronounced effects on
expenditures on at-home consumption of the seafood product forms. The
statistically insignificant education parameter estimate and relatively
low expenditures elasticity estimate associated with fresh seafood may
be the result of a confounded effect between an increase in the
opportunity cost of time of the meal planner which prevents him/her
from preparing fresh seafood and an increased desire for a nutritional
meal associated with additional education.
Number of Guest Meals
Increases in the number of guest meals (X27) were estimated to be
positively related to weekly expenditures on at-home consumption of


47
in nature. However, several of the variables evaluated in the present
study were binary in nature (X1-X23). The main consequence of such
a specification pertains to the evaluation of X3 given a binary
variable. Specifically, the probability of observing a positive
value of the dependent variable now becomes conditional on the binary
variable being evaluated. Thus, the value of the standard normal,
(Xg|Xi)/a, and hence the value of the cumulative normal density
function F(Z |X^) and the value of the unit normal density function
f(Z |X-^) all become dependent on the binary variable being evaluated.
Hence, equations (2.14a) through (2.21a) need to be adjusted accord
ingly when discussing binary variables. These adjustments are
E(y|x) =
(X6|X) FiZjX^ + ofiZjXi)
(2.14b)
E(y*|X) =
(XS|X.) + af (zi|xi)/F(zi| x^
(2.15b)
E(y |X) =
E(y*|X) F(ZilXi)
(2.16b)
8E(y|Xi)/8Xi =
F(ZilXi) OEirHx^/aXi)
+ E(y*|Xi) (3F(Zi|xi)3Xi)
(2.17b)
9F(Zi|Xi)/3Xi =
f (Zi|Xi) 3i/ 0
(2.18b)
=
Si + (a/F(Zi|Xi)) afiZilXii/aXi
- (afZilX^/FZilXi)2) BFCZilX^/SXi
(2.19b)
3E(y*|Xi)/3Xi =
B.[l (Z i |X i) f (Z i|X i)/F(Z i|Xi)
- f(Zi|Xi)2/F(Zi|Xi)2J
(2.20b)
3E(yfX.)/3X. =
FiZilxpSi
(2.21b)
Technically, it would be preferable to use the concept of a
limit rather than a derivative in evaluating equations (2.17b) through
(2.21b) due to the discrete nature of Xi in each of the equations.


Table 4-4. Estimated weekly expenditure, quantity and quality elasticities with respect to
before tax income for specific and total seafood product forms3
Category
Elasticity
among existing
households
(conditional elasticity)
Elasticity associated
with entry/exit of Total
households (elasticity elasticity
of participation)
Expenditures:
Fresh
Frozen
Canned
Finfish
Shellfish
Total
Quantities:
Fresh
Frozen
Canned
Finfish
Shellfish
Total
0.0899
0.0629
0.0561
0.0558
0.0958
0.3771
0.2403
0.1361
0.0927
0.4476
0.4670
0.3032
0.1922
0.1485
0.5434
0.0953
0.1436
0.2389
0.0795
0.0516
0.0310
0.0419
0.2059
0.3334
0.1994
0.0667
0.0744
0.7228
0.4129
0.2510
0.0977
0.1163
0.9287
0.0496
0.1318
0.1814
132


2
The reasons for the relatively low per capita consumption of fish
and shellfish in the U.S. compared to other developed countries of the
world are many and varied. First and foremost, the United States has
traditionally been the world's largest producer of beef and poultry
which has resulted in an abundant supply of these products at
relatively modest prices. The supply of edible fishery products, on
the other hand, has relied heavily upon imports to meet domestic
demand at acceptable prices. Second, in terms of ease of preparation,
fish and shellfish are typically rated poorly when compared to meat and
poultry products (Gillespie and Houston, 1975).1 This factor, in part,
has resulted in a large institutional and restaurant trade in seafood
products, while at-home consumption as a proportion of the total has
remained relatively low. Third, the demand for fish and shellfish in
the United States has been affected by the market distribution,
perishability, and preservation of these products (Christy and Scott,
1965). As one moves inland from those coastal states recognized as
major seafood producers, the availability of fresh seafood products
falls and the price increases. Finally, there remains a constant
concern among U.S. consumers regarding the quality of seafood being
sold in the various retail outlets. Meat and poultry products must be
inspected and certified by government representatives before sale
while inspection and certification of seafood products by government
representatives remains voluntary on the part of the seafood processor
(Becker, 1933). As such, inspection of seafood products has tradi
tionally been sporadic and minimal. For example, the most intensive
*83361 on a regional study in Texas conducted by the authors.


ACKNOWLEDGEMENTS
Numerous people have helped in making this study and my graduate
program possible. Sincerest appreciation is extended to my chairman,
Fred J. Prochaska. He was always there to give me encouragement and
advice even during strained times. He should be credited with my
success but not failure in the academic field.
Dr. Scott Shonkwiler gave freely of his time in helping me with
questions of a statistical nature. Similarly, Drs. Cato, Kilmer, and
Otwell devoted their time to assure a better quality product than would
otherwise have been the case. Hopefully, their constructive criticisms
of this study make it more useful for the groups for which it is
intended.
I wish also to thank the Food and Resource Economics Department at
the University of Florida and Florida Sea Grant for giving me the
opportunity to pursue a graduate program and for providing financial
support.
I wish to thank Janet Eldred for the typing she did on this
manuscript. It sometimes got complicated sending everything through
the mail.
Finally, I wish to thank my parents for giving me the opportunity
to pursue a graduate career. Without their support I would not have
made it this far. Unfortunately, now they want me to pay them back.
ii


Table B-l. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Actual mean
values
Interaction terms:
Income and race (X31)
12.745
11.756
12.943
Income and family size (X32)
46.628
47.275
46.498

Other seafood:
Dollars (X33)
0.780
0.783
0.780
Pounds (X33)
0.455
0.425
0.461

Weekly expenditures and quantity:
EXP
0.707
4.243
0.000
Q
0.516
3.097
0.000

Number of households:
10,689
1,780
8,909

Percent of households:
100
16.65
83.35

The data provided in this table associated with the binary variables (X1-X23) should be interpreted as
representing proportions rather than percentages. To obtain percentages, multiply data by 100.
155


15
Hence, for a good to be a Giffen good, the income effect must not only
be negative but also outweigh the own-price substitution effect.
Furthermore, a good whose income effect is negative is referred to as
inferior and conversely, a good whose income effect is positive is
referred to as a superior good. An Engel curve which relates consump
tion or expenditures on a good with income will thus be downward
sloping in the case of an inferior good and upward sloping in the case
of a superior good.
A second restriction offered by economic theory is that demand
functions (2.4a) are homogeneous of degree zero in all prices and
income. This suggests that an arbitrary scaling of all prices from
(p, m) to (ap, am) will have no effect on quantity demanded of any
good.
Unfortunately, these two restrictions have little to offer when
estimating the demand for a single good. Though the own-price
substitution effect is negative, the demand curve can be upward sloping
given that the good being analyzed is inferior and inferior to the
extent that the income effect outweighs the own-price substitution
effect. Thus, no restrictions can be imposed, a priori, on the sign of
2
the own-price coefficient. Furthermore, since a good can be either
inferior or superior, no restrictions can be imposed, a priori, on the
sign of the income coefficient. The second restriction is also of
little value in the estimation of single demand equations since rarely
if ever are all prices included in a single demand equation. Though
these restrictions offer little value in the estimation of single
2
This is of little concern in empirical demand analysis since few
if any goods have ever been shown to be Giffen goods.


Table 3-7.
Estimated effects of changes in before tax income on weekly expenditures and quantities of
seafood consumed3
Component
Before
tax income
5
10
15
20
25
Avg.
Weekly expenditures ($):
Total expected change in
consumption 8E(EXP)/3X^
.0260
.0260
.0258
.0254
.0248
.0252
Expected change attributable
to consuming households
[3E(EXP*)/3X1] F(Z)
.0070
.0088
.0090
.0092
.0093
.0088
Expected probability of
consumption F(Z^)
.4347
.4578
.4797
.5005
.5201
.4759
Change in expected probability
of consumption 3F(Z)/3X^
.0048
.0046
.0043
.0041
.0039
.0043
Weekly quantity consumed (lbs):
Total expected change in
consumption 3E(Q)/3X
.0137
.0134
.0130
.0125
.0119
.0129
Expected change attributable
to consuming households
[3E(Q*)/aXi] F(Zi)
.0044
.0044
.0044
.0043
.0042
.0043
Expected probability of
consumption F(Z^)
.4149
.4352
.4501
.4641
.4771
.4469
Change in expected probability
of consumption 3F(Z^)/3X
.0032
.0031
.0029
.0027
.0025
.0029
a
Evaluated at a family size (X24)
equal to 3.153
and a
proportion of
White households
(X18)
equal to
0.813.
102


Table B-5. Continued
Expected total
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z.)
change resulting
from change in X^
3E CEXP)b
9Xi
Expected change
among consuming units
3E(EXP*) .F(Z .)
3X .
i
(1)
(2)
(3)
(A)
(5)
(6)
Household life cycle:
Young single w/o
children (X9)
-0.6930
-1.205
.1230
-0.0852
-0.0154
Young married w/o
children (X10)
-0.6245
-1.309
.1271
-0.0794
-0.0145
Young single with
children (X11)
-1.0718
-1.688
.1093
-0.1171
-0.0203
Young married with
children (X12)
-0.3712
-0.905
.1357
-0.0504
-0.0095
Middle aged single
w/o children (X13)
-0.3847
-0.755
.1357
-0.0522
-0.0098
Middle aged married
w/o children (X14)
0.1864
0.492
.1587
0.0296
0.0059
Middle aged single
with children (X15)
-0.1134
-0.231
.1469
-0.0167
-0.0032
Middle aged married
with children (X16)
-0.3152
-0.773
.1379
-0.0435
-0.0082
Elderly single (X17)
-0.3425
-0.732
.1379
-0.0472
-0.0089
Elderly married (base)


.1515


172


Table B-12. Continued
Category
Parameter
estimates
Pi
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X,-
3E(Qr
3X
Expected change
among consuming units
3ECQ*) F(Z,)
3X
(1)
Income before taxes:
(2)
(3)
(A)
(5)
(6)
Thousand dollars (X29)
Thousand dollars
0.0289
1.747
.4280
0.0067
0.0021
squared (X30)
Interaction terms:
-0.0001
-1.376
Income and race (X31)
Income and family
-0.0229
-1.587



size (X32)
Other seafood:
0.0027
1.235
Pounds (X33)
Constant:
0.0385
1.355
.4280
0.0165
0.0053
SO
-2.2180
-5.534



a
The value of XB at the i
means of all X^1
s is equal
to -0.6566; o =
3.5125.
^The effects of the interaction and/or squared terms have been accounted for in the construction of the
associated linear terms.
210


Table B-l. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Receives food stamps:
Yes (X20)
No (base)
Caught fish for own use:
Yes (X21)
No (base)
Employment of meal planner:
Yes (X22)
No (base)
Sex of meal planner:
Female (X23)
Male (base)
Mean (percent)3
0.074 0.095 0.070 0.214
0.926 0.905 0.930 0.163
0.234 0.298 0.221 0.212
0.766 0.702 0.779 0.153
0.465 0.429 0.472 0.154
0.535 0.571 0.528 0.178
0.908 0.928 0.904 0.170
0.092 0.072 0.096 0.130
153


Table B-12. Continued
Category
Parameter
estimates
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X,-
3E(Q)b
3Xi
Expected change
among consuming units
8E(Q) F(Z-,)
3X
(1)
Sex of meal planner:
(2)
(3)
(4)
(5)
(6)
Female (X23)
0.4406
2.583
.4801
0.2115
0.0677
Male (base)
Family size:
Total number in
.3821
household (X24)
Total number
0.5427
4.380
.4280
0.1796
0.0557
squared (X25)
Education of meal planner:
-0.0257
-2.194
Years (X26)
Guest meals:
0.0391
2.509
.4280
0.0167
0.0053
Number of meals (X27)
Meals away from home:
0.0780
5.068
.4280
0.0334
0.0107
Dollars (X28)
-0.0121
-4.922
.4280
-0.0052
-0.0016
209


Table 4-2. Continued
Variable
Seafood
product
form
Total
Fresh
Frozen
Canned
Finfish
Shellfish
Interaction terms:
Income and race (X31)
_
Income and family size (X32)
Other seafood:
+
+
+*
+
+
Dollars (X33)
Constant:
NAC
-
+
+*
+*
o
Parameter estimates, t-values, and associated information are provided in Appendix B.
b
+ indicates that parameter estimates was positive; indicates that parameter estimates was negative;
* indicates that parameter estimates was statistically significant at the 90 percent level (t value >
1.654).
c
Not applicable.
118


Household Life Cycle
The household life cycle category (X9-X17) appears to be very
useful in explaining weekly household at-home seafood consumption as
judged by the number of statistically significant parameter estimates.
Household life cycle estimates suggest that the composition of the
household, independent of size, explains both expenditures and quanti
ties of seafood consumed at home in a rather systematic and logical
manner. Furthermore, the manner in which consumption of seafood can be
explained via the household life cycle is as expected, given the
current understanding of the at-home seafood market. For example,
there appears to be a general tendency for increased consumption of
seafood associated with the maturing of the household. Households with
the household head less than 35 years of age (X9-X12) consistently
consumed less seafood than households in more mature life cycle
categories. Households with the household head from 35 through
64 years of age (X13-X16) generally consumed less seafood than house
holds comprised of an elderly married couple (base). Given that the
difference in household size between that of an elderly individual
(X17) and that of an elderly married couple (base) has been accounted
for by the variables representing household size (X24 and X25), the
estimated difference in at-home seafood consumption between these two
groups of households may represent differences in eating habits. For
example, elderly individuals may not wish to cook only for themselves,
especially those items requiring any amount of preparation time. This
would preclude them from consuming all but canned seafood which takes a
minimal amount of preparation before being suitable for consumption.


Table B-5. Continued
Category
Parameter Asymptotic
estimates t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in
9E(EXP)b
ax.
Expected change
among consuming units
3E(EXP) FCZj)
3X i
(1)
Sex of meal planner:
(2)
(3)
(4)
(5)
(6)
Female (X23)
0.1402
0.363
.1401
0.0196
0.0037
Male (base)
Family size:
Total number in
.1539
household (X24)
Total number
0.7309
2.686
.1401
0.0641
0.0122
squared (X25)
Education of meal planner:
-0.0460
-1.803
Years (X26)
Guest meals:
0.1235
3.533
.1401
0.0173
0.0045
Number of meals (X27)
Meals away from home:
0.0901
2.906
.1401
0.0126
0.0033
Dollars (X28)
-0.0193
-3.472
.1401
-0.0027
-0.0005
174


Table B-8. Summary
statistics for Tobit
analysis of
weekly household expenditures on
canned seafood3
Category
Parameter
Asymptotic
Expected
Expected total
change resulting
Expected change
estimates
a.
X
t-ratio
probability
F(Z.)
from change in X^
SE(EXP)b
BX.
among consuming units
9E(EXP*) F(Z.)
ax
(1)
(2)
(3)
(4)
(5)
(6)
Region:
Northeastern (XI)
0.4819
6.550
.4404
0.2122
0.0704
North Central (X2)
-0.4122
-5.262
.2776
-0.1144
-0.0351
Southern (X3)
-0.3921
-5.174
.2810
-0.1102
-0.0282
Western (base)


.3520


Urbanization:
Central City (X4)
0.4555
6.874
.3859
0.1758
0.0536
Suburban (X5)
0.2882
4.701
.3520
0.1014
0.0293
Nonmetro (base)


.3015


Season:
Spring (X6)
-0.1090
-1.575
.3446
0.0376
-0.0107
Summer (X7)
-0.0537
-0.775
.3557
-0.0191
-0.0056
Fall (X8)
-0.1747
-2.602
.3336
-0.0587
-0.0163
Winter (base)


.3632


186


Table B-7. Descriptive statistics of variables in canned seafood models
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(A)
Proportion
of category
consuming
Mean (percent)3
Region:
Northeastern (XI)
0.248
0.350
0.200
0.447
North Central (X2)
0.241
0.203
0.258
0.267
Southern (X3)
0.342
0.251
0.384
0.233
Western (base)
0.169
0.196
0.158
0.368
Urbanization:
Central City (X4)
0.305
0.322
0.298
0.335
Suburban (X5)
0.355
0.402
0.333
0.359
Nonmetro (base)
0.340
0.276
0.369
0.257
Season:
Spring (X6)
0.238
0.230
0.241
0.306
Summer (X7)
0.232
0.236
0.230
0.322
Fall (X8)
0.268
0.256
0.273
0.303
Winter (base)
0.262
0.278
0.256
0.336
181


Table B-7. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(M
Proportion
of category
consuming
Actual mean
values
Family size:
Total number in household (X24)
2.946
3.300
2.782
Total number squared (X25)
11.457
13.826
10.358

Education of meal planner:
Years (X26)
11.732
12.203
11.514

Guest meals:
Number of meals (X27)
1.141
1.269
1.081

Meals away from home:
Dollars (X28)
11.658
12.564
11.238

Income before taxes:
Thousand dollars (X29)
14.109
15.844
13.305
Thousand dollars squared (X30)
324.590
389.335
294.558

184


Table B-7. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Mean (percent)
Household life cycle:
Young single w/o children (X9)
0.055
0.047
0.059
0.271
Young married w/o children (X10)
0.058
0.055
0.059
0.300
Young single with children (X11)
0.035
0.036
0.034
0.326
Young married with children (X12)
0.151
0.171
0.142
0.359
Middle aged single w/o children (X13)
0.075
0.056
0.085
0.236
Middle aged married w/o children
(X14)
0.115
0.107
0.119
0.295
Middle aged single with children
(X15)
0.056
0.065
0.052
0.368
Middle aged married with children
(X16)
0.261
0.332
0.229
0.403
Elderly single (X17)
0.099
0.057
0.118
0.182
Elderly married (base)
0.095
0.074
0.103
0.247
Race of respondent:
White (X18)
0.852
0.873
0.842
0.325
Other (X19)
0.030
0.037
0.026
0.390
Black (base)
0.118
0.090
0.132
0.242
182


LIST OF TABLES
Table Page
3-1 Descriptive statistics of variables in total
seafood models 52
3-2 Summary statistics for Tobit analysis of weekly
household expenditures on seafood 61
3-3 Summary statistics for Tobit analysis of weekly
household consumption of seafood 67
3-4 Percentage distribution of household head by age
for selected years 85
3-5 Estimated effects of changes in household size on
weekly expenditures and at-home seafood consumption 92
3-6 Percent of meals eaten away from home, by type of
meal and selected household characteristics, spring
1965 and spring 1977 98
3-7 Estimated effects of changes in before tax income
on weekly expenditures and quantities of seafood
consumed 102
3-8 Estimates of at-home seafood expenditure elastici
ties with respect to income 104
3-9 Median family income in constant (1982) dollars for
selected years 106
4-1 Descriptive statistics of data used in seafood
product form models 112
4-2 Signs of estimated parameter associated with vari
ables included in Tobit seafood expenditure models 115
4-3 Estimated weekly expenditure elasticities with
respect to family size for specific seafood product
forms 127
4-4 Estimated weekly expenditure, quantity and quality
elasticities with respect to before tax income for
specific and total seafood product forms 132
vi


Table B-9. Continued
Category
Parameter
estimates
f*i
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X-^
3E(Q)b
9Xi
Expected change
among consuming units
3E(Q) F(Z.)
aXi 1
(1)
Sex of meal planner:
(2)
(3)
(4)
(5)
(6)
Female (X23)
0.2498
3.989
.3557
0.0889
0.0258
Male (base)
Family size:
Total number in
.2776
household (X24)
Total number
0.1692
3.854
.3520
0.0488
0.0141
squared (X25)
Education of meal planner:
-0.0091
-2.201
Years (X26)
Guest meals:
0.0223
3.939
.3520
0.0078
0.0023
Number of meals (X27)
Meals away from home:
0.0259
4.833
.3520
0.0091
0.0026
Dollars (X28)
-0.0023
-2.746
.3520
-0.0008
-0.0002
194


Table 3-3. Summary statistics for Tobit analysis of weekly household quantity consumption of total seafood
Category
Parameter
estimates
ei
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
3E(Q)b
3X.
Expected change
among consuming units
3E(Q*) F(Z )
ax.
(1)
(2)
(3)
(4)
(5)
(6)
Region:
Northeastern (XI)
0.5888
4.382
.5010
0.2950
0.1073
North Central (X2)
-0.5883
-5.197
.3842
-0.2260
-0.0688
Southern (X3)
-0.0060
-0.044
.4414
-0.0026
-0.0009
Western (base)


.4420


Urbanization:
Central City (X4)
0.7393
6.343
.4807
0.3553
0.1256
Suburban (X5)
0.4234
3.895
.4490
0.1901
0.0639
Nonmetro (base)


.4071


Season:
Spring (6)
0.1739
1.417
.4535
0.0789
0.0267
Summer (7)
0.2623
2.132
.4623
0.1213
0.0416
Fall (X8)
0.0215
0.181
.4383
-0.0094
0.0031
Winter (X9)


.4361




144
seafood markets would provide information which could be used to
study the interrelationships between the two markets and help answer
questions which continually arise regarding various policy issues.
A second extension of this study would be a parallel study of
at-home and away-from-home consumption of specific species such as
shrimp, groundfish, etc. An analysis of this nature was omitted from
this study because of the relatively few observations in the data base
used pertaining to any specific species within socioeconomic and demo
graphic classes. However, since the data collected by the Marketing
Research Corporation includes away-from-home seafood consumption, this
data base is probably adequate for an analysis of this type.
With respect to the present study, there are at least two
research options available to improve the understanding of at-home
seafood consumption. First, additional variables and/or a change in
the specification of those variables used in the current analysis may
prove beneficial. For example, religion has often been hypothesized
to influence seafood consumption but was not noted in the data base in
the current study. Similarly, additional interaction terms might be
tested. Estimation and inclusion of a wage rate of the meal planner
similar to that discussed by Prochaska and Schrimper (1973) may avoid
some of the confounding effects associated with family size and educa
tion that were observed in this study.
A second option that may prove beneficial when investigating
at-home seafood consumption relates to the statistical technique used
in the present study. With the Tobit procedure the question of whether
to consume and the level of consumption are considered simultaneously.
In reality, however, these questions may be addressed independently


Table 3-5. Estimated effects of changes in household size on weekly expenditures and at-home
seafood consumption3
<
Component
Number of
people
residing in
household
1
2
3
4
5
Avg.
Weekly expenditures ($):
Total expected change in
consumption 9E(EXP)/9X.£
.201
.206
.207
.205
.199
.206
Expected change attributable
to consuming households
[aEEXP^/aXj F(Z)
.062
.068
.072
.075
.077
.072
Expected probability of
consumption F(Z^)
.397
.436
.472
.507
.538
.476
Change in expected probability
of consumption 9F(Z)/9X.£
.040
.038
.035
.033
.030
.035
Weekly quantity consumed (lbs):
Total expected change in
consumption 3E(Q)/9X
.149
.155
.158
.159
.158
.158
Expected change attributable
to consuming households
[9E(Q*)/9X] F(Z)
.044
.049
.053
.056
.058
.053
Expected probability of
consumption F(Z^)
.369
.407
.443
.478
.510
.447
Change in expected probability
of consumption 9F(Z^)/9X
.038
.037
.036
.034
.031
.035
kO
ho
aEvaluated at an income level of $15,080 thousand.


109
overriding factor that will probably determine the future status of
demand for seafood for at-home consumption is that of the change in the
market for food consumed away from home. Given the propensity to
consume seafood in the away-from-home market, increases in away-from-
home consumption can be expected to have a rather strong negative
influence on at-home consumption of seafood. Evidence suggests that
away-from-home consumption will continue to take a larger portion of
the consumers income in future years. In fact, many of those factors
found in this study to increase at-home consumption of seafood also
increase total away-from-home consumption of food. Redman (1980)
investigated those factors determining expenditure on meals away from
home and concluded that increased family income and a college educa
tion of the woman head of household, both of which were estimated to
increase at-home consumption of seafood also contribute to increased
expenditures on meals away from home. Additionally, Redman concluded
that a decline in family size, which the U.S. population is currently
undergoing, also contributes to increased expenditures on meals away
from home. Decreases in expenditure on meals away from home were
found among Black households and with increased age of the woman. As
concluded in the current study, at-home seafood consumption was rela
tively high among Black households and elderly couples, though consump
tion was much lower among elderly individuals many of which are women,
certis paribus.
As expenditures on meals consumed away from home increases,
seafood consumed away from home as a percentage of total consumption
will increase proportionately. Thus, the cyclical nature associated
with the demand for seafood as discussed in Chapter I will likely


Billion Pounds (whole weight)
Figure 1-2. U.S. supply of edible fishery products (round weight), 1960-83
SOURCE: United States Department of Commerce, NOAA, NMFS (various issues).


36
planner are likely to be associated with an increased awareness of the
nutritional value associated with consumption of seafood products.
This increased nutritional awareness and resultant increased consump
tion of seafood products are likely to offset the expected decline in
consumption associated with the increase in the opportunity cost of
time resulting from additional education. Similarly, though increases
in family size are expected to result in an increase in the opportunity
cost of time of the meal planner and hence a potential decline in
at-home consumption of seafood, increases in family size by definition
necessitates increased consumption in total. Hence the expected
decline in household consumption of seafood resulting from an increase
in the opportunity cost of time of the meal planner associated with an
increase in family size may be offset by increased consumption necessi
tated by an increase in family size.
The number of guest meals served from home food supplies (Z12),
found to be a significant factor by Capps (1982) in explaining expendi
ture on seafood consumed at home, was introduced into the analysis to
account for the expected increase in consumption of total seafood and
especially those seafood product forms most likely to be served when
entertaining guests. Since the less processed seafood product forms
are generally associated with a higher quality product and hence viewed
as more preferred items, increased weekly consumption of these product
forms is expected to be positively related with increases in the
number of guest meals.
Although Perry (1981) concluded that the money value of away-from-
home consumption was generally unimportant in explaining seafood
expenditures for at-home consumption, a similar variable was included


60
consumption to increase significantly with changes in the value of
these variables.
Regression Estimates of Total Seafood Analyses
Notwithstanding the general usefulness of the descriptive statis
tics just presented, there are several inherent weaknesses associated
with these types of statistics. The foremost weakness associated with
these types of statistics is that they do not control for confounded
effects among different variables. Thus, one cannot separate the
effect of one exogenous variable from that of another when examining
changes in the dependent variable. For example, the relatively low
probability of seafood consumption among households without children
when compared to those with children may be due to some factor such as
larger expenditures on meals consumed away from home among the former
group of households. The Tobit regression parameters presented in this
section can be considered as partial effects in that the confounded
effects among exogenous variables have been controlled for.
J
Results of the Tobit analysis relating to total at-home seafood
2
consumption are presented in Tables 3-2 and 3-3. The first column
gives a listing of the variables used in the analysis. The second
column in each table gives the Tobit parameter estimates associated
with each of the exogenous variables. The asymptotic t-values asso
ciated with the parameter estimates are presented in the third column.
The relatively large sample size employed in this study should assure
that the asymptotic t-values are representative of the true values.
2
The Tobit model used for this analysis was developed by the Rand
Corporation and is referred to as LIMDEP. Documentation of the model
is given by Phelps (1972).


49
, 9Xi E(y*). 3F(Z)
ni E(y*) ^ 9X )+ F(Z) ( 9Xi *
(2.23)
The total elasticity is comprised of two components. The first compo
nent measure the conditional elasticity associated with the nonlimit
observations. The second term measures the elasticity of the prob
ability of participation associated with a change in X^.
Data Source and Considerations
The 1977-78 Nationwide Food Consumption Survey (NFCS) provides the
data used in the analysis. This is the most recent of the household
food consumption surveys periodically conducted under the auspices of
the United States Department of Agriculture. The survey encompassed
approximately 15,000 households throughout the 48 continguous states
and contained detailed information on characteristics for each
household. In addition, the survey contained detailed information
pertaining to expenditures on and the corresponding quantities of a
continum of foods consumed at home (measured at the level at which the
foods came into the kitchen) by each of the households surveyed. The
survey was conducted over a one year period (April 1977 through March
1978) and was stratified according to a variety of factors including
season, region, and urbanization in an attempt to have the sample
represent the universe of households in the continental United States
as accurately as possible. Though information on 14,930 households
was provided on the original NFCS data tapes distributed through the
Department of Commerce's National Technical Information Service, only
10,689 observations of the original 14,930 were retained for the
current analysis. Of the 4,241 deleted observations, approximately 92


77
coastal fisheries "... are exploited by small, ill-equipped or part-
time fishermen, delivering their product to dockside warehouses for
fresh distribution to communities within a 50-mile radius." O'Rourke
further claims that the New England fisheries deliver much of their
specialty catches in this manner. Because of this factor "... large
areas of the continental U.S. have below average consumption of many
fish and shellfish species." The second segment which may account
for two-thirds of the U.S. consumption of seafood is comprised of
". . large, highly capitalized canners or prepackers selling
standardized, breaded, and heavily promoted products through nationwide
retail chains or institutional outlets." The relative unavailability
and expense of certain fresher seafood products in the North Central
region probably explains to some extent the relatively low estimates
of at-home seafood expenditures and consumption in this region compared
to the other regions of the United States. Furthermore, one would
expect regional differences to be relatively large for fresh seafood
products and somewhat less for the more processed seafood products sold
either frozen or canned. The validity of this hypothesis is examined
in the next chapter when results pertaining to the seafood product
forms are analyzed.
Given the differences and probable causes for these differences as
relating to at-home consumption of seafood, what are their implications
to the seafood industry and its support group? Most obvious and
probably the most important from a policy standpoint lies in the
jrC
realization that the North Central region provides a relatively large
and untapped source by which to increase national at-home seafood
consumption. To this extent, it may be beneficial to the seafood
/


Table 3-2. Continued
Category
Parameter Asymptotic
estimates t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in
3E(EXP)b
SXi
Expected change
among consuming units
9E(EXP'::~) ?{Z)
3Xi
(1) (2) (3)
Race of respondent:
White (X18) -1.2718 -4.635
Other (X19) -0.3428 -0.988
Black (base)
Receives food stamps:
Yes (X20) 0.2103 0.870
No (base)
Caught fish for own use:
Yes (X21) 1.3842 10.723
No (base)
Employment of meal planner:
-0.964
(4) (5) (6)
.4504 -0.8553 -0.2883
.6051 -0.2074 -0.0883
.6061
.4915 0.1034 0.0371
.4745
.5580 0.7724 0.3084
.4464
.4705 -0.0568
.4803
Yes (X22)
No (base)
-0.1207
-0.0199


Table B-6. Continued
Category
Parameter Asymptotic
estimates t-ratio
3.
i
Expected
probability
F(Z.)
Expected total
change resulting
from change in X-¡
3E(Q)b
3Xi
Expected change
among consuming units
MQ- f(z -)
9Xi
(1)
(2)
(3)
(A)
(5)
(6)
Race of respondent:
White (X18)
0.3523
1.139
.1401
-0.0290
-0.0054
Other (X19)
-0.7204
-1.752
.1093
-0.0787
-0.0146
Black (base)


.1515


Receives food stamps:
Yes (X20)
0.0442
0.1670
.1401
0.0062
0.0011
No (base)


.1379

Caught fish for own use:
Yes (X21)
0.7343
5.757
.1711
0.1256
0.0232
No (base)


.1259

Employment of meal planner:
Yes (X22)
-0.0468
-0.370
.1357
-0.0064
-0.0012
No (base)


.1379


178


Table 3-1. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Interaction terms:
Income and race (X31)
12.745
Actual mean
13.394
values
12.075
Income and family size
(X32)
46.628
52.460
40.601

Weekly expenditures and
EXP
quantity:
1.487
2.927
0.000
Q
0.971
1.911
0.000

Number of households:
10,689
5,430
5,259

Percent of households:
100
50.8
49.2

The data provided in this table associated with the binary variables (X1-X23) should be interpreted as
representing proportions rather than percentages. To obtain percentages, multiply data by 100.


87
parents were present in the household. Thus, households in this group
should have a higher demand for seafood products for which little
preparation time is required relative to households in which both
parents are present.
Race
Black households (base) had significantly higher at-home seafood
expenditures and quantities consumed than did White households (X18).
Similarly, Black households were estimated to consume greater quanti
ties of seafood than households of "Other" ethnic origins (X19) but
their weekly expenditures were not significantly different given the
insignificant t-value in column 3 of Table 3-2. Total weekly at-home
seafood consumption by a typical White household was estimated to be
0.78 pounds less than that of a similar Black household while expendi
tures by that same White household were estimated to be $0.86 less than
that of a similar Black household. Among households of some "Other"
ethnic origin, weekly at-home consumption of seafood was estimated to
be 0.42 pounds less than that of a similar Black household while
expenditures by this group were only $0.21 less than that of a similar
Black household, certis paribus.
The estimated probability of consuming seafood at home was
substantially lower among White households than among either Black
households or households of "Other" ethnic origins. Among White house
holds, the estimated probability of consuming seafood at home was 0.42,
compared to 0.61 among Black households, and 0.53 among households of
"Other" ethnic origins (column 4, Table 3-3). The large differences
associated with the probability of consumption among the different


Table B-ll. Continued
Category
Parameter
estimates
a.
i
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in Xi
9E(EXP)b
9X
Expected change
among consuming units
3ECEXP*) F(Z .)
9X 1
(1)
(2)
(3)
(4)
(5)
(6)
Household life cycle:
Young single w/o
children (X9)
-0.5339
-1.738
.4562
-0.2436
-0.0780
Young married w/o
children (X10)
-0.9250
-3.493
.4207
-0.3891
-0.1245
Young single with
children (X11)
-1.2980
-3.946
.3859
-0.5009
-0.1603
Young married with
children (X12)
-1.1509
-4.994
.4013
-0.4619
-0.1478
Middle aged single
w/o children (X13)
-0.4296
-1.585
.4681
-0.2011
-0.0644
Middle aged married
w/o children (X14)
-0.0933
-0.439
.4960
-0.0463
-0.0148
Middle aged single
with children (X15)
-0.2522
-0.942
.4840
-0.1221
-0.0391
Middle aged married
with children (X16)
-0.6745
-2.962
.4443
-0.2997
-0.0959
Elderly single (X17)
-0.4862
-1.941
.4602
-0.2237
-0.0718
Elderly married (base)


.5080

-
202


123
be relatively low when compared to other households, ceteris paribus.
Though expenditures on finfish products by households with young
children tended to be relatively low, a similar trend was not noticed
with shellfish expenditures. It is likely that the specification of
the life cycle category was overly refined to account for differences
in shellfish consumption among households in different life cycles
given the relatively few positive observations for this product form.
Race
Substantial differences in preferences for the alternative seafood
product forms exist among households of different races. Black house
holds (base) made significantly higher expenditures on fresh seafood
consumed at home than either White households (X18) or households of
other ethnic origins (X19), ceteris paribus. Alternatively, expendi
tures on canned seafood were significantly higher among White
households and households of "Other" ethnic origins than among Black
households. The difference in expenditures on frozen seafood consumed
at home was not statistically significant among White, Black, and
"Other" households.
Statistically significant differences in expenditure patterns
were estimated among households of different races in the purchasing
of finfish products but not shellfish products. Estimated expenditures
on finfish products consumed at home were highest among Black house
holds and lowest among White households, ceteris paribus.


31
In addition to the expenditure elasticities reported by Haidacher
et al., the authors also report the quantity elasticities with respect
to income. The quantity elasticities with respect to income were
positive for shellfish and negative for finfish and total seafood. The
negative quantity elasticity with respect to income for total seafood
in conjunction with a positive expenditure elasticity implies a posi
tive quality elasticity for total seafood which equalled 0.20. This
consists of a high estimate of the quality elasticity associated with
shellfish consumption (0.59) and a relatively low estimate of the
quality elasticity associated with finfish consumption (0.10).
Summarizing the research to date, evidence suggests inelastic
income expenditure and quantity elasticities for total seafood and
specific product forms. However, none of the studies conducted to
date has made complete use of all data and/or available statistical
options. An extension of the work provided by the authors discussed in
this chapter is the basis of the next two chapters. The remainder of
this chapter lays the groundwork for the models to be estimated.
Conceptual Model Development
The first task associated with specifying a cross-sectional con
sumption model involves that of defining the set of arguments compris
ing the column vector of socioeconomic characteristics, Zj, given in
equation (2.10). The concept of consumer demand in conjunction with
the seafood expenditure/quantity consumption studies discussed in the
previous section were of assistance in meeting this objective. The
respective expenditure and quantity equations were specified as a


Table B-3. Continued
Category
Parameter
estimates
h
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X.
8E(Q)b
aXi
Expected change
among consuming units
3E(Q*) F(Z.)
3X
(1)
(2)
(3)
(A)
(5)
(6)
Race of respondent:
White (X18)
-4.7697
-9.723
.1635
-0.8916
-0.1790
Other (X19)
-2.5689
-4.031
.2514
-0.6458
-0.1565
Black (base)


.3745


Receives food stamps:
Yes (X20)
0.3883
0.848
.2033
0.0789
0.0174
No (base)


.1894

Caught fish for own use:
Yes (X21)
3.0059
12.022
.2776
0.8344
0.1387
No (base)


.1587

Employment of meal planner:
Yes (X22)
-0.4019
-1.605
.1814
-0.0729
-0.0153
No (base)


.1977


163


Table B-3. Continued
Category
Parameter
estimates

Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in
8E(Q)b
9*i
Expected change
among consuming units
3E(Q) F(Zi)
ax
(1)
Sex of meal planner:
(2)
(3)
(A)
(5)
(6)
Female (X23)
0.7761
1.663
.1922
0.1492
0.0322
Male (base)
Family size:
Total number in
.1635
household (X24)
Total number
squared (X25)
Education of meal planner:
-0.2702
0.0496
-0.831
1.674
.1894
-0.0090
-0.0019
Years (X26)
Guest meals:
0.0505
1.214
.1894
0.0096
0.0020
Number of meals (X27)
Meals away from home:
0.2266
5.835
.1894
0.0429
0.0091
Dollars (X28)
-0.0182
-2.762
.1894
-0.0034
-0.0007
164


Table B-13. Descriptive statistics of variables in shellfish seafood models
(1)
(2)
(3)
(A)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Mean (percent)
Region:
Northeastern (XI)
0.248
0.308
0.243
0.095
North Central (X2)
0.241
0.160
0.247
0.051
Southern (X3)
0.342
0.314
0.344
0.070
Western (base)
0.169
0.218
0.166
0.098
Urbanization:
Central City (X4)
0.305
0.345
0.302
0.086
Suburban (X5)
0.355
0.385
0.352
0.083
Nonmetro (base)
0.340
0.270
0.346
0.061
Season:
Spring (X6)
0.238
0.214
0.240
0.069
Summer (X7)
0.232
0.218
0.233
0.072
Fall (X8)
0.268
0.298
0.265
0.085
Winter (base)
0.262
0.270
0.262
0.079
211


23
Leser, 1963), no single functional form has emerged as clearly superior
under all conditions. There are, however, certain criteria that should
be considered when selecting and judging the appropriateness of a given
functional form (Brown and Deaton, 1972; Tomek, 1977; Hassan et al.,
1977). First, the functional form should allow for the possibility
that the commodity will not be consumed given an income below some
initial level. Second, the functional form should allow for a declin
ing marginal propensity to consume with increased income. Third, the
functional form should allow for a satiety level which provides an
upper bound on quantity consumed. Finally, simplicity and convenience
of estimation need to be considered. Though these criteria are valid
when considering quantity consumption functions, they appear to be
overly restrictive when considering expenditure consumption functions.
Given the additional aspect of demand for quality and services, there
is no reason to necessarily expect the marginal propensity of expendi
tures associated with some commodities to decline with increased
income. Similarly, a satiety level associated with expenditures on
some commodities is not necessarily expected, a priori.
As a commodity becomes more narrowly defined in an analysis, the
percentage of individual households not consuming that commodity
naturally increases. Therefore, a decision whether to include in the
analysis those households not consuming the commodity being analyzed
needs to be made. Currie et al. (1972) provide a good discussion
concerning under what conditions it is logical to exclude (include)
nonconsuming households from (in) the analysis. Basically, the issue
reduces to the following premise: if the consuming and nonconsuming
households can be considered as having identical behavioral patterns,


Table B-6. Continued
Category
Parameter
estimates
Asymptotic
t-ratio
Expected
probability
F(Zi)
Expected total
change resulting
from change in X.
3E(Q)b
3X-:
Expected change
among consuming units
3E(Q*) .F(Z.)
3Xi
(1)
Sex of meal planner:
(2)
(3)
(4)
(5)
(6)
Female (X23)
0.0342
0.135
.1379
0.0047
0.0087
Male (base)
Family size:
Total number in
.1357
household (X24)
Total number
0.5570
3.103
.1379
0.0450
0.0083
squared (X25)
Education of meal planner:
-0.0366
-2.165
Years (X26)
Guest meals:
0.0763
3.328
.1379
0.0105
0.0019
Number of meals (X27)
Meals away from home:
0.0455
2.207
.1379
0.0063
0.0012
Dollars (X28)
-0.0139
-3.738
.1379
-0.0019
-0.0004
179


CHAPTER III
TOTAL SEAFOOD ANALYSIS
The discussion of the results associated with the total seafood
models is given in two sections. First, descriptive statistics
comparing/contrasting consumers and nonconsumers of seafood as
established by the survey data are presented and briefly discussed.
In the second section, the results of the Tobit analysis associated
with the total seafood models are presented and discussed.
Descriptive Statistics Associated with
Total Seafood Analysis
Descriptive statistics of the 10,689 households included in the
analysis are presented in Table 3-1. Though the information presented
in this table is of a descriptive nature without any attempt to
separate partial effects, the information does provide an overall
comparison of those households which consumed seafood during the one
week survey period versus those households which did not.
The statistics provided in Table 3-1 are assigned to one of four
categories. The first category, labelled total sample, gives the mean
values of all variables used in the analysis. For example, as
indicated in the table, 24.8 percent of the households in the analysis
resided in the Northeast region (XI). The second category, labelled
nonlimit observations, provides the mean values of all variables for
those households consuming seafood at home. For example, of those
51


Table 3-1. Continued
(1)
(2)
(3)
(4)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Mean (percent)3
Household life cycle:
Young single w/o children (X9)
0.055
0.045
0.065
0.416
Young married w/o children (X10)
0.058
0.054
0.061
0.473
Young single with children (Xll)
0.035
0.035
0.034
0.508
Young married with children (X12)
0.151
0.158
0.143
0.532
Middle aged single w/o children (X13)
0.075
0.063
0.088
0.427
Middle aged married w/o children (X14)
0.115
0.119
0.111
0.526
Middle aged single with children (X15)
0.056
0.062
0.050
0.562
Middle aged married with children (X16)
0.261
0.299
0.223
0.582
Elderly single (X17)
0.099
0.073
0.126
0.375
Elderly married (base)
0.095
0.092
0.099
0.492
Race of respondent:
White (X18)
0.852
0.831
0.874
0.495
Other (X19)
0.030
0.035
0.024
0.592
Black (base)
0.118
0.134
0.102
0.577


Page
III TOTAL SEAFOOD ANALYSIS 51
Descriptive Statistics Associated with Total
Seafood Analysis 51
Regression Estimates of Total Seafood Analyses 60
Region 75
Urbanization 78
Season 80
Household Life Cycle 81
Race 87
Household Receives Food Stamps 89
Household Caught Fish for Own Use 90
Employment of the Meal Planner 90
Sex of the Meal Planner 90
Household Size 91
Education of Meal Planner 95
Number of Guest Meals 96
Expenditures on Meal Consumed Away from Home 97
Income before Taxes 100
Outlook for Increasing At-Home Demand for Seafood
and Implications 107
IV SPECIFIC SEAFOOD PRODUCT FORM ANALYSIS Ill
Introduction Ill
Comparison of Consumption Parameters 114
Region 114
Urbanization 120
Season 121
Household Life Cycle 121
Race 123
Food Stamps 124
Fish Caught for Own Use 124
Employment of the Meal Planner 124
Sex of the Meal Planner 125
Family Size 125
Education of Meal Planner 129
Number of Guest Meals 129
Expenditures on Meals Away from Home 130
Income before Taxes 130
Other Seafood Expenditures 135
Outlook for Increasing At-Home Demand for Specific
Product Forms and Implications 136
iv


73
In column 4 of each table, the vlue of the cumulative normal
distribution function associated with each variable is presented.
The value of this function varies with each of the discrete variables
due to variations in E(X3|X). For discrete variables in the models,
the values provided for the cumulative normal distribution function are
interpreted as the expected probability of observing a positive level
of the dependent variable given the occurrence of X^, holding all non-
mutually exclusive variables at their mean levels. Mutually exclusive
variables are set equal to zero. For example, to determine the
expected probability of seafood consumption by average Northeastern
households, the value of XI is set equal to one, the values for X2 and
X3 are set equal to zero, and the values for all remaining variables
are set at their respective means. For continuous variables, the value
provided for the cumulative normal distribution function represents the
probability of observing a positive level of the dependent variable
given the mean values for all exogenous variables. Of course, the
value of the cumulative normal distribution function varies with
changes in the level of exogenous variables^
Multiplication of the appropriate parameter estimates, given by
those values in column 2 by their respective expected probabilities of
occurrence (column 4) provides the unconditional or total expected
change in the dependent variable due to a change in X^. These esti
mates are given in column 5 of Tables 3-2 and 3-3. The unconditional
or total effect of a change in expenditures or quantity consumed with
respect to a change in the independent variable X^ can be decomposed
into two parts. The first part represents the change in the value of
expenditures or quantity consumed among existing consumers weighted by


Table B-12. Continued
Category
Parameter
estimates
h
Asymptotic
t-ratio
Expected
probability
F(Zi)
Expected total
change resulting
from change in X^
3E(Q)b
8Xi
Expected change
among consuming units
3E(Q*) .F(Z .)
3Xi
(1)
(2)
(3)
(4)
(5)
(6)
Race of respondent:
White (X18)
-1.3642
-6.925
.4013
-0.6690
-0.2007
Other (X19)
-0.9009
-3.597
.4920
-0.4432
-0.1374
Black (base)


.5910


Receives food stamps:
Yes (X20)
0.1591
0.916
.4404
0.0701
0.0224
No (base)


.4257


Caught fish for own use:
Yes (X21)
1.1117
11.885
.5199
0.5780
0.1965
No (base)


.3966


Employment of meal planner:
Yes (X22)
-0.0119
-0.131
.4247
-0.0051
-0.0016
No (base)


.4247


208


Table B-13. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Actual mean
values
Family size:
Total number in household (X24)
2.946
3.043
2.939
Total number squared (X25)
11.457
11.777
11.431

Education of meal planner:
Years (X26)
11.732
12.763
11.647

Guest meals:
Number of meals (X27)
1.141
1.793
1.087

Meals away from home:
Dollars (X28)
11.658
15.093
11.375

Income before taxes:
Thousand dollars (X29)
14.109
18.144
13.776
Thousand dollars squared (X30)
324.590
489.020
311.017

214


10
If as in the first case, at-home consumption of seafood is increased at
the expense of the away-from-home market, little change in the total
demand for seafood may be experienced. If, on the other hand, at-home
demand for seafood is increased independent of the away-from-home
market, total demand for seafood by definition will be increased.
This could put pressure on many of the already heavily fished stocks
(as of 1974 about 62 percent of the economically important fisheries in
the United States were fully utilized or overfished (Eckert, 1979)) and
lead to an even greater demand for imported seafood.
In order to effectively increase at-home demand for seafood, a
thorough understanding of those factors hypothesized to determine
at-home consumption of seafood is required. Only then can the effects
of the at-home seafood market on the away-from-home market and
ultimately on the domestic fishing industry and on import demand for
seafood be fully comprehended. Thus, the overall objective of this
study is to examine and quantify those factors hypothesized to
determine at-home consumption of total seafood and specific product
forms (fresh, frozen, canned, shellfish, and finfish). In relation to
this objective, this study is designed to provide information concern
ing seafood marketing implications based on empirical findings and
historical trends related to factors which determine seafood
consumption. This information is essential in evaluating fishery
market legislation, management alternatives, long-term trends, and
promotion and marketing programs.
The format of this study proceeds as follows. A review of
literature, the models to be estimated, and a discussion of statistical
considerations and data used in analyzing the models are provided in


Table 3-1. Descriptive statistics of variables in total seafood models
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Mean (percent)
Region:
Northeastern (XI)
0.248
0.299
0.195
0.612
North Central (X2)
0.241
0.213
0.270
0.449
Southern (X3)
0.342
0.305
0.379
0.453
Western (base)
0.169
0.183
0.156
0.550
Urbanization:
Central City (X4)
0.305
0.330
0.280
0.550
Suburban (X5)
0.355
0.374
0.335
0.535
Nonmetro (base)
0.340
0.296
0.385
0.442
Season:
Spring (X6)
0.238
0.233
0.243
0.497
Summer (X7)
0.232
0.234
0.229
0.512
Fall (X8)
0.268
0.259
0.278
0.491
Winter (base)
0.262
0.274
0.250
0.531


Table B-4. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(A)
Proportion
of category
consuming
3
Mean (percent)
Receives food stamps:
Yes (X20)
0.074
0.054
0.078
0.095
No (base)
0.926
0.946
0.922
0.133
Caught fish for own use:
Yes (X21)
0.234
0.303
0.224
0.168
No (base)
0.766
0.697
0.776
0.118
Employment of meal planner:
Yes (X22)
0.465
0.460
0.466
0.128
No (base)
0.535
0.540
0.534
0.131
Sex of meal planner:
Female (X23)
0.908
0.938
0.903
0.134
Male (base)
0.092
0.062
0.097
0.087
168


20
With respect to household composition, the authors generally
attempted to standardize the household in order to show the impact that
different household members, varying in terms of sex and age, had on
the consumption behavior of the household. The one exception to this
rule is provided by Murphy and Staples, who attempted to explain
differences in consumption among different households by examining the
different stages of a household's growth and maturity.
With respect to the opportunity cost of time, it has long been
recognized that a commodity, when consumed at home, is not in the same
form as when purchased. Rather, value is added to the commodity after
it is purchased to transform it into some "new" commodity suitable for
consumption. Mincer (1963) cognizant of this fact demonstrated that
estimated income elasticities for a variety of commodities will tend to
be biased if the opportunity cost of time is omitted. Prochaska and
Schrimper (1973) analyzed away-from-home consumption with respect to
the opportunity cost of time of the meal preparer. Redman (1980)
analyzed the impact of women's time on away-from-home consumption and
for prepared meals based upon the concept of the opportunity cost of
time as presented by Gronau (1977). Though conceptually similar, the
studies of Prochaska and Schrimper and Redman differ somewhat in the
treatment of the opportunity cost of time. Prochaska and Schrimper
estimated a wage rate for the meal preparer based upon a set of
arguments (education and age) and then included the estimated wage rate
as an argument in the demand equation for away-from-home consumption.
Redman, on the other hand, introduced those arguments hypothesized to
affect women's opportunity cost of time (education of homemaker, age of


Table B-10. Descriptive statistics of variables in finfish seafood models
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Mean (percent)
Region:
Northeastern (XI)
0.248
0.303
0.198
0.579
North Central (X2)
0.241
0.215
0.264
0.423
Southern (X3)
0.342
0.301
0.378
0.442
Western (base)
0.169
0.181
0.160
0.491
Urbanization:
Central City (X4)
0.305
0.330
0.283
0.513
Suburban (X5)
0.355
0.372
0.339
0.496
Nonmetro (base)
0.340
0.298
0.378
0.415
Season:
Spring (X6)
0.238
0.235
0.240
0.468
Summer (X7)
0.232
0.234
0.230
0.478
Fall (X8)
0.268
0.256
0.279
0.453
Winter (base)
0.262
0.275
0.251
0.497
196


Table B-14. Continued
Category
Parameter
estimates
a .
i
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
9E(EXP)b
ax.
i
Expected change
among consuming units
9E(EXP*) F(Z.)
ax.
i
(1)
(2)
(3)
(4)
(5)
(6)
Sex of meal planner:
Female (X23)
Male (base)
-0.1093
-0.152
.0934
.0968
0.0102
-0.0017
Family size:
Total number in
household (X24)
Total number
squared (X25)
-1.5323
0.0966
-2.865
2.017
.0934
-0.0770
-0.0128
Education of meal
planner:
Years (X26)
0.3520
5.079
.0934
0.0329
0.0054
Guest meals:
Number of meals
(X27)
0.3398
6.382
.0934
0.0317
0.0053
Meals away from home:
Dollars (X28)
0.0033
0.379
.0934
0.0003
0.00005
219


14
Since the solution to (2.3) depends only on prices, income, and
the utility function, (2.3) can be solved yielding n + 1 equations, one
for each of the q.'s and one for X in terms of p and m.
i
q = q (p,m) (i = 1, 2, ..., n) (2.4a)
X = X(p>m) (2.4b)
The expressions in (2.4a) are referred to as demand functions, with the
demand for each of the n goods being expressed in terms of its own
price, prices of all substitutes, and income.
Price and Income Considerations
Economic theory suggests that demand functions satisfy certain
restrictions. One restriction is derived from the "fundamental equa
tion of value" theory which decomposes the effect of a price change
into the substitution and income effects, i.e.,
3
3q
(r-M q
op u=const j
3 m
(i,j = 1, 2, ..., n) (2.5)
A good is said to be normal if it has a downward sloping demand curve
or equivalently, a negative own price elasticity. Otherwise it is a
Giffen good which implies an upward sloping demand curve. The first
restriction which follows from the assumption of a strictly convex
indifference map implies that the own-price substitution effect is
1
negative, i.e.,
3qi
(^)
dp. u=const
i
< 0
(2.6)
See Hicks (1957) for a proof of this and subsequently discussed
restrictions.


Table B-8. Continued
Category
Parameter
estimates
a.
i
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in Xi
9E(EXP)b
8Xi
Expected change
among consuming units
3E(EXP*) F(Z1 )
axi
(1)
(2)
(3)
(4)
(5)
(6)
Sex of meal planner:
Female (X23)
Male (base)
0.4326
3.922
.3520
.2776
0.1523
0.0440
Family size:
Total number in
household (X24)
Total number
squared (X25)
0.3004
-0.0184
3.864
-2.512
.3483
0.0822
0.0236
Education of meal
planner:
Years (X26)
0.0408
4.090
.3483
0.0142
0.0041
Guest meals:
Number of meals
(X27 )
0.0446
4.722
.3483
0.0155
0.0045
Meals away from home:
Dollars (X28)
-0.0034
-2.305
.3483
0.0012
0.0045
189


Table B-ll.
Summary statistics for Tobit analysis of weekly household expenditures on finfish seafood
a
Category
Parameter
estimates
i
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
8E(EXP)b
3Xi
Expected change
among consuming units
BE (EXP*) F(Z.j)
3Xi
(1)
(2)
(3)
(A)
(5)
(6)
Region:
Northeastern (XI)
0.6735
4.628
.5239
0.3528
0.1329
North Central (X2)
-0.7195
-4.741
.3974
-0.2859
-0.1000
Southern (X3)
-0.5607
-3.824
.4090
-0.2293
-0.0725
Western (base)


.4641


Urbanization:
Central City (X4)
0.8832
6.979
.4880
0.4310
0.1537
Suburban (X5)
0.4787
4.057
.4522
0.2165
0.0731
Nonmetro (base)


.4090


Season:
Spring (X6)
0.0974
0.732
.4602
0.0448
0.0143
Summer (X7)
0.1032
0.774
.4602
0.0475
0.0152
Fall (X8)
-0.2455
-1.898
.4286
-0.1052
-0.0337
Winter (base)


.4522


201


151
Table B-l. Descriptive statistics of variables in fresh seafood models
(1)
(2)
(3)
(A)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Mean (percent)3
Region:
Northeastern (XI)
0.248
0.303
0.237
0.203
North Central (X2)
0.241
0.156
0.258
0.108
Southern (X3)
0.342
0.366
0.337
0.178
Western (base)
0.169
0.175
0.168
0.172
Urbanization:
Central City (X4)
0.305
0.397
0.287
0.217
Suburban (X5)
0.355
0.333
0.359
0.156
Nonmetro (base)
0.340
0.270
0.354
0.132
Season:
Spring (X6)
0.238
0.235
0.238
0.164
Summer (X7)
0.232
0.262
0.226
0.188
Fall (X8)
0.268
0.254
0.271
0.158
Winter (base)
0.262
0.249
0.265
0.158


126
products, and finfish products with increases in family size. On the
other hand, a negative linear term combined with a positive quadratic
term was estimated for expenditures on both fresh seafood products and
shellfish products. These results suggested that expenditures on fresh
seafood products and shellfish products consumed at home declined, at
least within initial ranges, with increases in family size. Given the
increase in the opportunity cost of time of the meal planner associated
with increases in family size, these results are not totally
unexpected. Convenience in seafood products is most often associated
with certain frozen products, such as fish sticks, and canned products.
Furthermore, the majority of these products are made from finfish
rather than shellfish. As family size increaes, it is logical to
assume that the meal planner becomes more dependent on these conve
nience products.
The estimated expenditure elasticities with respect to family
size for the various seafood products are presented in Table 4-3.
Expenditures on at-home consumption of frozen seafood products, canned
seafood products, finfish seafood products, and total seafood products
have positive elasticities with respect to family size while expendi
tures on at-home consumption of fresh seafood products and shellfish
seafood products exhibit negative elasticities with respect to family
size. Furthermore, the total elasticities (comprised of the elasticity
of participation and the conditional elasticity) ranged from a low of
-0.73 associated with that of expenditures on shellfish to a high of
0.59 associated with that of expenditures on canned seafood products.
The above discussion and analysis suggest convenience in terms of
purchasing, preparation, and consumption of the various seafood product


Table 3-3. Continued
Category
Parameter
estimates
*i
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X.-
3E(Q)b 1
aXi
Expected change
among consuming units
3E(Q*) F(Z)
ax
(1)
(2)
(3)
(A)
(5)
(6)
Race of respondent:
White (X18)
-1.3990
-6.344
.4181
-0.7838
-0.2516
Other (X19)
-0.7943
-2.839
.5263
-0.4180
-0.1579
Black (base)


.6051


Receives food stamps:
Yes (X20)
0.2263
1.164
.4677
0.1058
0.0366
No (base)


.4457


Caught fish for own use:
Yes (X21)
1.2961
12.451
.5426
0.7033
0.2724
No (base)


.4129


Employment of meal planner:
Yes (X22)
-0.0252
-0.249
.4455
-0.0112
-0.0037
No (base)


.4480




Table B-12. Continued
Category
Parameter
estimates
6i
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X.
3E(Q)b
3Xi
Expected change
among consuming units
3E(Q*) .F(Z.)
3Xi
(1)
(2)
(3)
(4)
(5)
(6)
Household life cycle:
Young single w/o
children (X9)
-0.4494
-1.765
.4247
-0.1909
-0.0608
Young married w/o
children (X10)
-0.7208
-3.284
.3974
-0.2864
-0.0916
Young single with
children (X11)
-1.0293
-3.788
.3632
-0.3738
-0.1234
Young married with
children (X12)
-0.9062
-4.752
.3745
-0.3394
-0.1120
Middle aged single
w/o children (X13)
-0.3346
-1.491
.4404
-0.1474
-0.0457
Middle aged married
w/o children (X14)
-0.0200
-0.114
.4761
-0.0095
-0.0029
Middle aged single
with children (X15)
-0.1468
-0.663
.4602
-0.0676
-0.0210
Middle aged married
with children (X16)
-0.5033
-2.670
.4207
-0.2117
-0.0656
Elderly single (X17)
-0.3777
-1.823
.4325
-0.1634
-0.0522
Elderly married (base)


.4761


207


Table B-14. Continued
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
3E(EXP)b
3X
Expected change
among consuming units
3E(EXP*) F(Z.)
3X
(1)
(2)
(3)
(4)
(5)
(6)
Household life cycle:
Young single w/o
children (X9)
-1.1032
-1.005
.0559
-0.0617
-0.0087
Young married w/o
children (X10)
1.3229
1.446
.0934
0.1236
0.0205
Young single with
children (X11)
-0.1944
-0.151
.0681
-0.0132
-0.0020
Young married with
children (X12)
1.5329
1.781
.0985
0.1510
0.0253
Middle aged single
w/o children (X13)
-1.1672
-1.143
.0559
-0.0652
-0.0092
Middle aged married
w/o children (X14)
1.3829
1.781
.0951
0.1315
0.0220
Middle aged single
with children (X15)
1.6982
1.703
.1003
0.1703
0.0292
Middle aged married
with children (X16)
2.7643
3.285
.1230
0.3400
0.0613
Elderly single (X17)
-1.4028
-1.397
.0526
-0.0738
-0.0102
Elderly married (base)


.0721
___
217


Table B-15. Continued
Category
Parameter
estimates
Pi
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
8E(Q)b
9Xi
Expected change
among consuming units
3E(Q*) .F(Z)
3Xi
(1)
Sex of meal planner:
(2)
(3)
(4)
(5)
(6)
Female (X23)
0.0150
0.029
.1190
0.0018
0.0018
Male (base)
Family size:
Total number in
.1190
household (X24)
Total number
-1.2057
-3.203
.1190
-0.0759
-0.0131
squared (X25)
Education of meal planner:
0.0858
2.582
Years (X26)
Guest meals:
0.2529
5.141
.1190
0.0301
0.0052
Number of meals (X27)
Meals away from home:
0.2397
6.379
.1190
0.0285
0.0049
Dollars (X28)
-0.0015
-0.244
.1190
0.0018
-0.0003
224


Table B-5. Continued
Category
Parameter
estimates
a.
i
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
9E(EXP)b
3X.
Expected change
among consuming units
9E(EXP*) F(Z-t)
9X.
(1)
Income before taxes:
(2)
(3)
(A)
(5)
(6)
Thousand dollars (X29)
Thousand dollars
0.0985
2.653
.1401
0.0072
0.0014
squared (X30)
Interaction terms:
-0.0004
-2.004
Income and race (X31)
Income and family
-0.0451
-1.383



size (X32)
Other seafood:
0.0018
0.396
Dollars (X33)
Constant:
-0.0503
-1.738
.1401
-0.0070
-0.0013
a0
-10.5679
-11.539



a
The value of Xg at the means of all X-j_'
's is equal
to -5.05160; a
= 5.500982.
^The effects of the interaction and/or squared terms have been accounted for in the construction of the
associated linear terms.
175


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EFDV77QD6_KND24H INGEST_TIME 2014-11-17T20:52:44Z PACKAGE AA00026376_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


23
seafood products. Capps' study does provide considerable information
for assistance in determining which variables should be included in a
seafood consumption equation. Using a quadratic expenditure equation
Capps found that region, urbanization, race, martial status, household
size, household income, and price all contributed in a statistically
significant manner in explaining seafood expenditures. In agreement
with the results provided by Salathe, Capps found the income elasticity
of seafood expenditures to be extremely inelastic, equalling 0.1651.
With respect to family size, Capps found an elasticity of 0.2296 which
is somewhat less than that reported by Salathe.
Perry's analysis of seafood expenditures was by far the most
complete of those utilizing the 1972-74 BLS Consumer Expenditure
Survey. In addition to specifying a rather complete model describing
expenditures on total seafood and specific product forms (shellfish,
canned fish, whole fish, and filleted/steaks fish) in terms of
variables introduced into the equations, the analysis incorporated all
households in the survey. Furthermore, to avoid the likelihood of
biased estimates of the true parameters associated with using ordinary
least squares when a large concentration of zero observations for the
dependent variable is presented in the data, Perry estimated the
equations via a Tobit procedure. This procedure provides asymptoti
cally consistent estimates of true parameters given a correct model
specification. The variables included in the various seafood/seafood
product expenditure equations were household income, race, urbaniza
tion, expenditures on food consumed away from home, occupation of
household head, education level of household head, and household
composition. By estimating separate expenditure functions for the four


Abstract of Dissertation Presented to the Graduate School of
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
SOCIOECONOMIC DETERMINANTS OF AT-HOME SEAFOOD
CONSUMPTION: A LIMITED DEPENDENT VARIABLE
ANALYSIS OF EXISTING AND LATENT CONSUMERS
By
Walter R. Keithly, Jr.
August, 1985
Chairman: Frederick J. Prochaska
Major Department: Food and Resource Economics
Weekly household at-home seafood consumption in the United States
was analyzed using 1977-1978 Nationwide Food Consumption Survey data.
The cross-sectional consumption study related expenditure and quanti
ties consumed of total seafood and five specific products (fresh,
frozen, canned, finfish, and shellfish) to a set of socioeconomic and
demographic factors which influence at-home seafood consumption
patterns.
A Tobit procedure was used in the estimation of the various
seafood product equations. The model, though used for statistical
reasons, provided considerable information which was used in examining
existing seafood consumers as well as potential seafood consumers.
The results of the analysis appear logical and useful. For the
most part, the estimated parameters were consistent with theoretical
expectations and/or results of previous studies. Region, urbanization,
race, household size, the stage of household growth and maturity,
ix


SOCIOECONOMIC DETERMINANTS OF AT-HOME SEAFOOD
CONSUMPTION: A LIMITED DEPENDENT VARIABLE
ANALYSIS OF EXISTING AND LATENT CONSUMERS
BY
WALTER R. KEITHLY, JR.
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR
THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1985


16
demand equations, they do show the importance of including prices and
income in single demand equations.
Though the above restrictions offer little assistance in the
formulation of single equation demand models, they are of value in the
formulation of complete systems of demand equations. There are
additional restrictions offered by economic theory which are useful
in the formulation of complete systems of demand equations. First,
as illustrated by the budget constraint (equation 2.1), the sum of
expenditures on individual commodities must equal income in
equilibrium. A second restriction suggests that the sum of marginal
expenditures is unity, i.e.,
PiS1 = 1 (i= 1. 2, .... n) (2.7)
This restriction guarantees that an increase in income is associated
with increased expenditures on at least one good. Third, the zero
homogeneity of the demand functions yields another n restrictions on
the slope coefficients of the following form:
m
3m
n
+ £ Pa a
J dp
9qi
= 0
(i,j = 1, 2, ..., n)
j=l
(2.8)
Finally, another in(n 1) restrictions can be obtained through
symmetry conditions
9q. 3q 3q. 3q.
i i i + 1
3p. qj 3m 3p^ i 3m
(i,j = 1, 2, ..., n) (2.9)


Table B-2. Continued
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
9E(EXP)b
SXi
Expected change
among consuming units
9ECEXP*) F(Z-i)
3X
(1)
Income before taxes:
(2)
(3)
(4)
(5)
(6)
Thousand dollars (X29)
Thousand dollars
squared (X30)
Interaction terms:
0.1926
-0.0006
3.958
-1.963
.1949
0.0234
0.0051
Income and race (X31)
Income and family
-0.0615
-1.505



size (X32)
Other seafood:
-0.0067
-0.961
Dollars (X33)
Constant:
-0.0787
-1.130
.1949
-0.0153
-0.0033
a0
-6.9439
-5.575



The value of Xg at the means of all X^'
s is equal
to -7.43939; a
= 8.68865.
^The effects of the interaction and/or squared terms have been accounted for in the construction of the
associated linear terms.
160


LIST OF FIGURES
Figure Page
1-1 U.S. annual per capita consumption of commercial
fish and shellfish (edible weight), 1960-83 4
1-2 U.S. supply of edible fishery products (round
weight), 1960-83 7
viii


CHAPTER II
THEORETICAL CONSIDERATIONS, REVIEW OF
RELATED WORK, AND MODEL DEVELOPMENT
Cross-sectional consumption studies have become increasingly
accepted and utilized during the past two decades in conjunction with
the increasing availability of appropriate data sets and the advent of
the high speed computer. With the increased volume of literature deal
ing with the estimation of cross-sectional demand has come increased
sophistication in terms of model development and estimation techniques.
Regardless of the degree of sophistication employed in the analysis of
cross-sectional data, the fact remains that the theory of demand is the
basis for all model development in this area. Given this fact, the
present chapter begins with a review of the theory of demand. This
theory is then adapted to estimation of cross-sectional quantity
demanded and expenditure models. Following this section, a review of
related seafood consumption studies is presented. The chapter then
concludes with a discussion dealing with specification of the models
used in this study, the statistical procedures employed in the estima
tion of these models, and the data used in the analysis.
Theoretical Considerations of Demand and Consumption
Theoretical Considerations of Utility Maximization
The theory of demand is developed on the postulate that consumers
maximize utility subject to a resource (income) constraint, i.e.,
12


99
percentage of meals consumed away from home and the size of the
household. Perhaps the most important information contained in Table
3-6 is that associated with the percentage of meals consumed away from
home according to race. Among Black households, the percentage of
lunch and dinner meals consumed away from home increased 70 and 204
percent, respectively, compared to only 37 and 66 percent,
respectively, among nonblack households.
Given the propensity for away-from-home seafood consumption com
pared to at-home consumption of seafood, it is important to recognize
the emerging patterns. Just as important, however, is the recognition
that the away-from-home market is likely to expand even more in the
long run, largely at the expense of the at-home market. A growing body
of research suggests that the total expenditure/income elasticity for
food consumed away from home is approximately twice that of the at-home
consumption market (e.g., Eastwood and Craven, 1981; Haidacher et al.,
1982). Second, though the estimated price elasticity associated with
food consumed at home is apparently more inelastic than that associated
with food consumed away from home, the importance of this factor in
stabilizing at-home food consumption will probably be negated by
changes in the socioeconomic structure of the population. For example,
increases in the education level and the proportion of females entering
the work force, and a decline in household size are all believed to
lead to increased expenditures on and the number of meals consumed
away from home (see Prochaska and Schrimper (1973) and Redman (1980)
for a discussion of those factors which determine away-from-home
consumption).


41
X24 =
family size (total number living in household)
X25 =
family size squared
X26 =
number of years of schooling of meal planner
X27 =
number of guest meals served from household food
supply in previous 7 days
X28 =
dollar value of meals purchased and consumed away
from home (excluding snacks)
X29 =
before tax income (thousand dollars)
X30 =
before tax income squared
X31 =
interaction between before tax income and race
(X18 X29)
X32 =
interaction between before tax income and family size
(X24 X29)
X33 =
expenditures on (or quantity consumed) of alternative
product forms
l, a2, a3,
... 33 =
estimated coefficients associated with weekly
expenditure equations
81 Bo Bo
... ^33 =
estimated coefficients associated with weekly
quantity equations
U1 "
normally distributed random disturbance specific to
the expenditure equations
u2 =
normally distributed random disturbance specific to
the quantity consumed equations
Though most of the independent variables included in equations
(2.12a) and (2.12b) enter in a binary manner (X1-X23), family size
(X24), education level of the meal planner (X26), guest meals (X27),
money value of meals consumed away from home (X28), before tax income
(X29), and expenditures (or quantities) of alternative seafood product
forms (X33) enter the equations in a continuous manner. Among this
latter group of variables, family size and income were specified in a


105
Assuming that the quality elasticity of at-home consumption of
seafood is positive, one would expect the quantity elasticity to be
somewhat less than the expenditure elasticity. The estimate of the
quantity elasticity, 0.1814, was in fact less than the estimated
expenditure elasticity which equalled 0.2389. This translates to a
quality elasticity for seafood consumed at home equal to 0.0515. Thus,
an increase in consumption resulting from an increase in income is
expected to result in a greater increase in expenditures than quantity
consumed; the difference measuring a demand for quality and/or
services.
Household income, measured in 1982 dollars, for selected years and
among different races is given in Table 3-9. As indicated from the
data in the table, after growing steadily throughout the 1950s and
1960s, the real household income stagnated during the 1970s and even
decreased in the early 1980s. The decline in growth in real household
income during the 1970s compared to the previous ten year period and
the actual decline in real household income during the early 1980s has
probably affected the growth in per capita consumption of commercial
fish and shellfish (Figure 1-1) and especially at-home seafood
consumption. Black households, estimated to have a higher propensity
to consume seafood at home with respect to income than did White
households, experienced a 13 percent decline in real income between
1970 and 1982 compared to only a 3 percent decline among White
households. Given the growing proportion of Black households in the
United States, the decline in real income among this group poses an
obstacle in any attempt to increase at-home consumption of seafood.


107
Given the relatively low estimate of the at-home seafood consump
tion elasticity with respect to income, even large increases in real
household income such as that of the 34 percent increase experienced
during the decade of the 1960s, will not have an overbearing effect
on at-home consumption of seafood. For instance, an increase in
real household income during the decade of the 1980s equal to that
experienced during the decade of the 1960s was estimated to result in
only a 6.2 percent increase in the quantity of seafood consumed at home
with related expenditures increasing about 8.1 percent.
Outlook for Increasing At-Home Demand
for Seafood and Implications
The outlook for increased at-home seafood consumption over the
next several years does not appear promising without significant
advances by the seafood industry and its support groups in terms of
more effective marketing and promotional efforts. For example, generic
advertising on seafood is extremely small when compared to most other
O
food sectors. Furthermore, nongeneric advertising on seafood in major
media outlets (excluding newspapers) declined 10 percent between 1977
and 1982 compared to a 55 percent increase in nongeneric advertising
associated with meat and a 167 percent increase in nongeneric advertis
ing associated with poultry (Anonymous, 1984). These factors alone
would tend to indicate a disadvantage to the seafood industry vis-a-vis
other food sectors; all competing for a limited household food budget.
Furthermore, as discussed throughout the analysis, movement in the
g
Generic advertising on seafood averaged $85,000 annually during
1981-82 compared to $4.2 million on red meats and almost $27 million on
milk and other dairy products (Morrison and Armbruster, 1983).


APPENDIX B
DISAGGREGATED SEAFOOD STATISTICS


228
Longwoods Research Group Limited (The). "A Usage Segmentation Analysis
of the 1981 U.S. Seafood Consumption Study (final report),"
prepared for the Fisheries Council of Canada, October, 1984.
Maddala, G. S. Econometrics. New York: McGraw-Hill Book Co.,
1977.
McDonald, J. F., and R. A. Moffitt. "The Uses of Tobit Analysis,"
Review of Economics and Statistics 62(1980):318-321.
Mincer, J. "Market Prices, Opportunity Costs, and Income Effects,"
In Carl Chrest (editor), Measurement in Economics.
Stanford: Stanford University Press, 1963:66-82.
Morrison, R. V/., and W. Armbruster. "Generic Adverstising of Farm
Products," National Food Review NFR-23(1983):14-18.
Muellbauer, J. "Household Composition, Engel Curves, and Welfare
Comparisons between Households: A Duality Approach," European
Economic Review 5(1974):102-122.
Murphy, P. E., and W. A. Staples. "A Modernized Family Life Cycle,"
Journal of Consumer Research 6(1979):12-22.
O'Rourke, A. D. "Marketing and Distribution Problems with Extended
Jurisdiction." Jin L. G. Anderson (editor), Economic Impacts
of Extended Jurisdiction. Ann Arbor: Ann Arbor Science
Publishers Inc., 1977:237-246.
Perry, J. S. "An Econometric Analysis of Socioeconomic and Demo
graphic Determinants of Fish and Shellfish Consumption in the
United States." Ph.D. dissertation, Gainesville: University
of Florida, 1981.
Phelps, C. E. "LIMDEPA Regression Program for Limited Dependent
Variables." International Publication, The Rand Corporation,
1972.
Prais, S. J., and H. S. Houthakker. An Analysis of Family Budgets.
Cambridge, England: Cambridge University Press, 1955.
Prochaska, F. J., and R. A. Schrimper. "Opportunity Cost of Time and
Other Socioeconomic Effects on Away-From-Home Consumption,"
American Journal of Agricultural Economics 55(1973):595-603.
Purcell, J. C., and R. Raunikar. Analysis of Demand for Fish and
Shellfish. Athens: University of Georgia, Department of
Agricultural Economics Bulletin 51, December 1968.
Redman, B. J. "The Impact of Women's Time Allocation on Expenditure
for Meals Away from Home and Prepared Foods," American Journal
of Agricultural Economics 62(1980):234-237.


Table B-3. Continued
Category
Parameter
estimates
gi
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
3E(Q)b
3X
Expected change
among consuming units
9E(Q*) F(Z )
3X
(1)
Income before taxes:
(2)
(3)
(A)
(5)
(6)
Thousand dollars (X29)
Thousand dollars
0.1420
3.445
.1894
0.0151
0.0032
squared (X30)
Interaction terms:
-0.0005
-1.735
Income and race (X31)
Income and family
-0.0469
-1.358



size (X32)
Other seafood:
-0.0049
-0.825
Pounds (X33)
Constant:
-0.3641
-3.282
.1894
-0.0690
-0.0147
-6.2095
-5.889



aThe value of Xg at the means of all X^'s is equal to -6.41971; a = 7.31141.
^The effects of the interaction and/or squared terms have been accounted for in the construction of the
associated linear terms.
165


100
The results of the analysis indicate a rather strong negative
relationship between expenditures on meals away from home and weekly
at home consumption of seafood which is in contrast to the results
reported by Perry (1981). A $10 increase in expenditures on meals
consumed away from home was estimated to reduce at-home consumption of
seafood by $0,057 and 0.059 pounds, respectively. This relationship is
especially important considering the growing away-from-home consumption
market and one which must be considered in any attempt to increase the
at-home demand for seafood.
Income before Taxes
Income was found to be an important determinant of seafood consumption
as has been the case in most related studies (e.g., Capps, 1982;
Salathe, 1979; Haidacher et al., 1982; Perry, 1981). The statistically
significant positive linear effect (X29) and the negative coefficient
estimated for income squared (X30) imply a positive but declining
marginal propensity to purchase and consume seafood at home with
increasing income. The statistically significant parameter estimate
associated with the interaction term between income and race (X31)
suggests a different marginal propensity of at-home seafood consump
tion among households of different races. The negative estimate of
this term indicates that white households have a lower marginal
propensity to purchase and consume seafood at home at all levels of
income than nonwhite households, ceteris paribus. The positive
estimate of the parameter associated with the interaction between
family size and income (X32), though statistically insignificant,


Table 4-2. Signs of estimated parameters
models3
associated with
variables
included
in Tobit
seafood ex
penditure
Seafood
product
form
Variable
Total
Fresh
Frozen
Canned
Finfish
Shellfish
Region:
Northeastern (XI)
+*
+*
_*
+*
+*
+
North Central (X2)
'/c
+
.it.
Southern (X3)
-
-
-V.
+
Urbanization:
Central City (X4)
+*
+*
+
+*
+*
+*
Suburban (X5)
+*
+*
+
+*
+*
+
Season:
Spring (X6)
+
+
+
-
+
-
Summer (X7)
+
+*
-
+
~
Fall (X8)
+

>C
+
Household life cycle:
Young single w/o children
(X9)
.JJ.
-
+*
-
Young married w/o children
(X10)
_*
-
+
+
Young single with children
(Xll)
-V.
_-;s-
+
-
Young married with children (X12)
_*
-
+
_
+*
Middle aged single w/o children (X13)
"
+
115


22
As discussed by George and King (1971), the increase in average price
per unit of the commodity associated with an increase in household
income can be viewed as a demand for quality or services. Furthermore,
an estimate of the quality elasticity for commodity q^ can be defined
as the difference between the expenditure elasticity for that commodity
with respect to income and the quantity elasticity for that commodity
with respect to income (George and King, 1971). Given that other price
variations related to space, package size, etc., have been adequately
accounted for by inclusion of variables such as region, urbanization,
family size, etc., in the estimated equation, the quality elasticity
hence measures the percentage change in average price paid for a
commodity with respect to a percentage change in income^ and can be of
considerable interest "as a measure of consumers' desire for improved
quality or services, given the present average or standard quality"
(George and King, 1971, p. 72). Though the "quality/services" concept
as generally discussed is associated only with income, it can easily be
extended to reflect other economic variables which take continuous
values.
Statistical Considerations
The statistical considerations most often addressed with respect
to cross-sectional demand analyses concern the functional form of the
demand equations and the treatment of nonconsumers in the analyses.
Though considerable research has been conducted in an effort to find
the "ideal" functional form (e.g., Prais and Houthakker, 1955;
3
For proof of the relationship see George and King (1971, p. 72).


24
then there exists no rationale for excluding the latter group from the
analysis. The parameter estimates associated with such an analysis can
then be interpreted as reflecting the average of both consumers and
nonconsumers. In order to determine whether consumers and nonconsumers
represent a homogeneous group, the reasons why nonconsumers may be
present in a given sample are addressed by Currie et al.
The first reason provided by Currie et al. for observing non
consumers in the sample proceeds as follows. Assume the distinction
between consumers and nonconsumers can be adequately summarized in
terms of some qualitative variable(s), say religion. In this case, two
options are available to the researcher. First, the researcher can
exclude the nonconsumers from the analysis, in which case the results
should be interpreted as exclusive of that religious segment of the
population. Alternatively, the researcher has the option of including
all households in the analysis by explicitly accounting for the
qualitative difference of religion.
As a second example, the authors consider the case in which non
consumers can be explained as a result of a difference in the level of
some quantitative variable, say income, between them and the consuming
group. This being the case, there is no reason to expect behavioral
differences between the consuming and nonconsuming groups if incomes
were equal. With an increase in the level of income, nonconsumers
should enter the market and react in a similar manner to that of the
consuming group. Hence there is no rationale, from a theoretical
standpoint, for excluding the nonconsuming group from the analysis
under this condition.


11
Chapter II. The empirical results associated with weekly total seafood
consumption are analyzed in Chapter III. In Chapter IV, a discussion
of the results associated with the specific seafood product form models
is presented. In the last chapter of the main text, Chapter V, conclu
sions of the study are presented as are suggestions for future research
in this area.


3
Federal seafood inspection program which operates under the auspices
of the National Marine Fisheries Service inspected only about 20
percent of the 2.8 billion pounds of seafood consumed in the U.S. in
1982 (Becker, 1983). Though the average consumer probably does not
realize that seafood requires no federal inspection before sale to the
public, he/she is often reminded of some of the adverse health related
issues associated with consumption of certain seafood products. For
example, periodic newspaper headline scares such as those in the early
to mid 1970s regarding high mercury content in certain finfish species
and those related to occasional outbreaks of cholera resulting from the
consumption of contaminated raw oysters has left the consumer in a
quandry concerning the safety of eating these seafood products.
Though the preceding discussion points a bleak picture of the
future of the seafood industry in the United States, evidence to the
contrary suggests that consumption of seafood will be an increasingly
important component of the American household diet. For example, the
desire among American consumers to increase consumption of lower
calorie, natural, and more nutritious foods will likely translate to
increased seafood consumption (Slavin, 1984). Recent trends in the per
capita consumption of fish and shellfish also suggest that it will be
an increasingly important component of the American household diet.
As illustrated in Figure 1-1, per capita consumption of commercial
fish and shellfish has gradually been trending upwards over the past
two and one-half decades. During the 1960-64 period per capita
consumption of commercial fish and shellfish averaged 10.56 pounds
annually. By the 1979-83 period, annual per capita consumption had
increased 21 percent to 12.78 pounds (United States Department of


101
indicates propensity to increase at-home seafood consumption with
increases in family size, ceteris paribus.
The nonlinearity specification of the income component in the
expenditure and quantity models makes it important to evaluate the
effect of income on weekly expenditures and quantities of seafood
consumed at various levels of income. The more important effects are
presented in Table 3-7.
The information given in Table 3-7 indicates a declining, albeit
small, marginal propensity to consume seafood at home with increases in
household income. For example, at an income level of $5,000, the total
expected change in weekly expenditures with respect to a $1,000 change
in income equalled $0,026. At an annual household income level equal
to $25,000, a $1,000 change in income was predicted to change the total
expected weekly expenditures by $0.0242, or about 93 percent of the
estimated change at an annual income level of $5,000.
About 25 to 35 percent of the change in at-home seafood consump
tion was estimated to reflect changes among those households already in
the market in terms of either increases or decreases in weekly expendi
tures and quantities consumed. The remaining 65 to 75 percent of the
change in at-home consumption of seafood, therefore, reflects changes
in the probability of market participation, either entry or exit
weighted by expected expenditures or quantities consumed.
The positive estimates of the total change in seafood consumption
and change in seafood consumption among participating households at
various levels of income as specified in Table 3-7 indicates that the
level of income required to reach a saturation level of seafood
consumption was far in excess of that reported by most households in


149
Season (X6-X8)
Surveyed seasons of the year are
Spring (X6)Months of April, May and June, 1977.
Summer (X7)Months of July, August, and September, 1977.
Fall (X8)Months of October, November, and December, 1977.
Winter (base)Months of January, February, and March, 1978.
Household Life Cycle Stage
Young households with or without children (X9-X12)head of
household is less than 35 years old.
Middle aged households with or without children (X13-X16)
head of household is greater than or equal to 35 years of
age but less than 65 years of age.
Older households (X17-base)head of households is 65 years
of age or greater.
Seafood Expenditures and Quantity Consumed
The seafood expenditure and quantity data used in this analysis
includes food commodities with the first four digit codes equalling
4521 or 4522 in the 1977-78 National Food Consumption Survey data
tapes. As such, food items which included a seafood product as only
one component of the total item were excluded from the analysis.
Seafood quantities were reported in the form brought into the kitchen.


Table B-13. Continued
(1)
(2)
(3)
(4)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Mean (percent)
Household life cycle:
Young single w/o children (X9)
0.055
0.053
0.055
0.073
Young married w/o children (X10)
0.058
0.077
0.056
0.101
Young single with children (Xll)
0.035
0.022
0.036
0.048
Young married with children (X12)
0.151
0.140
0.152
0.071
Middle aged single w/o children (X13)
0.075
0.058
0.077
0.059
Middle aged married w/o children (X14)
0.115
0.142
0.113
0.094
Middle aged single with children (X15)
0.056
0.052
0.057
0.071
Middle aged married with children (X16)
0.261
0.342
0.255
0.100
Elderly single (X17)
0.099
0.048
0.103
0.037
Elderly married (base)
0.095
0.066
0.096
0.053
Race of respondent:
White (X18)
0.852
0.864
0.851
0.077
Other (X19)
0.030
0.033
0.029
0.084
Black (base)
0.118
0.103
0.120
0.067
212


APPENDICES


Table B-9. Continued
Category
Parameter
estimates
*1
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
3EC Q)b
9X*
Expected change
among consuming units
3E(Q*) F(Z.)
9Xi
(1)
(2)
(3)
(4)
(5)
(6)
Household life cycle:
Young single w/o
children (X9)
0.2294
2.518
.3974
0.0912
0.0283
Young married w/o
children (X10)
0.0349
0.443
.3336
0.0116
0.0033
Young single with
children (X11)
0.0996
1.022
.3557
0.0354
0.0103
Young married with
children (X12)
0.0323
0.468
.3336
0.0108
0.0030
Middle aged single
w/o children (X13)
0.1245
1.513
.3632
0.0452
0.0133
Middle aged married
w/o children (X14)
0.0953
1.465
.3520
0.0335
0.0097
Middle aged single
with children (X15)
0.2560
3.207
.4052
0.1037
0.0324
Middle aged married
with children (X16)
0.1046
1.527
.3557
0.0372
0.0108
Elderly single (X17)
-0.0533
-0.685
.3050
-0.0163
-0.0043
Elderly married (base)


.3228


192


Table B-ll. Continued
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X.
3E(EXP)b
3Xt
Expected change
among consuming units
3E(EXP*) F(Z,)
3X
(1)
Income before taxes:
(2)
(3)
(4)
(3)
(6)
Thousand dollars (X29)
Thousand dollars
0.0490
2.451
.4522
0.0124
0.0040
squared (X30)
Interaction terms:
-0.0001
-1.440
Income and race (X31)
Income and family
-0.0283
-1.619



size (X32)
Other seafood:
0.0020
0.780
Dollars (X33)
Constant:
0.1321
4.802
.4522
0.0597
0.0197
a0
-2.7392
-5.645



aThe value of Xg at the means of all X^'s is equal to -0.5239; a = 4.2606.
^The effects of the interaction and/or squared terms have been accounted for in the construction of the
associated linear terms.


141
in center city areas consumed more seafood in total and more of all
product forms with the exception of frozen. Season was of significance
in explaining seafood consumption only in isolated instances.
Within the second category, variables denoting family size and the
composition of the household were used to explain at-home seafood
consumption. At-home consumption of seafood in total and for frozen,
canned, and finfish were found to be positively related to initial
increases in family size while at-home consumption of fresh seafood
and shellfish were negatively related to increases in family size.
To measure household composition, a set of variables reflecting the
life cycle of the 'typical' household was included in the analysis.
The results pertaining to this set of variables indicate that the
composition of the household, independent of size, is very important
in explaining at-home seafood consumption patterns in total and for
specific product forms. For example, households in the younger stages
of their life cycle, especially those with children, avoided consump
tion of fresh and finfish seafood products and total seafood relative
to households in more mature stages of their life cycle. Elderly
households, on the other hand, consumed the least canned seafood of
any life cycle stage, ceteris parabus.
Within the third category, variables denoting the race of the
household, the employment status of the meal planner, the sex of the
meal planner, the education of the meal planner, expenditures on meals
consumed away-from-home, the number of guest meals, and the household
catching fish were used to explain seafood consumption. Race of house
hold was used in the analysis to account for variations in tastes and
preferences among households of different races which in turn would


84
the more mature household life cycles. A related factor pertains to
the hesitency among parents of serving seafood to younger children out
of fear that the bones in the fish may injure the children or that the
children will have problems eating certain types of seafood (such as
shellfish). All of these reasons suggest a marketing strategy aimed at
promoting highly processed/ready-to-eat types of seafood products to
this segment of the population. Additionally, one would expect to find
fresh seafood consumption, which requires the most preparation time and
with which bones are most frequently associated, to be lower among
younger households, especially those with children, than among other
households. However, for frozen and canned seafood for which prepara
tion time is minimal and bones are generally not a problem, one would
expect to observe little differences in consumption patterns among the
households categorized in the different life cycles. This hypothesis
will be examined in greater detail in the following chapter.
The characteristics and composition of the American household is
in a continual state of transition. Knowledge of this transition in
conjunction with the information provided by the results pertaining to
the household life cycle category can further aid the seafood industry
and its support groups in the planning stage of a long term marketing
strategy.
The first factor the seafood industry and its support groups
may wish to consider when planning a long term seafood marketing
strategy is the changing age structure of the American household.
Table 3-4 provides some statistics on the age distribution of
household heads for selected years. The statistics suggest that
younger households (those with household heads less than 35 years of


114
than shellfish products. Second, shellfish is considerably more
expensive than finfish (Table 4-1).
Comparisons of Consumption Parameters
Information concerning the signs and statistical significance of
the estimated parameters associated with the specific product form and
total seafood expenditure models is presented in Table 4-2. Since the
signs and statistical significance associated with the estimated
quantity models are generally in agreement with those of the expendi
ture models they are not presented here. Summarizing the results in
this manner allows for an easy and logical comparison of the estimated
models. The complete descriptive and summary statistics associated
with each of the specific seafood product form models are presented in
Appendix B.
Region
Overall, household expenditures on most seafood product forms were
highest among households residing in the Northeastern region of the
United States (XI) and lowest among households residing in the North
Central region (X2). This difference holds true for expenditures on
fresh and canned seafood. However, expenditures on frozen seafood
showed a reversal of that of the other two processed product forms with
respect to Northeastern and North Central regions. Weekly household
expenditures on frozen seafood were estimated to be greatest in the
North Central region of the United States and lowest in the North
eastern region.


Table B-2. Summary statistics for Tobit analysis of weekly household expenditures on fresh seafood3
Category
Parameter Asymptotic
estimates t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X.
9E(EXP)b
Expected change
among consuming units
3E(EXP-) F(Z.)
3X.
(1)
(2)
(3)
(4)
(5)
(6)
Region:
Northeastern (XI)
1.1633
3.035
.2389
0.2779
0.0658
North Central (X2)
-2.6341
-6.209
.1271
-0.3348
-0.0616
Southern (X3)
-0.2033
-0.524
.1949
-0.0396
-0.0086
Western (base)


.2005


Urbanization:
Central City (X4)
1.9322
5.718
.2266
0.4378
0.1011
Suburban (X5)
0.7099
2.181
.1867
0.1325
0.0281
Nonmetro (base)


.1660


Season:
Spring (X)
0.5887
1.639
.1977
0.1164
0.0254
Summer (X7)
1.2707
3.583
.2206
0.2803
0.0641
Fall (X8)
0.1812
0.517
.1867
0.0338
0.0072
Winter (base)


.1788


156


13
max u = u (q) J (2.1)
subject to p'q = m
where
u = utility derived from the consumption of q
q = an n-element column vector of quantities of
commodities available in the market place
p = an n-element column vector of market prices
m = consumer income
To maximize u(q) subject to the income constraint, the Lagrangian
expression
L(q,A) = u(q) + A(m p'q) (2.2)
is differentiated with respect to the arguments q and A. The
resultant derivatives are then set equal to zero. The following n + 1
first order conditions are the result of such a procedure:
- Ap =0 (i = 1, 2, . ., n) (2.3)
3q 1
m p' q = 0
The first n equations of (2.3) satisfy the condition that in
equilibrium the consumer has equated the ratio of marginal utilities
derived from the consumption of any two commodities to the ratio of
their respective prices. Additionally, in equilibrium, the marginal
utility derived from the consumption of each commodity divided by its
respective price equals A, the marginal utility of income. The last
equation of (2.3) reinforces the condition that in equilibrium the
consumer has exhausted his total resources (income).


95
United States declined more than 18 percent to 2.72 individuals (United
States Department of Commerce, Bureau of the Census, various issues).
Thus, holding all other factors constant, the decline in household size
between 1960 and 1982 would have resulted in about a 7 percent decline
in expenditures and approximately an 8 percent decline in weekly
quantity of seafood consumed at home. Of course, other factors have
not remained constant. The declining family size, for instance,
reflects the gradual change in household composition. Increases in the
proportion of younger and older households, as discussed earlier, at
the expense of middle aged households will yield a smaller household
size. All of these factors must be viewed simultaneously when con
sidering expected changes in at-home seafood consumption throughout
the nation.
The declining household size has another important implication
that the seafood marketing sector may wish to consider when conducting
and planning a long term marketing program. Primarily, the seafood
industry and its support groups should recognize and react accordingly
to the fact that consumers of seafood products are going to desire
smaller portions of seafood due to the decline in family size. /
Education of Meal Planner
Seafood consumption was estimated to be positively related to the
education level of the meal planner (X26). Each additional year of
education of the meal planner increased the total household expendi
tures on seafood consumed at home by just over $0.04 per week, ceteris
paribus. Similarly, the total household at-home consumption of seafood
was estimated to increase by almost 0.03 pounds per week with each


17
Though these restrictions are vital in the formulation of complete
systems of demand equations, again they are of little value in the
formulation of single demand equations.
Aggregation Considerations
Though the demand functions developed earlier and specified in
equation (2.4a) provide the foundation for demand analyses, estimation
of the equations as specified in equation (2.4a) is generally difficult
because the equations refer to only a specific individual or household
and hence variation in the data is not present. Empirical demand
studies, on the other hand, are generally conducted by aggregating
consumption and income across all households for several different
time periods or, alternatively, treating the household as the main
consuming unit while estimating the demand equations for a cross-
section of households. The first approach, referred to as a time-
series analysis of demand, is used to study the effect of changes in
prices and income on consumption through time. With the second
approach, referred to as a cross-sectional demand analysis, differences
among household units are explicitly accounted for in the estimation
^procedures. Thus, determining the effects of differences in socio
economic factors and income across households on consumption patterns
becomes the primary objective of cross-sectional studies. Given the
cross-sectional nature of the present study, the discussion to follow
centers on the estimation of cross-sectional demand functions.


Table B-10. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(A)
Proportion
of category
consuming
Actual mean
values
Family size:
Total number in household (X24)
2.946
3.181
2.735
Total number squared (X25)
11.457
13.040
10.031

Education of meal planner:
Years (X26)
11.732
11.885
11.594

Guest meals:
Number of meals (X27)
1.141
1.223
1.066

Meals away from home:
Dollars (X28)
11.658
11.361
11.926

Income before taxes:
Thousand dollars (X29)
14.109
14.938
13.363
Thousand dollars squared (X30)
324.590
353.983
298.117

199


Table B-2. Continued
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X-^
9E(EXP)b
9X.
Expected change
among consuming units
9E(EXP*) F(Z-¡ )
9X.
(1)
(2)
(3)
(4)
(5)
(6)
Race of respondent:
White (X18)
-5.2222
-8.982
.1736
-1.0623
-0.2185
Other (X19)
-2.4925
-3.329
.2676
-0.667
-0.1663
Black (base)


.3707

Receives food stamps:
Yes (X20)
0.3583
0.660
.2061
0.0738
0.0164
No (base)


.1949


Caught fish for own use:
Yes (X21)
3.1191
10.557
.2743
0.8556
0.2167
No (base)


.1685


Employment of meal planner:
Yes (X22)
-0.6360
-2.154
.1841
-0.1171
-0.0248
No (base)


.2061

158


households (X3) by $0.83 and $0.61, respectively. The information
provided in column 5 of Table 3-3 suggests the total expected weekly
consumption of seafood by households in the Northeastern region
exceeded that of households in the North Central region, Southern
region, and Western region by 0.52 pounds, 0.30 pounds, and 0.30
pounds, respectively, ceteris paribus.
The relatively high estimates associated with at-home seafood
expenditures for a household in the Northeastern region compared to
households in other regions of the United States are the result of two
factors. First, the estimated probabilities of a household having
positive weekly consumption expenditures and quantities, given in
column 4 of Tables 3-2 and 3-3, exceed those associated with any other)
region. Second, the expected expenditures among consuming households
residing in the Northeast exceeded those of households in other
regions, as noted in the last column of Tables 3-2 and 3-3.
Given the estimated differences in at-home seafood expenditures
among households residing in the different regions, establishing
probable causes for these differences may be beneficial to the seafood
industry and its support groups. Traditionally, the Northeast region
has had an extensive fishing industry. This factor, in conjunction
with the close proximity of most of the Northeastern States to the
ocean, has resulted in a relatively steady supply of fresh seafood to
the households in this region. Transporting fresh seafood products
from coastal states to the inland states, such as many of those located
in the North Central region, is risky and expensive. O'Rourke (1977)
categorizes the U.S. seafood marketing system into two distinct
segments. The first segment, defined by O'Rourke (p. 239) as the


Table 3-3. Continued
Category
Parameter
estimates
*1
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in Xn-
9E(Q)b
Expected change
among consuming units
9E(Q*) F(Z.)
axt
(1)
(2)
(3)
(4)
(5)
(6)
Other seafood:
Pounds (X33)
NAC
NA
NA
NA
NA
Percent of households:
*0
-2.5837
-5.772



aThe value of Xg at the
means of all X^'
's is equal
to -0.52941; a
= 3.96192.
bThe effects of the interaction and/or squared terms have been accounted for in the construction of the
linear g^ terms associated with those variables.
Not applicable.


80
Season
At-home quantity of seafood consumed was estimated to be greatest
in the summer quarter (X7), followed by the spring quarter (X6), fall
quarter (X8), and winter quarter (base). Weekly expenditures, however,
were found to be much more constant across quarters with no statistical
differences noted for any season, as judged by the nonsignificance of
the asymptotic t-values associated with the parameter estimates (column
3 of Table 3-2). A quantity change not reflected by a corresponding
expenditure change suggests that price must also be changing.
During the one year period in which the survey data used in this
analysis were collected (April 1977-March 1978), the price for seafood
consumed at home as measured by the Consumer Price Index increased by
7.6 percent (United States Department of Agriculture, 1983). Given the
7.6 percent increase in the price of seafood consumed at home, the
question arises as to why the estimated weekly expenditures on seafood
consumed at home did not show a similar increase. In fact, weekly
expenditures were estimated to be lower (though only marginally) in the
later half of the survey year than in the first half. A possible
explanation put forward to answer this question is that households may
have reacted to the increase in the price of seafood consumed at home
by reducing quantities purchased leaving weekly household expenditures
on seafood unchanged. This would imply the quantity elasticity with
respect to price must equal approximately unity.
.rA


119
The relatively high expenditures on fresh seafood by households
living in the Northeastern region of the United States is consistent
with the steady supply of fresh edible seafood products harvested and
sold in the Northeastern region. Similarly, the relatively high
expenditures on frozen seafood by households residing in the North
Central region may be explained by the relative unavailability of
fresh seafood in that region and hence the need for substitute frozen
products. Apparently, households in the Northeastern region can sub
stitute fresh seafood for frozen seafood which explains the relatively
low expenditures on frozen seafood in that region. The relatively high
expenditures on canned seafood in the Northeastern region compared to
the other regions of the United States was, however, unanticipated
given the availability of this product form in all regions of the
country at expected comparable prices. A possible higher demand for
at-home consumption of seafood in total in the Northeastern region due
to differences in tastes and preferences may help to explain the higher
expenditures on both fresh and canned seafood products in that region
compared to other regions of the United States.
The results of the analysis indicated that expenditures on both
finfish and shellfish were highest in the Northeastern region of the
United States and lowest in the North Central region. These differ
ences in expenditures between the two regions probably reflect differ
ences in availability of fresh quality seafood as well as differences
in consumer preference between the two regions.
The preceding discussion should be considered by the seafood
industry and its support groups when considering what products and
product forms to provide in different regions of the country.


REFERENCES
Amemiya, T. "Regression Analysis When the Dependent Variable is
Truncated Normal," Econometrica 41(1973):997-1016.
Amemiya, T. "Tobit Models: A Survey," Journal of Econometrics
24(1984):3-61.
Anonymous. "Conference Told to Step Up Advertising, PR," Seafood
Business Report 3(Winter 1984):21.
Becker, G. S. Mandatory Federal Inspection System: An Overview,
Washington, D.C.: Congressional Research Service, The Library of
Congress, TX 501 B, Report No. 83-198-ENR, 1983.
Blokland, J. Continuous Consumer Equivalence Scales. The Hague:
Martinus Nighoff, 1976.
Bockstael, N. E. "Uncertainty about Consumption and Consumer
Uncertainty," Marine Resource Economics 1(1984):67-77.
Brown, J. A. C., and A. S. Deaton. "Surveys of Applied Economics:
Models of Consumer Behavior," The Economic Journal 82(1972):
1145-1236.
Burk, M. C. Influences of Economic and Social Factors on U.S.
Consumption. Minneapolis: Burgess Publishing Co., 1961.
Buse, R. C. Data Problems in the BLS/CES PU-2 Dairy Tape: The
Wisconsin 1972-1973 CES Dairy Tape. Madison: University of
Wisconsin, Agricultural Economics Staff Paper 164, July 1979.
Buse, R. C., and L. E. Salathe. "Adult Equivalent Scales: An
Alternative Approach," American Journal of Agricultural Economics
60(1978):460-468.
Capps, 0., Jr. Analysis of Aggregate Fish and Shellfish Expenditure.
Blacksburg: Virginia Polytechnic Institute and State University,
Agricultural Economics Bulletin 82-1, May 1982.
Cato, J. C., and F. J. Prochaska. "Market Development in the Mid
western U.S. for Gulf and South Atlantic Seafoods from 1977 to
1980," Proceedings of the Sixth Annual Tropical and Subtropical
Conference of the Americas. College Station: Texas A&M
University, TAMU-SG-82-101, 1981.
226


Table 3-2. Summary statistics for Tobit analysis of weekly household expenditures on total seafood0
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
3E(EXP)
aXi
t\)
Expected change
among consuming units
3E(EXP*) F(Z-¡)
9Xi
(1)
(2)
(3)
(4)
(5)
(6)
Region:
Northeastern (X1
0.8147
4.903
(. 5469^)
0.4456
0.1738
North Central (X2)
-0.9383
-5.412
.4061
-0.3810
-0.1216
Southern (X3)
-0.3578
-2.146
.4523
-0.1618
-0.0547
Western (base)


.4812


Urbanization:
Central City (X4)
1.0843
7.507
.5198
0.5636
0.2104
Suburban (X5)
0.4888
3.628
.4717
0.2306
0.0802
Nonmetro (base)


.4322


Season:
Spring (X6)
0.0154
0.101
.4771
0.0073
0.0026
Summer (X7)
0.1428
0.937
.4873
0.0696
0.0248
Fall (X8)
-0.1414
-0.959
.4644
-0.0657
-0.0226
Winter (base)


.4759




229
Salathe, L. E. "Household Expenditure Patterns in the U.S.,"
Washington, D.C.: U.S. Department of Agriculture, Economics,
Statistics, and Cooperative Service, Technical Bulletin No.
1603, April 1979.
Slavin, J. W. "Review of U.S. Seafood Market," Infofish 2(1984):
22-26.
Tobin, J. "Estimation of Relationships for Limited Dependent
Variables," Econometrica 26(1958):24-26.
Tomek, W. G. "Empirical Analyses of the Demand for Food: A Review."
In R. Raunikar (editor), Food Demand & Consumption Behavior.
Athens: University of Georgia, March 1977.
United States Department of Agriculture, Human Nutrition Service.
Food Consumption: Households in the United States, Seasons and
Year 1977-78, Nationwide Food Consumption Survey 1977-78,
Washington, D.C.: United States Department of Agriculture, Human
Nutrition Service, Report No. H-6, 1983.
United States Department of Commerce, Bureau of the Census.
Statistical Abstract of the United States. Washington, D.C.:
Bureau of the Census, 1970-1985.
United States Department of Commerce, NOAA, NMFS. Fisheries of the
United States. Washington, D.C.: NOAA, NMFS, 1965-1984.
Van Dress, M. G. Dinning Out: Separate Eating Places Keep Customers
Happy, Suppliers Busy. Washington, D.C.: United States Depart
ment of Agriculture, Economics Research Service, Agriculture
Information Bulletin No. 459, July 1983.
Vondruska, J. "U.S. Consumer Attitudes toward Fish," Infofish
3(1984):30-34.
Vondruska, J. "Trends in U.S. Markets for Processed Shrimp."
Unpublished paper presented at the annual meeting of the National
Shrimp Processors Association, Lake Buena Vista, Florida, February
20-23, 1985.
Wilson, J. R. "Seafood Production, Markets, and Policies: United
States." In^ B. Thaker (editor), Seafood Production, Markets,
and Policies: Belgium, Federal Republic of Germany, Ireland, The
Netherlands, The United States. Corvallis: Oregon State
University, ORESU-X-82-002, 1982:108-157.


121
extent, possibly the other product forms as well. Households residing
in nonmetro areas of the country are often farming families or live in
farming communities. As such, generations of households have been
raised on meat products and introduction of seafood products may not
have been initiated to an extent necessary to establish a strong
at-home demand for these products among these households.
Season
Generally, season of the year was estimated to be statistically
insignificant in explaining weekly household expenditure on at-home
consumption of seafood. Weekly expenditures on fresh seafood products
consumed at home were found to be highest in the summer months (July,
August, September, 1977), while expenditures on frozen seafood products
were highest in the spring months (April, May, June, 1977), and
expenditures on canned seafood products were estimated to be highest in
the winter months (January, February, March, 1978), ceteris paribus.
Household Life Cycle
The household life cycle category (X9-X17, base) was very useful
in explaining weekly expenditures on at-home consumption of the
specific seafood product forms. For example, expenditures on fresh
seafood tended to increase with the maturing of the household, with
elderly households generally having the greatest expenditures on fresh
seafood products (Table 4-2). Expenditures on canned seafood consumed
at home, however, were lowest among elderly households (X17, base),
ceteris paribus. Expenditures on frozen seafood products attributable
to differences in the stages of the household life cycle were


Table B-ll. Continued
Category
Parameter
estimates
a.
i
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X-^
3E(EXP)b
dX
Expected change
among consuming units
9E(EXP*) F(Z. )
3X.
(1)
Sex of meal planner:
(2)
(3)
(M
(5)
(6)
Female (X23)
0.6162
2.990
.4562
0.2811
0.0899
Male (base)
Family size:
Total number in
.3974
household (X24)
Total number
0.64569
4.305
.4522
0.2249
0.0760
squared (X25)
Education of meal planner:
-0.0280
-1.980
Years (X26)
Guest meals:
0.0455
2.413
.4522
0.0206
0.0068
Number of meals (X27)
Meals away from home:
0.0960
5.155
.4522
0.0434
0.0143
Dollars (X28)
-0.0131
-4.435
.4522
-0.0059
-0.0019
204


33
The remaining socioeconomic variable entered into equations
(2.11a) and (2.11b), Z4-Z13, M, and S were included to specifically
account for variations in seafood consumption among different house
holds resulting from underlying socioeconomic differences which are
expected to influence tastes and preferences. The measurement of
household composition, Z4, used in this study is a modification of the
household life cycle classification proposed by Murphy and Staples
(1979) and is more directly applicable to the study than measurements
generally proposed for reasons discussed below. For purposes of this
study, households were stratified according to ten mutually exclusive
life cycle classifications: young single without children, young
married without children, young single with children, young married
with children, middle aged single without children, middle aged married
without children, middle aged single with children, middle aged married
with children, older single, and older married (where young is defined
as the head of household being less than 35 years old, middle aged is
defined as head of household being from 35 years old to 65 years old,
and elder is defined as head of household being equal to or greater
than 65 years old). There were two reasons for using the household
life cycle measurement of family composition in this study as opposed
to a more traditional measurement. First, it is useful to investigate
the reasons for the changes in apparent consumption of total seafood
and specific product forms over the past two decades. Available time
series data pertaining to household composition are related to the
family life cycle measurement of household composition more closely
than with the other measurements. Thus, the life cycle classification
was a preferable measurement of household composition in terms of


Table 3-2. Continued
Category
Parameter
estimates
a.
i
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
9E(EXP)b
BXi
Expected change
among consuming units
9E(EXP"-) .F(Zi)
9Xi
(1)
Sex of meal planner:
(2)
(3)
(A)
(5)
(6)
Female (X23)
0.5679
2.425
.4790
0.2720
0.0958
Male (base)
Family size:
Total number in
.4417
household (X24)
Total number
0.4689
2.721
.4759
0.2062
0.0722
squared (X25)
Education of meal planner:
-0.0169
-1.036
Years (X26)
Guest meals:
0.0858
3.982
.4759
0.0408
0.0143
Number of meals (X27)
0.1649
7.831
.4759
0.0785
0.0275


Table B-13. Continued
(1)
(2)
(3)
(4)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Receives food stamps:
Yes (X20)
No (base)
Caught fish for own use:
Yes (X21)
No (base)
Employment of meal planner:
Yes {122)
No (base)
Sex of meal planner:
Female (X23)
Male (base)
Mean (percent)
0.074
0.926
0.049
0.951
0.234
0.766
0.281
0.719
0.465
0.535
0.513
0.487
0.908
0.092
0.921
0.079
0.076
0.924
0.050
0.078
0.230
0.770
0.092
0.072
0.461
0.539
0.084
0.069
0.906
0.094
0.077
0.065
213


21
homemaker, age of children, employment status of homemaker) directly
into the demand functions for meals away from home and prepared food.
The rationale for excluding prices from cross-sectional demand
analyses (equation 2.10) is based on the concept that households
observe the same price for any given commodity at a point in time.
Though the validity of this assumption has been addressed (e.g.,
Mincer, 1963) and relaxed in applied research (e.g., Capps, 1982; Cox
et al., 1984), the exclusion of prices from cross-sectional demand
analyses remains the prevailing practice. Those who include prices in
a cross-sectional demand study do so on the premise that prices can be
expected to vary in some systematic manner across space and time (given
a sufficient period over which the data were collected). However, when
prices are included in the model, the interpretation of the estimated
price coefficients becomes exceedingly difficult due to price varia
tions independent of shifts in supply. For example, price variations
can result from differences in average prices per unit due to quality
variation, price discounts associated with larger purchases, and/or
price variations resulting from added services provided with the
"basic" commodity. When price is excluded from the estimated model,
variables which relate to these price variations are included in the
model. However, often the parameters associated with these variables
represent a composition of two effects: a direct effect associated
with the specific variable being analyzed and a price effect. For
example, assume the average price paid per unit of a commodity, q
to be positively related with income, m. Hence, as household income
increases total expenditures on that commodity, equaling p^q^, will
tend to increase proportionately more than the increase in q^.


75
consuming households resulting from a change in consistently
averaged from 30-40 percent of the total change. This implies that
approximately 35 percent of the change in at-home consumption of
seafood with respect to a change in X-¡^ is due to increased/decreased
consumption of seafood by those households currently consuming seafood
as opposed to entry or exit among households. Thus, approximately 65
percent of the total change is attributable to entry/exit into (from)
a fy*
the at-home seafood market by households. This finding has significant )
implications to the seafood industry and its support groups who would
r
like to know the relative merits and associated costs of increasing
at-home seafood consumption by enticing new consumers into the
at-home seafood market as opposed to increasing consumption among
currently consuming households.
Region
Total weekly expenditures on seafood consumed at home were esti
mated to be highest among the households residing in the Northeastern
region of the United States (XI) (Table 3-2) which is consistent with
the results presented by Perry (1981) and Capps (1982). Similarly,
total weekly quantity of seafood consumed by households residing in the
Northeastern region was estimated to be higher than that for other
regions (Table 3-3). Based on the information provided in column 5 of
Table 3-2, the total expected expenditures on seafood consumed at home
for a household in the Northeastern region were estimated to exceed
those for a household in the base region (West) by $0.45. Similarly,
expenditures among Northeastern households were estimated to exceed
expenditures among North Central households (X2) and Southern


78
industry and the support groups to look for ways of reducing costs and
preserving the freshness of seafood when transporting fresh seafood to
the inland regions of the country. Unfamiliarity with many seafood
products among North Central households may also help to explain the
differences in the at-home seafood consumption patterns across regions.
Thus, promotion aimed at familiarizing these households with the dif
ferent available products may prove useful in the short run. In the
long run, as the regional structure of the population shifts due to
increased population mobility and the transportation of fresh seafood
products becomes more economically feasible due to improved methods of
preserving the freshness of seafood, regional differences in seafood
consumption will probably decline naturally. For example, between 1950
and 1980, the proportion of the U.S. population living in the Northeast
and Midwest sections of the United States declined by 17 percent and 12
percent, respectively, while the proportion of the U.S. population
residing in the South and West increased by 7 percent and 44 percent,
respectively (United States Department of Commerce, Bureau of the
Census, 1984). Additional declines in the Northeast and Midwest
sections of the country are expected until at least the year 2000.
With these demographic shifts in population should come an exchange of
knowledge among households concerning different types of seafood and
methods of preparation which may eventually lead to the disappearance
of regional differences in at-home seafood consumption patterns.
Urbanization
Households in the central city (X4) were estimated to have higher
weekly consumption of seafood at home than those households in either


83
For example, the relatively low estimates of at home seafood consump
tion among those households categorized in the younger life cycle
categories establish the premise that seafood promotion/marketing
targeted towards this segment of the population may provide the seafood
industry and its support groups with greater net returns than that of
targeting households categorized in more mature stages of their life
cycle. Of course, the validity of this premise depends on the relative
costs associated with promoting seafood to households in the different
life cycles relative to the returns per dollar expended. However,
before targeting this group of households the reasons why this group of
households exhibits a relatively low level of seafood consumption needs
to be addressed. A couple of reasons can be offered to help explain
these results. First, the meal planner in the "younger" households
probably tends to be somewhat less experienced at preparing meals than
the meal planner in more mature households. Due to this factor, these
**- IBMJ.. > l*njr ,-!
meal planners are more likely to avoid cooking a meal which involves
any much of preparation. Seafood, especially fresh seafood, has a
reputation for being difficult to properly prepare. Thus, seafood may
not be prepared and consumed as often by "younger" households as would
be expected among more mature households.
A second explanation is specific only to those younger households
with children (X11-X12). This segment of the population exhibited the
lowest weekly seafood consumption among the different categories in
the household life cycle. The children in this household group will
generally be of a lower average age than in households categorized in
the more mature life cycles. Hence, these children will place a
larger burden on the meal planner's time than would older children in


Table B-9. Summary statistics for Tobit analysis of weekly household quantity consumption of canned
seafood
Category
Parameter
estimates
3i
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
9E(Q)b
9Xi
Expected change
among consuming units
9E(Q*) F(Z.)
(1)
(2)
(3)
(4)
(5)
(6)
Region:
Northeastern (XI)
0.2388
5.726
.4404
0.1052
0.0350
North Central (X2)
-0.2428
-5.473
.2843
-0.0690
-0.0178
Southern (X3)
-0.2396
-5.575
.2843
-0.0681
-0.0175
Western (base)


.3594


Urbanization:
Central City (X4)
0.2488
6.623
.3821
0.0951
0.0288
Suburban (X5)
0.1767
5.085
.3594
0.0635
0.0186
Nonmetro (base)


.3050


Season:
Spring (X6)
-0.0426
-1.085
.3483
-0.0148
-0.0043
Summer (X7)
-0.0212
-0.539
.3557
-0.0075
-0.0022
Fall (X8)
-0.0987
-2.589
.3336
-0.0329
-0.0092
Winter (base)


.3632


191


120
For example, the Gulf and South Atlantic Fisheries Development Founda
tion has coordinated the effort of several groups aimed at expanding
demand for many of the underutilized species landed in the Southeastern
United States which enjoy only regional acceptance. Much of the effort
has gone to developing markets for these products in the Midwestern
section of the United States. (See Cato and Prochaska (1981) for a
discussion of the program.) Given the familiarity with frozen seafood
products in this section of the country, promoting these underutilized
species in frozen form appears logical.
Urbanization
Households residing in central city (X4) and suburban areas (X5)
had higher estimated expenditures on all seafood product types and
forms than did households residing in nonmetro areas (base), control
ling for regional differences, income differences, etc. Furthermore,
with the exception of frozen seafood purchases, households in central
city areas consistently had higher expenditures on all seafood product
forms than did households residing in suburban areas, ceteris paribus.
The proximity of many of the larger cities in the United States to
major fishing ports would explain, in part, the higher expenditures on
fresh seafood consumed at home among households residing in central
city areas in the United States. However, proximity to the coast does
not explain the relatively high expenditures on canned seafood products
by central city households since canned seafood products can probably
be transported to the inland areas of the country at a minimal cost.
Therefore, some other factor must account for the estimated differences
in weekly household expenditures on canned seafood products and to some


Table B-15. Continued
Category
Parameter
estimates
h
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
3E(Q)
9X
Expected change
among consuming units
9E(Q*) .F(Z<)
3Xi
(1)
(2)
(3)
(4)
(5)
(6)
Income before taxes:
Thousand dollars (X29)
0.1259
2.737
.1190
0.0104
0.0028
Thousand dollars
squared (X30)
-0.0007
-2.632



Interaction terms:
Income and race (X31)
-0.0242
-0.603
Income and family
size (X32)
0.0025
0.405



Other seafood:
Pounds (X33)
0.1258
3.423
.1190
0.0150
0.0026
Constant:
eo
-12.7426
-9.941



The value of XB at the means of all X'
1 s is equal
to -7.45469; a
= 6.340.
The effects of the interaction and/or squared terms have been accounted for in the construction of the
associated linear terms.
225


Table 3-1. Continued
(1)
(2)
(3)
(4)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Mean
(percent)
a
Receives food stamps:
Yes (X20)
No (base)
Caught fish for own use:
Yes (X21)
No (base)
Employment of meal planner:
Yes {122)
No (base)
Sex of meal planner:
Female (X23)
Male (base)
0.074
0.926
0.077
0.923
0.234
0.766
0.264
0.736
0.465
0.535
0.454
0.546
0.908
0.092
0.932
0.068
0.071
0.929
0.529
0.506
0.202
0.798
0.573
0.488
0.476
0.524
0.496
0.518
0.882
0.118
0.521
0.370


Table B-7. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(A)
Proportion
of category
consuming
Actual mean
values
Interaction terms:
Income and race (X31)
12.745
14.514
11.925
Income and family size (X32)
46.628
57.549
41.562

Other seafood:
Dollars (X33)
1.076
1.204
1.017
Pounds (X33)
0.740
0.767
0.728

Weekly expenditures and quantity:
EXP
0.411
1.296
0.000
Q
0.231
0.729
0.000

Number of households:
10,689
3,387
7,302

Percent of households:
100
31.68
68.32

a
The data provided in this table associated with the binary variables (X1-X23) should be interpreted as
representing proportions rather than percentages. To obtain percentages, multiply data by 100.
185


19
characteristics of households, thought to influence tastes and prefer
ences and thus expenditures and/or quantity demanded of commodity i,
are controlled for by way of the vector of socioeconomic character
istics (Zj) particular to each household. Correct specification of the
vector Z continues to create controversy even though considerable
J
groundwork in this area was provided more than two decades ago by Prais
and Houthakker (1955) and Burk (1961). The problem with specifying
the vector Zj "correctly" is that it can potentially vary depending
upon the commodity being analyzed. For example, the socioeconomic
characteristics determining household demand for alcoholic beverages
may depend on a completely different set of cultural factors than
those factors determining household demand for milk. Thus, a review
of both economic literature and literature from disciplines specific
to the commodity being analyzed is essential for the correct model
specification. Prais and Houthakker and Burk discuss the importance
of examining factors such as family composition, social class,
religion, and demographic characteristics when analysing household
demand for food. These exploratory studies have subsequently been
extended and refined in an attempt to more closely model the actions of
the household. Two areas specific to the estimation of the household
demand for food frequently addressed by subsequent researchers are
those pertaining to measurement of household composition (e.g.,
Blokland, 1976; Muellbauer, 1974; Buse and Salathe, 1978; Murphy and
Staples, 1979) and the incorporation of the opportunity cost of time
of the homemaker (e.g., Mincer, 1963; Prochaska and Schrimper, 1973;
Redman, 1980).


CHAPTER V
CONCLUSIONS AND IMPLICATIONS FOR FURTHER RESEARCH
Conclusions
This study was conducted to enhance the current understanding of
socioeconomic and demographic factors believed to influence at-home
consumption of total seafood and specific seafood products (fresh,
frozen, canned, finfish, and shellfish). A better understanding of the
factors determining at-home consumption of seafood also provides
information which can be used in examining the away-from-home seafood
market, seafood import demand, and ultimately the total demand for
seafood in the United States.
In addition to the "traditional" estimates associated with
regression analysis, the Tobit procedure used in this analysis provided
a method for decomposing the change in the level of at-home seafood
consumption resulting from a change in any exogenous variable into two
components. The first component measured the change in consumption
resulting from increased (decreased) consumption among existing
(consuming) households. The second component measured the change in
total consumption related to an increase (decrease) in the number of
participating households. For the total seafood consumption models,
it was estimated that approximately 35 percent of the change in
at-home seafood consumption resulting from a change in an exogenous
variable results from increased/decreased consumption among
139


5
Commerce, NOAA, NMFS, 1983). Though per capita consumption of poultry
has increased at a much faster rate than that observed for seafood, per
capita consumption of red meats has been declining since 1971
(Vondruska, 1984).
An increase in per capita consumption of fresh and frozen seafood
from 1960 through 1983 has accounted for most of the increase in total
per capita seafood consumption during this period (Figure 1-1).
Averaging 5.82 pounds during the 1960-64 period, per capita consumption
of fresh and frozen fish and shellfish increased 35 percent to 7.86
pounds during the 1979-83 period. By comparison, per capita
consumption of canned fish and shellfish which averaged 4.22 pounds
annually during the 1960-64 period increased only 9 percent to 4.6
pounds during the 1979-83 period. Per capita consumption of cured fish
and shellfish which represents a very small portion of the total per
capita consumption of fish and shellfish has actually declined in
recent decades. Wilson (1982) offers two reasons for the increase in
per capita consumption of fresh and frozen fish and shellfish relative
to that of canned and cured fish and shellfish. First, there has been
an increased availability of fresh and frozen seafood products in
retail outlets in recent years. Much of this increase has resulted
from the introduction and acceptability by the consumer of highly
processed fish and frozen seafood products such as sticks and portions
for which per capita consumption increased 130 percent between the
1960-64 period and the 1979-83 period. The second reason offered by
Wilson for the increased per capita consumption of fresh and frozen
fish and shellfish relates to the recent increase in the number of
restaurants in the United States, especially those specializing in


48
For evaluation of the continuous variables in the model, use of equa
tions (2.14a) through (2.21a) is valid.
The total change in the dependent variable y given a change is
as specified in equation (2.17a) can be broken into two components.
The first component [F(Z)( 3E(y-55")represents the change in the
expected value of the dependent variable if above the limit (positive)
weighted by the probability of being above the limit (positive). The
second component [Eiy^KBFiZi/sX^) ] represents the change in the prob
ability of being above the limit (positive) weighted by the expected
value of the dependent variable if above. The breakdown of the total
change of the dependent variable into these two components is probably
very consistent with the actions of consumers in the market and is thus
useful in studying the consumption patterns of households.
It is interesting to note the simularity between the Tobit
estimates and OLS estimates. From equation (2.21a) it can be observed
that the effect of a change in X^ on the dependent variable y is equal
to only when F(Z) is equal to one. This is in fact what one would
expect since as F(Z) approaches one, OLS estimates should be obtained.
Multiplying equation (2.17a) by X^/E(y) gives the Tobit
elasticity, n^, which equals
[ F( ^
(2.22)
i
Substituting equation (2.16a) for E(y) and making the appropriate
reductions provides the following specification of the elasticity of y
with respect to Xi:


94
As presented in equation (2.23) an elasticity associated with
Tobit analysis can be broken into two components where the first term
reflects the percentage change in consumption among consuming units
due to a change in while the second component reflects the
elasticity of the change in the probability of consuming associated
with the changes in X^. Approximately 60 percent of the estimated
household size expenditure and quantity elasticities reflect an
increased probability of consuming seafood associated with increases
in household size while the remaining portion reflects increased
expenditures and consumption among those households already consuming
seafood as noted in the following two elasticity estimates:^
NEXp = 0.1635 + 0.2450 = 0.4085
Nq = 0.1957 + 0.2834 = 0.4791
Subtracting the quantity elasticity with respect to household size from
the expenditure elasticity with respect to household size provides a
measure of the quality elasticity which in this example was estimated
to equal -0.0706. The estimated negative quality elasticity indicates
that households tend to purchase relatively less expensive seafood
items with increases in household size.
Household size remained relatively stable from 1950 to 1960,
declining from an average of 3.37 to 3.33. However, since 1960, the
average household size has declined significantly reflecting both a
decline in birth rates and an increase in the number of single person
households. Between 1960 and 1982, the average household size in the
^Evaluated at the mean values of all variables.


A3
resulted in severe collinearity problems among the regressors.
Therefore, those used were only those considered most appropriate.
Statistical Considerations
The model developed in the previous section can be expressed in
matrix form as follows:
y
y
t
t
where,
yt
xt
ut
xte + ut if xte + ut > o
o if xt e + ut _< o
t = 1, 2, . N
dependent variable
vector of independent variables
2
error term assumed iid N(0, o )
(2.13)
This model specification referred to as the Tobit model after its
founder is well known and used extensively in economic studies of a
cross-sectional nature (see Amemiya (1984) for a good review of the
basic model and its uses). Given the specification in (2.13), an
assumption is implicitly made that an underlying stochastic index equal
to Xtg + U is observed only when strictly positive. In other words,
yt will only be positive given a value of XtB + Ut greater than zero.
Otherwise, yt will equal zero. For example, assume two households with
identical attributes with the exception of income. Furthermore, assume
that the household with the higher income consumed seafood while the
household with the lower income did not consume seafood. This would
imply that the first household with the higher income had exceeded that
threshold level required to consume seafood (i.e., Xt8 + Ut > 0), while


Table B-l. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(A)
Proportion
of category
consuming
Actual mean
values
Family size:
Total number in household (X24)
2.946
2.963
2.943
Total number squared (X25)
11.457
11.672
11.414

Education of meal planner:
Years (X26)
11.732
11.559
11.767

Guest meals:
Number of meals (X27)
1.141
1.333
1.102

Meals away from home:
Dollars (X28)
11.658
10.641
11.862

Income before taxes:
Thousand dollars (X29)
14.109
14.572
14.017
Thousand dollars squared (X30)
324.590
355.264
318.461

154


Page
V SUMMARY AND IMPLICATIONS FOR FURTHER RESEARCH 139
Summary 139
Implications for Further Research 143
APPENDICES
DEFINITIONS OF SELECTED VARIABLES 148
DISAGGREGATED SEAFOOD STATISTICS 151
REFERENCES 226
BIOGRAPHICAL SKETCH 230
v


SOCIOECONOMIC DETERMINANTS OF AT-HOME SEAFOOD
CONSUMPTION: A LIMITED DEPENDENT VARIABLE
ANALYSIS OF EXISTING AND LATENT CONSUMERS
BY
WALTER R. KEITHLY, JR.
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR
THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1985

ACKNOWLEDGEMENTS
Numerous people have helped in making this study and my graduate
program possible. Sincerest appreciation is extended to my chairman,
Fred J. Prochaska. He was always there to give me encouragement and
advice even during strained times. He should be credited with my
success but not failure in the academic field.
Dr. Scott Shonkwiler gave freely of his time in helping me with
questions of a statistical nature. Similarly, Drs. Cato, Kilmer, and
Otwell devoted their time to assure a better quality product than would
otherwise have been the case. Hopefully, their constructive criticisms
of this study make it more useful for the groups for which it is
intended.
I wish also to thank the Food and Resource Economics Department at
the University of Florida and Florida Sea Grant for giving me the
opportunity to pursue a graduate program and for providing financial
support.
I wish to thank Janet Eldred for the typing she did on this
manuscript. It sometimes got complicated sending everything through
the mail.
Finally, I wish to thank my parents for giving me the opportunity
to pursue a graduate career. Without their support I would not have
made it this far. Unfortunately, now they want me to pay them back.
ii

TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS ii
LIST OF TABLES vi
LIST OF FIGURES viii
ABSTRACT ix
CHAPTERS
I INTRODUCTION 1
Review of Seafood Consumption in the United States 1
Objectives 9
II REVIEW OF RELATED WORK AND MODEL DEVELOPMENT 12
Theoretical Considerations of Demand and Consumption 12
Theoretical Considerations of Utility
Maximization 12
Price and Income Considerations 14
Aggregation Considerations 17
Cross-Section Demand Analysis Considerations 18
Price, Income, and Socioeconomic Variable
Considerations 18
Statistical Considerations 22
Related Seafood Consumption Studies 25
Conceptual Model Development 31
Econometric Models and Statistical Considerations 39
Econometric Models 39
Statistical Considerations 43
Data Source and Considerations 49
iii

Page
III TOTAL SEAFOOD ANALYSIS 51
Descriptive Statistics Associated with Total
Seafood Analysis 51
Regression Estimates of Total Seafood Analyses 60
Region 75
Urbanization 78
Season 80
Household Life Cycle 81
Race 87
Household Receives Food Stamps 89
Household Caught Fish for Own Use 90
Employment of the Meal Planner 90
Sex of the Meal Planner 90
Household Size 91
Education of Meal Planner 95
Number of Guest Meals 96
Expenditures on Meal Consumed Away from Home 97
Income before Taxes 100
Outlook for Increasing At-Home Demand for Seafood
and Implications 107
IV SPECIFIC SEAFOOD PRODUCT FORM ANALYSIS Ill
Introduction Ill
Comparison of Consumption Parameters 114
Region 114
Urbanization 120
Season 121
Household Life Cycle 121
Race 123
Food Stamps 124
Fish Caught for Own Use 124
Employment of the Meal Planner 124
Sex of the Meal Planner 125
Family Size 125
Education of Meal Planner 129
Number of Guest Meals 129
Expenditures on Meals Away from Home 130
Income before Taxes 130
Other Seafood Expenditures 135
Outlook for Increasing At-Home Demand for Specific
Product Forms and Implications 136
iv

Page
V SUMMARY AND IMPLICATIONS FOR FURTHER RESEARCH 139
Summary 139
Implications for Further Research 143
APPENDICES
DEFINITIONS OF SELECTED VARIABLES 148
DISAGGREGATED SEAFOOD STATISTICS 151
REFERENCES 226
BIOGRAPHICAL SKETCH 230
v

LIST OF TABLES
Table Page
3-1 Descriptive statistics of variables in total
seafood models 52
3-2 Summary statistics for Tobit analysis of weekly
household expenditures on seafood 61
3-3 Summary statistics for Tobit analysis of weekly
household consumption of seafood 67
3-4 Percentage distribution of household head by age
for selected years 85
3-5 Estimated effects of changes in household size on
weekly expenditures and at-home seafood consumption 92
3-6 Percent of meals eaten away from home, by type of
meal and selected household characteristics, spring
1965 and spring 1977 98
3-7 Estimated effects of changes in before tax income
on weekly expenditures and quantities of seafood
consumed 102
3-8 Estimates of at-home seafood expenditure elastici
ties with respect to income 104
3-9 Median family income in constant (1982) dollars for
selected years 106
4-1 Descriptive statistics of data used in seafood
product form models 112
4-2 Signs of estimated parameter associated with vari
ables included in Tobit seafood expenditure models 115
4-3 Estimated weekly expenditure elasticities with
respect to family size for specific seafood product
forms 127
4-4 Estimated weekly expenditure, quantity and quality
elasticities with respect to before tax income for
specific and total seafood product forms 132
vi

Table Page
B-l Descriptive statistics of variables in fresh seafood
models 151
B-2 Summary statistics for Tobit analysis of weekly
household expenditures on fresh seafood 156
B-3 Summary statistics for Tobit analysis of weekly
household quantity consumption of fresh seafood 161
B-4 Descriptive statistics of variables in frozen
seafood models 166
B-5 Summary statistics for Tobit analysis of weekly
household expenditures on frozen seafood 171
B-6 Summary statistics for Tobit analysis of weekly
household quantity consumption of frozen seafood 176
B-7 Descriptive statistics of variables in canned
seafood models 181
B-8 Summary statistics for Tobit analysis of weekly
household expenditures on canned seafood 186
B-9 Summary statistics for Tobit analysis of weekly
household quantity consumption of canned seafood 191
B-10 Descriptive statistics of variables in finfish
seafood models 196
B-ll Summary statistics for Tobit analysis of weekly
household expenditures on finfish seafood 201
B-12 Summary statistics for Tobit analysis of weekly
household quantity consumption of finfish seafood 206
B-13 Descriptive statistics of variables in shellfish
seafood models 211
B-14 Summary statistics for Tobit analysis of weekly
household expenditures on shellfish seafood 216
B-15 Summary statistics for Tobit analysis of weekly
household quantity consumption of shellfish seafood 221
vii

LIST OF FIGURES
Figure Page
1-1 U.S. annual per capita consumption of commercial
fish and shellfish (edible weight), 1960-83 4
1-2 U.S. supply of edible fishery products (round
weight), 1960-83 7
viii

Abstract of Dissertation Presented to the Graduate School of
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
SOCIOECONOMIC DETERMINANTS OF AT-HOME SEAFOOD
CONSUMPTION: A LIMITED DEPENDENT VARIABLE
ANALYSIS OF EXISTING AND LATENT CONSUMERS
By
Walter R. Keithly, Jr.
August, 1985
Chairman: Frederick J. Prochaska
Major Department: Food and Resource Economics
Weekly household at-home seafood consumption in the United States
was analyzed using 1977-1978 Nationwide Food Consumption Survey data.
The cross-sectional consumption study related expenditure and quanti
ties consumed of total seafood and five specific products (fresh,
frozen, canned, finfish, and shellfish) to a set of socioeconomic and
demographic factors which influence at-home seafood consumption
patterns.
A Tobit procedure was used in the estimation of the various
seafood product equations. The model, though used for statistical
reasons, provided considerable information which was used in examining
existing seafood consumers as well as potential seafood consumers.
The results of the analysis appear logical and useful. For the
most part, the estimated parameters were consistent with theoretical
expectations and/or results of previous studies. Region, urbanization,
race, household size, the stage of household growth and maturity,
ix

number of guest meals, money value of meals consumed away from home,
the household having caught fish, and income were all contributing
factors which helped to explain at-home seafood consumption patterns.
The estimated income elasticities associated with total seafood con
sumption and consumption of all product categories were positive and
inelastic.
The analysis is distinguished from previous studies in two major
areas. First, the consumption effects were partitioned into those
for existing seafood consumers and those for potential consumers.
Second, consumption effects were separated into quantity and quality
components. These distinctions allow for a separate study of current
consumers and potential seafood consumers and for separation of
consumer expenditures into those for additional volume and those for
different qualities and associated marketing services.
The seafood industry and its support groups may wish to consider
the study when designing and implementing a long term promotion/
marketing program. Factors which determine at-home seafood consumption
are constantly changing. Consideration of these changes is given to
determine possible changes in at-home seafood consumption.
x

CHAPTER I
INTRODUCTION
Review of Seafood Consumption in the United States
Annual worldwide per capita consumption of fish and shellfish
averaged 27.1 pounds (live weight equivalent) during the 1975-77 period
(United States Department of Commerce, NOAA, NMFS, 1983). Japan, with
a per capita consumption of fish and shellfish equal to 148.6 pounds
during this period, was the world leader in terms of per capita
consumption. The United States, with a per capita consumption equal to
35.1 pounds, ranked 39th among all reported countries. Among developed
countries of the world, per capita consumption of fish and shellfish in
the United States is slightly less than two-thirds of the average.
However, per capita consumption of fish and shellfish in the United
States is approximately twice that of the undeveloped countries
(Wilson, 1982).
Though per capita consumption of fish and shellfish in the United
States trails that of several countries worldwide, total consumption of
fish and shellfish in the United States exceeds that of most other
countries due to the relatively large population in the United States
compared to other countries. After accounting for population, total
consumption of fish and shellfish in the United States is surpassed
only by Japan, the U.S.S.R., and China.
1

2
The reasons for the relatively low per capita consumption of fish
and shellfish in the U.S. compared to other developed countries of the
world are many and varied. First and foremost, the United States has
traditionally been the world's largest producer of beef and poultry
which has resulted in an abundant supply of these products at
relatively modest prices. The supply of edible fishery products, on
the other hand, has relied heavily upon imports to meet domestic
demand at acceptable prices. Second, in terms of ease of preparation,
fish and shellfish are typically rated poorly when compared to meat and
poultry products (Gillespie and Houston, 1975).1 This factor, in part,
has resulted in a large institutional and restaurant trade in seafood
products, while at-home consumption as a proportion of the total has
remained relatively low. Third, the demand for fish and shellfish in
the United States has been affected by the market distribution,
perishability, and preservation of these products (Christy and Scott,
1965). As one moves inland from those coastal states recognized as
major seafood producers, the availability of fresh seafood products
falls and the price increases. Finally, there remains a constant
concern among U.S. consumers regarding the quality of seafood being
sold in the various retail outlets. Meat and poultry products must be
inspected and certified by government representatives before sale
while inspection and certification of seafood products by government
representatives remains voluntary on the part of the seafood processor
(Becker, 1933). As such, inspection of seafood products has tradi
tionally been sporadic and minimal. For example, the most intensive
*83361 on a regional study in Texas conducted by the authors.

3
Federal seafood inspection program which operates under the auspices
of the National Marine Fisheries Service inspected only about 20
percent of the 2.8 billion pounds of seafood consumed in the U.S. in
1982 (Becker, 1983). Though the average consumer probably does not
realize that seafood requires no federal inspection before sale to the
public, he/she is often reminded of some of the adverse health related
issues associated with consumption of certain seafood products. For
example, periodic newspaper headline scares such as those in the early
to mid 1970s regarding high mercury content in certain finfish species
and those related to occasional outbreaks of cholera resulting from the
consumption of contaminated raw oysters has left the consumer in a
quandry concerning the safety of eating these seafood products.
Though the preceding discussion points a bleak picture of the
future of the seafood industry in the United States, evidence to the
contrary suggests that consumption of seafood will be an increasingly
important component of the American household diet. For example, the
desire among American consumers to increase consumption of lower
calorie, natural, and more nutritious foods will likely translate to
increased seafood consumption (Slavin, 1984). Recent trends in the per
capita consumption of fish and shellfish also suggest that it will be
an increasingly important component of the American household diet.
As illustrated in Figure 1-1, per capita consumption of commercial
fish and shellfish has gradually been trending upwards over the past
two and one-half decades. During the 1960-64 period per capita
consumption of commercial fish and shellfish averaged 10.56 pounds
annually. By the 1979-83 period, annual per capita consumption had
increased 21 percent to 12.78 pounds (United States Department of

Pounds (edible weight)
Figure 1-1. U.S. annual per capita consumption of commercial fish and shellfish (edible weight), 1960-83
SOURCE: United States Department of Commerce, NOAA, NMFS (1983).

5
Commerce, NOAA, NMFS, 1983). Though per capita consumption of poultry
has increased at a much faster rate than that observed for seafood, per
capita consumption of red meats has been declining since 1971
(Vondruska, 1984).
An increase in per capita consumption of fresh and frozen seafood
from 1960 through 1983 has accounted for most of the increase in total
per capita seafood consumption during this period (Figure 1-1).
Averaging 5.82 pounds during the 1960-64 period, per capita consumption
of fresh and frozen fish and shellfish increased 35 percent to 7.86
pounds during the 1979-83 period. By comparison, per capita
consumption of canned fish and shellfish which averaged 4.22 pounds
annually during the 1960-64 period increased only 9 percent to 4.6
pounds during the 1979-83 period. Per capita consumption of cured fish
and shellfish which represents a very small portion of the total per
capita consumption of fish and shellfish has actually declined in
recent decades. Wilson (1982) offers two reasons for the increase in
per capita consumption of fresh and frozen fish and shellfish relative
to that of canned and cured fish and shellfish. First, there has been
an increased availability of fresh and frozen seafood products in
retail outlets in recent years. Much of this increase has resulted
from the introduction and acceptability by the consumer of highly
processed fish and frozen seafood products such as sticks and portions
for which per capita consumption increased 130 percent between the
1960-64 period and the 1979-83 period. The second reason offered by
Wilson for the increased per capita consumption of fresh and frozen
fish and shellfish relates to the recent increase in the number of
restaurants in the United States, especially those specializing in

6
seafood preparation. In support of Wilson's contention, Van Dress
(1983) has estimated that the proportion of eating establishments
specializing in seafood preparation equalled 4.9 percent in 1979
compared to 2.1 percent in 1966.
Two features, already briefly alluded to, distinguish the seafood
industry in the United States from those industries associated with
most other major food products. First, the seafood industry is highly
dependent on an imported seafood product to satisfy domestic demand.
Second, consumption of seafood, as opposed to most other food products,
is highly related to the away-from-home food market.
The role of the international seafood market in meeting the
increasing per capita and total demand for seafood in the United States
can be observed with the aid of Figure 1-2. Between the 1960-64 period
and the 1979-83 period, total consumption of edible fishery products
increased by over 75 percent (Figure 1-2). This increase occurred in a
time period during which the population of the United States increased
only about 23 percent. As the consumption of edible fishery products
trended upwards, the composition of total supply shifted significantly
(Figure 1-2). Domestic supply of edible fishery products remained
relatively stable from 1960 through 1975 at approximately 2.5 billion
pounds annually (round weight). Associated with the passage of the
Magnuson Fisheries Conservation and Management Act of 1976, domestic
supply of edible fishery products in the United States experienced a
sizeable increase, and has averaged approximately 3.25 billion pounds
(round weight) annually since that time. Domestic supply appears to
have stabilized at this higher level in recent years. Overall, a 35.5
percent increase in the domestic supply of edible fishery products

Billion Pounds (whole weight)
Figure 1-2. U.S. supply of edible fishery products (round weight), 1960-83
SOURCE: United States Department of Commerce, NOAA, NMFS (various issues).

8
occurred between the 1960-64 and 1979-83 periods. While the domestic
supply of edible fishery products remained relatively stable through
out the 1960s and early 1970s, considerable growth occurred in the
imports of edible fishery products. Imports of edible fishery products
exceeded that of domestic supply beginning in 1966 and have since con
tinued to surpass domestic landings (Figure 1-2). Since the passage of
the Magnuson Fisheries Conservation and Management Act of 1976, imports
of edible fishery products have stabilized somewhat, averaging 4.75
billion pounds annually (round weight). Compared to a 35.5 percent
increase in the domestic supply of edible fishery products between the
1960-64 and 1979-83 periods, imports of edible fishery products have
increased by 131 percent, or almost four times that of the increase in
domestic supply. Imports of many of the higher valued seafood products
such as shrimp, scallops, and lobsters and imports of those fishery
products used in the preparation of processed frozen seafood items and
items served by the fast service eating establishments, such as fillets
and steaks frozen in blocks, have risen especially sharply in recent
decades. As the above discussion suggests, the growth in imported
edible fishery products is at least partially in response to a growing
domestic demand not met by domestic supply at acceptable prices.
Furthermore, given the apparent stability of domestic supply, further
increases in domestic demand will have to be met v/ith concurrent
increases in imports.
The other feature that distingushes the seafood industry from
those industries associated with most other food products concerns
the extent of the away-from-home market in the sales of seafood
products. Though no precise data is available indicating the extent

9
of away-from-home versus at-home consumption of fishery products, it
is estimated that anywhere from one-third (Vondruska, 1985) to two-
thirds (Bockstael, 1984) of all seafood is consumed away from home. As
noted by Bockstael (1984) the demand for many species (e.g., Crustacea,
fresh finfish) is highly sensitive to changes in income due to the
predominate restaurant trade in these species/products. Consequently,
in recessionary time periods when travelling, vacationing, and hence
restaurant trade are depressed, demand for these species/products will
also be depressed (Bockstael, 1984). During the 1973-74 recession and
the more recent economic slowdown starting in 1979, the depressed
demand for seafood resulted in a lowering of prices to the fishermen to
such an extent that the National Marine Fisheries Service was compelled
to initiate emergency programs such as the Catch America Program with
the intent of increasing consumer demand. Economic slowdowns and/or
recessions are a frequent occurrence in all or most developed
countries. Thus, the cyclical demand for seafood products in the
United States will likely be a recurring theme given the dependence of
seafood consumption on restaurant trade.
Objectives
Given the apparent relationship between seafood consumption and
restaurant trade, demand for seafood can be expected to rise and fall
in a cyclical manner in conjunction with oscillations in the general
economy. One means of alleviating the cyclical nature of the demand
for seafood is to encourage increased at-home consumption of seafood.
This can be done with increased at-home demand either at the expense of
the away-from-home market or independent of the away-from-home market.

10
If as in the first case, at-home consumption of seafood is increased at
the expense of the away-from-home market, little change in the total
demand for seafood may be experienced. If, on the other hand, at-home
demand for seafood is increased independent of the away-from-home
market, total demand for seafood by definition will be increased.
This could put pressure on many of the already heavily fished stocks
(as of 1974 about 62 percent of the economically important fisheries in
the United States were fully utilized or overfished (Eckert, 1979)) and
lead to an even greater demand for imported seafood.
In order to effectively increase at-home demand for seafood, a
thorough understanding of those factors hypothesized to determine
at-home consumption of seafood is required. Only then can the effects
of the at-home seafood market on the away-from-home market and
ultimately on the domestic fishing industry and on import demand for
seafood be fully comprehended. Thus, the overall objective of this
study is to examine and quantify those factors hypothesized to
determine at-home consumption of total seafood and specific product
forms (fresh, frozen, canned, shellfish, and finfish). In relation to
this objective, this study is designed to provide information concern
ing seafood marketing implications based on empirical findings and
historical trends related to factors which determine seafood
consumption. This information is essential in evaluating fishery
market legislation, management alternatives, long-term trends, and
promotion and marketing programs.
The format of this study proceeds as follows. A review of
literature, the models to be estimated, and a discussion of statistical
considerations and data used in analyzing the models are provided in

11
Chapter II. The empirical results associated with weekly total seafood
consumption are analyzed in Chapter III. In Chapter IV, a discussion
of the results associated with the specific seafood product form models
is presented. In the last chapter of the main text, Chapter V, conclu
sions of the study are presented as are suggestions for future research
in this area.

CHAPTER II
THEORETICAL CONSIDERATIONS, REVIEW OF
RELATED WORK, AND MODEL DEVELOPMENT
Cross-sectional consumption studies have become increasingly
accepted and utilized during the past two decades in conjunction with
the increasing availability of appropriate data sets and the advent of
the high speed computer. With the increased volume of literature deal
ing with the estimation of cross-sectional demand has come increased
sophistication in terms of model development and estimation techniques.
Regardless of the degree of sophistication employed in the analysis of
cross-sectional data, the fact remains that the theory of demand is the
basis for all model development in this area. Given this fact, the
present chapter begins with a review of the theory of demand. This
theory is then adapted to estimation of cross-sectional quantity
demanded and expenditure models. Following this section, a review of
related seafood consumption studies is presented. The chapter then
concludes with a discussion dealing with specification of the models
used in this study, the statistical procedures employed in the estima
tion of these models, and the data used in the analysis.
Theoretical Considerations of Demand and Consumption
Theoretical Considerations of Utility Maximization
The theory of demand is developed on the postulate that consumers
maximize utility subject to a resource (income) constraint, i.e.,
12

13
max u = u (q) J (2.1)
subject to p'q = m
where
u = utility derived from the consumption of q
q = an n-element column vector of quantities of
commodities available in the market place
p = an n-element column vector of market prices
m = consumer income
To maximize u(q) subject to the income constraint, the Lagrangian
expression
L(q,A) = u(q) + A(m p'q) (2.2)
is differentiated with respect to the arguments q and A. The
resultant derivatives are then set equal to zero. The following n + 1
first order conditions are the result of such a procedure:
- Ap =0 (i = 1, 2, . ., n) (2.3)
3q 1
m p' q = 0
The first n equations of (2.3) satisfy the condition that in
equilibrium the consumer has equated the ratio of marginal utilities
derived from the consumption of any two commodities to the ratio of
their respective prices. Additionally, in equilibrium, the marginal
utility derived from the consumption of each commodity divided by its
respective price equals A, the marginal utility of income. The last
equation of (2.3) reinforces the condition that in equilibrium the
consumer has exhausted his total resources (income).

14
Since the solution to (2.3) depends only on prices, income, and
the utility function, (2.3) can be solved yielding n + 1 equations, one
for each of the q.'s and one for X in terms of p and m.
i
q = q (p,m) (i = 1, 2, ..., n) (2.4a)
X = X(p>m) (2.4b)
The expressions in (2.4a) are referred to as demand functions, with the
demand for each of the n goods being expressed in terms of its own
price, prices of all substitutes, and income.
Price and Income Considerations
Economic theory suggests that demand functions satisfy certain
restrictions. One restriction is derived from the "fundamental equa
tion of value" theory which decomposes the effect of a price change
into the substitution and income effects, i.e.,
3
3q
(r-M q
op u=const j
3 m
(i,j = 1, 2, ..., n) (2.5)
A good is said to be normal if it has a downward sloping demand curve
or equivalently, a negative own price elasticity. Otherwise it is a
Giffen good which implies an upward sloping demand curve. The first
restriction which follows from the assumption of a strictly convex
indifference map implies that the own-price substitution effect is
1
negative, i.e.,
3qi
(^)
dp. u=const
i
< 0
(2.6)
See Hicks (1957) for a proof of this and subsequently discussed
restrictions.

15
Hence, for a good to be a Giffen good, the income effect must not only
be negative but also outweigh the own-price substitution effect.
Furthermore, a good whose income effect is negative is referred to as
inferior and conversely, a good whose income effect is positive is
referred to as a superior good. An Engel curve which relates consump
tion or expenditures on a good with income will thus be downward
sloping in the case of an inferior good and upward sloping in the case
of a superior good.
A second restriction offered by economic theory is that demand
functions (2.4a) are homogeneous of degree zero in all prices and
income. This suggests that an arbitrary scaling of all prices from
(p, m) to (ap, am) will have no effect on quantity demanded of any
good.
Unfortunately, these two restrictions have little to offer when
estimating the demand for a single good. Though the own-price
substitution effect is negative, the demand curve can be upward sloping
given that the good being analyzed is inferior and inferior to the
extent that the income effect outweighs the own-price substitution
effect. Thus, no restrictions can be imposed, a priori, on the sign of
2
the own-price coefficient. Furthermore, since a good can be either
inferior or superior, no restrictions can be imposed, a priori, on the
sign of the income coefficient. The second restriction is also of
little value in the estimation of single demand equations since rarely
if ever are all prices included in a single demand equation. Though
these restrictions offer little value in the estimation of single
2
This is of little concern in empirical demand analysis since few
if any goods have ever been shown to be Giffen goods.

16
demand equations, they do show the importance of including prices and
income in single demand equations.
Though the above restrictions offer little assistance in the
formulation of single equation demand models, they are of value in the
formulation of complete systems of demand equations. There are
additional restrictions offered by economic theory which are useful
in the formulation of complete systems of demand equations. First,
as illustrated by the budget constraint (equation 2.1), the sum of
expenditures on individual commodities must equal income in
equilibrium. A second restriction suggests that the sum of marginal
expenditures is unity, i.e.,
PiS1 = 1 (i= 1. 2, .... n) (2.7)
This restriction guarantees that an increase in income is associated
with increased expenditures on at least one good. Third, the zero
homogeneity of the demand functions yields another n restrictions on
the slope coefficients of the following form:
m
3m
n
+ £ Pa a
J dp
9qi
= 0
(i,j = 1, 2, ..., n)
j=l
(2.8)
Finally, another in(n 1) restrictions can be obtained through
symmetry conditions
9q. 3q 3q. 3q.
i i i + 1
3p. qj 3m 3p^ i 3m
(i,j = 1, 2, ..., n) (2.9)

17
Though these restrictions are vital in the formulation of complete
systems of demand equations, again they are of little value in the
formulation of single demand equations.
Aggregation Considerations
Though the demand functions developed earlier and specified in
equation (2.4a) provide the foundation for demand analyses, estimation
of the equations as specified in equation (2.4a) is generally difficult
because the equations refer to only a specific individual or household
and hence variation in the data is not present. Empirical demand
studies, on the other hand, are generally conducted by aggregating
consumption and income across all households for several different
time periods or, alternatively, treating the household as the main
consuming unit while estimating the demand equations for a cross-
section of households. The first approach, referred to as a time-
series analysis of demand, is used to study the effect of changes in
prices and income on consumption through time. With the second
approach, referred to as a cross-sectional demand analysis, differences
among household units are explicitly accounted for in the estimation
^procedures. Thus, determining the effects of differences in socio
economic factors and income across households on consumption patterns
becomes the primary objective of cross-sectional studies. Given the
cross-sectional nature of the present study, the discussion to follow
centers on the estimation of cross-sectional demand functions.

18
Cross-Sectional Demand Analysis Considerations
Price, Income, and Socioeconomic Variable Considerations
Two features of cross-sectional demand analysis of concern in
doing applied research in this field relate to proper model specifica
tion and method of estimation. In terms of model specification,
concern has traditionally centered around variable selection. In
practice, researchers have estimated demand functions (Equation 2.4a)
when using cross-sectional data as follows:
qiJ qij(mj Zj)
where
(i = 1, 2, ..., n)
(J = 1, 2, ..., k)
(2.10)
= quantity of the ith commodity demanded by the
Jth household
mj = income of Jth household
Zj = a column vector of socioeconomic characteristics
particular to the Jth household
The demand functions specified in (2.10) differ from those given by
(2.4a) in three aspects. First, a vector of socioeconomic characteris
tics (Zj) whose values are specific to the Jth household has been
included in the latter equation. Second, the vector of prices (p) has
been excluded from the latter equation. Third, all variables have been
subscripted denoting the Jth household.
Given the specification of the cross-sectional demand functions
(2.10), estimation entails the use of a sample of households differing
in socioeconomic characteristics and income rather than that of a
representative household as presented in (2.4a). Differences in

19
characteristics of households, thought to influence tastes and prefer
ences and thus expenditures and/or quantity demanded of commodity i,
are controlled for by way of the vector of socioeconomic character
istics (Zj) particular to each household. Correct specification of the
vector Z continues to create controversy even though considerable
J
groundwork in this area was provided more than two decades ago by Prais
and Houthakker (1955) and Burk (1961). The problem with specifying
the vector Zj "correctly" is that it can potentially vary depending
upon the commodity being analyzed. For example, the socioeconomic
characteristics determining household demand for alcoholic beverages
may depend on a completely different set of cultural factors than
those factors determining household demand for milk. Thus, a review
of both economic literature and literature from disciplines specific
to the commodity being analyzed is essential for the correct model
specification. Prais and Houthakker and Burk discuss the importance
of examining factors such as family composition, social class,
religion, and demographic characteristics when analysing household
demand for food. These exploratory studies have subsequently been
extended and refined in an attempt to more closely model the actions of
the household. Two areas specific to the estimation of the household
demand for food frequently addressed by subsequent researchers are
those pertaining to measurement of household composition (e.g.,
Blokland, 1976; Muellbauer, 1974; Buse and Salathe, 1978; Murphy and
Staples, 1979) and the incorporation of the opportunity cost of time
of the homemaker (e.g., Mincer, 1963; Prochaska and Schrimper, 1973;
Redman, 1980).

20
With respect to household composition, the authors generally
attempted to standardize the household in order to show the impact that
different household members, varying in terms of sex and age, had on
the consumption behavior of the household. The one exception to this
rule is provided by Murphy and Staples, who attempted to explain
differences in consumption among different households by examining the
different stages of a household's growth and maturity.
With respect to the opportunity cost of time, it has long been
recognized that a commodity, when consumed at home, is not in the same
form as when purchased. Rather, value is added to the commodity after
it is purchased to transform it into some "new" commodity suitable for
consumption. Mincer (1963) cognizant of this fact demonstrated that
estimated income elasticities for a variety of commodities will tend to
be biased if the opportunity cost of time is omitted. Prochaska and
Schrimper (1973) analyzed away-from-home consumption with respect to
the opportunity cost of time of the meal preparer. Redman (1980)
analyzed the impact of women's time on away-from-home consumption and
for prepared meals based upon the concept of the opportunity cost of
time as presented by Gronau (1977). Though conceptually similar, the
studies of Prochaska and Schrimper and Redman differ somewhat in the
treatment of the opportunity cost of time. Prochaska and Schrimper
estimated a wage rate for the meal preparer based upon a set of
arguments (education and age) and then included the estimated wage rate
as an argument in the demand equation for away-from-home consumption.
Redman, on the other hand, introduced those arguments hypothesized to
affect women's opportunity cost of time (education of homemaker, age of

21
homemaker, age of children, employment status of homemaker) directly
into the demand functions for meals away from home and prepared food.
The rationale for excluding prices from cross-sectional demand
analyses (equation 2.10) is based on the concept that households
observe the same price for any given commodity at a point in time.
Though the validity of this assumption has been addressed (e.g.,
Mincer, 1963) and relaxed in applied research (e.g., Capps, 1982; Cox
et al., 1984), the exclusion of prices from cross-sectional demand
analyses remains the prevailing practice. Those who include prices in
a cross-sectional demand study do so on the premise that prices can be
expected to vary in some systematic manner across space and time (given
a sufficient period over which the data were collected). However, when
prices are included in the model, the interpretation of the estimated
price coefficients becomes exceedingly difficult due to price varia
tions independent of shifts in supply. For example, price variations
can result from differences in average prices per unit due to quality
variation, price discounts associated with larger purchases, and/or
price variations resulting from added services provided with the
"basic" commodity. When price is excluded from the estimated model,
variables which relate to these price variations are included in the
model. However, often the parameters associated with these variables
represent a composition of two effects: a direct effect associated
with the specific variable being analyzed and a price effect. For
example, assume the average price paid per unit of a commodity, q
to be positively related with income, m. Hence, as household income
increases total expenditures on that commodity, equaling p^q^, will
tend to increase proportionately more than the increase in q^.

22
As discussed by George and King (1971), the increase in average price
per unit of the commodity associated with an increase in household
income can be viewed as a demand for quality or services. Furthermore,
an estimate of the quality elasticity for commodity q^ can be defined
as the difference between the expenditure elasticity for that commodity
with respect to income and the quantity elasticity for that commodity
with respect to income (George and King, 1971). Given that other price
variations related to space, package size, etc., have been adequately
accounted for by inclusion of variables such as region, urbanization,
family size, etc., in the estimated equation, the quality elasticity
hence measures the percentage change in average price paid for a
commodity with respect to a percentage change in income^ and can be of
considerable interest "as a measure of consumers' desire for improved
quality or services, given the present average or standard quality"
(George and King, 1971, p. 72). Though the "quality/services" concept
as generally discussed is associated only with income, it can easily be
extended to reflect other economic variables which take continuous
values.
Statistical Considerations
The statistical considerations most often addressed with respect
to cross-sectional demand analyses concern the functional form of the
demand equations and the treatment of nonconsumers in the analyses.
Though considerable research has been conducted in an effort to find
the "ideal" functional form (e.g., Prais and Houthakker, 1955;
3
For proof of the relationship see George and King (1971, p. 72).

23
Leser, 1963), no single functional form has emerged as clearly superior
under all conditions. There are, however, certain criteria that should
be considered when selecting and judging the appropriateness of a given
functional form (Brown and Deaton, 1972; Tomek, 1977; Hassan et al.,
1977). First, the functional form should allow for the possibility
that the commodity will not be consumed given an income below some
initial level. Second, the functional form should allow for a declin
ing marginal propensity to consume with increased income. Third, the
functional form should allow for a satiety level which provides an
upper bound on quantity consumed. Finally, simplicity and convenience
of estimation need to be considered. Though these criteria are valid
when considering quantity consumption functions, they appear to be
overly restrictive when considering expenditure consumption functions.
Given the additional aspect of demand for quality and services, there
is no reason to necessarily expect the marginal propensity of expendi
tures associated with some commodities to decline with increased
income. Similarly, a satiety level associated with expenditures on
some commodities is not necessarily expected, a priori.
As a commodity becomes more narrowly defined in an analysis, the
percentage of individual households not consuming that commodity
naturally increases. Therefore, a decision whether to include in the
analysis those households not consuming the commodity being analyzed
needs to be made. Currie et al. (1972) provide a good discussion
concerning under what conditions it is logical to exclude (include)
nonconsuming households from (in) the analysis. Basically, the issue
reduces to the following premise: if the consuming and nonconsuming
households can be considered as having identical behavioral patterns,

24
then there exists no rationale for excluding the latter group from the
analysis. The parameter estimates associated with such an analysis can
then be interpreted as reflecting the average of both consumers and
nonconsumers. In order to determine whether consumers and nonconsumers
represent a homogeneous group, the reasons why nonconsumers may be
present in a given sample are addressed by Currie et al.
The first reason provided by Currie et al. for observing non
consumers in the sample proceeds as follows. Assume the distinction
between consumers and nonconsumers can be adequately summarized in
terms of some qualitative variable(s), say religion. In this case, two
options are available to the researcher. First, the researcher can
exclude the nonconsumers from the analysis, in which case the results
should be interpreted as exclusive of that religious segment of the
population. Alternatively, the researcher has the option of including
all households in the analysis by explicitly accounting for the
qualitative difference of religion.
As a second example, the authors consider the case in which non
consumers can be explained as a result of a difference in the level of
some quantitative variable, say income, between them and the consuming
group. This being the case, there is no reason to expect behavioral
differences between the consuming and nonconsuming groups if incomes
were equal. With an increase in the level of income, nonconsumers
should enter the market and react in a similar manner to that of the
consuming group. Hence there is no rationale, from a theoretical
standpoint, for excluding the nonconsuming group from the analysis
under this condition.

25
As a final example, the authors consider the case in which non
consumers are observed only because the time period represented by the
survey does not cover a sufficient span of time necessary to observe
consumption by most or all households. As in the previous case, there
exists no reason to expect nonconsumers to exhibit a different
behavioral pattern from that of consumers and hence there exists no
economic reason for excluding them from the analysis.
Summarizing these three cases, if the differences between con
sumers and nonconsumers can adequately be accounted for, then there
exists no reason, from a methodological viewpoint, for excluding the
nonconsumers. Though there is no methodological rationale for exclud
ing the nonconsumers, caution must be taken when including nonconsumers
in the analysis because of statistical problems. As discussed later,
use of ordinary least squares when the data includes a large percentage
of nonconsumers will generally be inappropriate because of resulting
biased and inconsistent estimates of the true parameters. An appro
priate statistical technique to be used in conjunction with problems of
this nature will be presented towards the end of this chapter.
Related Seafood Consumption Studies
When compared to cross-sectional demand studies on those food
commodities which comprise a large percentage of the consumers food
budget, cross-sectional demand analyses for seafood products tend to be
somewhat limited. This probably reflects a lack of consideration of
"nontraditional" agricultural commodities in such surveys until recent
years.

26
Purcell and Raunikar (1968) provided one of the first comprehen
sive demand studies for seafood and seafood products. Though their
results may be of limited use in providing an understanding of current
U.S. seafood consumption patterns due to the regional specificity and
age of the study (the data consisted of quarterly observations on 160
households in Atlanta, Georgia, over the 1958 through 1962 period),
the study does provide some valuable information. In the analysis,
the following set of arguments were used to explain expenditures on
seafood: race, household composition, annual household income,
seasonality, a trend variable, gifts, and price. The results indi
cated that all variables with the exception of price and seasonality
were statistically significant in explaining expenditures on seafood.
Using data based on the 1972-74 BLS Consumer Expenditure Survey,
Salathe (1979), Capps (1982), and Perry (1981) each analyzed the
consumer demand for seafood and/or seafood products. Salathe expressed
expenditures on total seafood and two specific forms (canned fish and
fresh/frozen fish) as a function of household income and household
size. The analysis presented by Salathe may not be directly applicable
to this study for at least two reasons. First, arguments other than
household income and size undoubtedly influence expenditures on seafood
and seafood products. To the extent that the excluded variables are
correlated with household income and household size, the estimated
parameters associated with these two variables will tend to be biased.
The second reason that the results presented by Salathe may not be
directly applicable to this study relates to the econometric technique
employed by the author. In the analysis, Salathe included both con
sumers and nonconsumers of seafood products and proceeded to estimate

27
the expenditure equations via ordinary least squares. Given the
relatively large proportion of nonconsuming households of seafood
products in the 1972-74 Consumer Expenditure Survey, the statistical
technique used by Salathe is probably inappropriate, which will also
lead to biased estimates of the true parameters. Given the possible
bias of the parameter estimates presented by Salathe, they must be
interpreted with caution and viewed as only a rough approximation of
the true underlying parameters. Salathe found the expenditure
elasticities with respect to income for aggregate seafood and its two
components to be highly inelastic, ranging from 0.21 to 0.38. The
estimated household size elasticities of expenditures on aggregate
seafood and its two components were somewhat less inelastic, ranging
from 0.36 to 0.57.
The analysis presented by Capps is more complete than that of
Salathe's in some areas. The strength of the model developed by Capps
lies in the specification of the seafood expenditure equation which was
expressed as a function of region, urbanization, race, marital status,
education, occupation, tenure class, employment status of the female
household head, season, household size, household income, and price.
The drawback of the model presented by Capps lies in the exclusion of
those households who reported no expenditures on seafood during the
survey period. The consequence of excluding the nonpurchasing house
holds from the analysis is a loss of valuable information which could
potentially help to explain why some households purchased seafood
during the survey period while others did not. In addition, the
exclusion of nonpurchasers from the analysis indicates that the results
must be interpreted with respect to only those households purchasing

23
seafood products. Capps' study does provide considerable information
for assistance in determining which variables should be included in a
seafood consumption equation. Using a quadratic expenditure equation
Capps found that region, urbanization, race, martial status, household
size, household income, and price all contributed in a statistically
significant manner in explaining seafood expenditures. In agreement
with the results provided by Salathe, Capps found the income elasticity
of seafood expenditures to be extremely inelastic, equalling 0.1651.
With respect to family size, Capps found an elasticity of 0.2296 which
is somewhat less than that reported by Salathe.
Perry's analysis of seafood expenditures was by far the most
complete of those utilizing the 1972-74 BLS Consumer Expenditure
Survey. In addition to specifying a rather complete model describing
expenditures on total seafood and specific product forms (shellfish,
canned fish, whole fish, and filleted/steaks fish) in terms of
variables introduced into the equations, the analysis incorporated all
households in the survey. Furthermore, to avoid the likelihood of
biased estimates of the true parameters associated with using ordinary
least squares when a large concentration of zero observations for the
dependent variable is presented in the data, Perry estimated the
equations via a Tobit procedure. This procedure provides asymptoti
cally consistent estimates of true parameters given a correct model
specification. The variables included in the various seafood/seafood
product expenditure equations were household income, race, urbaniza
tion, expenditures on food consumed away from home, occupation of
household head, education level of household head, and household
composition. By estimating separate expenditure functions for the four

29
seafood/seafood products classifications by region and different income
groups, Perry in total estimated 85 equations explaining seafood/
seafood product expenditures. Though there are certain advantages to
estimating separate equations for different income groups, regions,
etc., the value of such a study in terms of answering national policy
questions becomes increasingly limited with increased refinements.
Given the relatively large number of equations estimated by Perry, a
discussion of the results necessitates generality. Those variables
found significant most often in explaining seafood expenditures were
income, race, and household composition. Other factors, most notably
urbanization and education, were important determinants of seafood
expenditures only in isolated instances.
The results presented by Perry were for the most part consistent
with those reported by other researchers. As in the studies conducted
by Salathe and Capps, Perry reported the income elasticities of
seafood/seafood products to be extremely inelastic. By region, the
income elasticities for total seafood expenditures ranged from a low of
0.069 in the South to a high of 0.204 in the Northeast. In terms of
specific seafood product forms, income elasticities were generally
statistically significant by region for shellfish, canned fish, and
filleted/steaks fish and insignificant for whole fish. The reported
income elasticities for shellfish expenditures were consistently higher
than that for the other individual products and ranged from a low of
0.069 in the South to a high of 0.344 in the Northeast.
In a recent study employing the 1977-78 Nationwide Food
Consumption Survey data, Haidacher et al. (1982) analyzed expenditures
and quantity demanded for total seafood, shellfish, and finfish.

30
Including all households in the analysis, the authors estimated the
various equations using ordinary least squares. As discussed in the
context of the previous studies, this estimation procedure may be
inappropriate when a large proportion of the households did not consume
the commodity. Given the large percentage of households not reporting
consumption of seafood products in the 1977-78 Nationwide Food Consump
tion Survey, caution should be used in the interpretation of the
results reported by Haidacher et al. In the analysis, the authors
expressed expenditures and quantities consumed as functions of region,
race, urbanization, income, household size, household composition,
season, and the number of guest meals. Given the similarity between
the study by Haidacher et al. and those by Capps and Perry, one would
expect the results reported in the respective studies to be similar,
which in fact was the case. The income elasticity for total seafood
expenditures as reported by Haidacher et al. equalled 0.16 which was
the same as that reported by Capps and within the range of those
reported by Perry for the various regions. Income elasticities for
shellfish and finfish expenditures were given as 0.73 and 0.03,
respectively. These estimates are in agreement with those presented by
Perry to the extent that the income elasticity of shellfish expendi
tures tended to be somewhat higher than that of finfish expenditures.
However, in absolute magnitude, the income elasticity of shellfish
expenditures reported by Haidacher et al. was two to three times the
size of that reported by Perry. Some of the discrepency in results
probably reflects the different statistical techniques employed in the
two studies.

31
In addition to the expenditure elasticities reported by Haidacher
et al., the authors also report the quantity elasticities with respect
to income. The quantity elasticities with respect to income were
positive for shellfish and negative for finfish and total seafood. The
negative quantity elasticity with respect to income for total seafood
in conjunction with a positive expenditure elasticity implies a posi
tive quality elasticity for total seafood which equalled 0.20. This
consists of a high estimate of the quality elasticity associated with
shellfish consumption (0.59) and a relatively low estimate of the
quality elasticity associated with finfish consumption (0.10).
Summarizing the research to date, evidence suggests inelastic
income expenditure and quantity elasticities for total seafood and
specific product forms. However, none of the studies conducted to
date has made complete use of all data and/or available statistical
options. An extension of the work provided by the authors discussed in
this chapter is the basis of the next two chapters. The remainder of
this chapter lays the groundwork for the models to be estimated.
Conceptual Model Development
The first task associated with specifying a cross-sectional con
sumption model involves that of defining the set of arguments compris
ing the column vector of socioeconomic characteristics, Zj, given in
equation (2.10). The concept of consumer demand in conjunction with
the seafood expenditure/quantity consumption studies discussed in the
previous section were of assistance in meeting this objective. The
respective expenditure and quantity equations were specified as a

32
function of the following set of variables (household subscripts have
been deleted for notational convenience):
EXP
= f(Zl,
Z2,
..., Z13,
M,
S)
(2.11a)
Q.
= f(zi,
Z2,
..., Z13,
M,
S)
(2.11b)
The variables EXP^ and refer to the expenditures on and quantity
consumed of total seafood or specific product form by the Jth
household, respectively. The values these two variables take range
from zero upwards, with the frequency of observed zero values varying
with respect to the specific product form.
The region of the country (Northeast, North Central, South,
West), the degree of urbanization (Central City, Suburban, Nonmetro),
and the season during which the household was interviewed (Spring,
4
Summer, Fall, Winter) are defined as Z1, Z2, and Z3, respectively.
The rationale for including these variables in (2.11a and 2.11b) is
two-fold. First, the prices associated with total seafood and the
specific product forms vary by region (Zl), urbanization (Z2), and
season (Z3) due to differences in aggregate demand and supply. As
such, households in different regions and/or levels of urbanization or
interviewed in different seasons encounter different prices for the
same product. Second, consumption of total seafood and the specific
product forms is likely to differ among households by region, urbaniza
tion, and season for reasons independent of a price effect, such as
tastes and preferences associated with cultural or institutional
factors.
^See Appendix A for a description of these and the following
variables used in this study.

33
The remaining socioeconomic variable entered into equations
(2.11a) and (2.11b), Z4-Z13, M, and S were included to specifically
account for variations in seafood consumption among different house
holds resulting from underlying socioeconomic differences which are
expected to influence tastes and preferences. The measurement of
household composition, Z4, used in this study is a modification of the
household life cycle classification proposed by Murphy and Staples
(1979) and is more directly applicable to the study than measurements
generally proposed for reasons discussed below. For purposes of this
study, households were stratified according to ten mutually exclusive
life cycle classifications: young single without children, young
married without children, young single with children, young married
with children, middle aged single without children, middle aged married
without children, middle aged single with children, middle aged married
with children, older single, and older married (where young is defined
as the head of household being less than 35 years old, middle aged is
defined as head of household being from 35 years old to 65 years old,
and elder is defined as head of household being equal to or greater
than 65 years old). There were two reasons for using the household
life cycle measurement of family composition in this study as opposed
to a more traditional measurement. First, it is useful to investigate
the reasons for the changes in apparent consumption of total seafood
and specific product forms over the past two decades. Available time
series data pertaining to household composition are related to the
family life cycle measurement of household composition more closely
than with the other measurements. Thus, the life cycle classification
was a preferable measurement of household composition in terms of

34
examining changes in seafood consumption through time. The second
reason for using the household life cycle measurement of family compo
sition is based on the hypothesis that household consumption of total
seafood and specific product forms is more directly related to the life
cycle classification of the household than to other measurements of
family composition. For example, households with young children may
avoid the purchase and consumption of specific seafood product forms
that are known to have bones. While most measures of household
composition do not consider the case in which certain households are
unlikely to consume a given commodity as a result of certain charac
teristics of the household members, the life cycle classification of
the household does to some extent account for this possibility by
stratifying households into mutually exclusive categories according to
given characteristics of the household.
The race of the household head (Z5), found in previous studies to
be of importance in explaining household consumption of seafood, was
included in the analysis to account for variations in tastes and
preferences among households of different races which would lead to
differences in at-home seafood consumption. For purposes of this study
the race of the household head was assigned to one of the categories:
White, Black, or "Other," where "Other" refers to any ethnic origin
other than that of White or Black.^
Food stamps (Z6) in essence are an additional source of income to
households which can be used for the purchase of most food items.
Similarly, a household having caught fish for its own use (Z7) has a
"Other" represents an all-inclusive term referring to households
of various ethnic origins such as Asian, Indian, etc.

35
home produced good intended for consumption. As such, there should be
a positive relationship between the catching of fish and at-home con-
6
sumption of seafood.
Employment of the meal planner (Z8), the sex of the meal planner
(Z9), family size (Z10), and the education level of the meal planner
(Zll) are expected to affect the opportunity cost of the meal planner's
time. Following Gronau's (1977) premise, an increase in the education
level of the meal planner should result in an increase in the oppor
tunity cost of his/her time, ceteris paribus. Similarly, meal planners
employed outside the home are expected to have a higher opportunity
cost of time than their counterparts. The planning of household meals
has traditionally been associated with the female members of the
household. Increases in the family size are expected to be associated
with increases in the opportunity cost of the meal planner's time,
ceteris paribus. Thus, an increase in family size is expected to
result in an increased consumption of the highly processed seafood
product forms such as canned seafood products relative to the non-
processed seafood products such as fresh seafood, ceteris paribus.
Though increases in the education level of the meal planner and
family size are expected to increase the opportunity cost of time of
the meal planner and hence result in a movement of household consump
tion patterns towards heavily processed seafood products, results
supporting this contention are likely to be masked by offsetting
factors. For example, increases in the education level of the meal
6
A price representing the market price for a similar product in a
given region and season was assigned in those cases where the seafood
product was not purchased in the market.

36
planner are likely to be associated with an increased awareness of the
nutritional value associated with consumption of seafood products.
This increased nutritional awareness and resultant increased consump
tion of seafood products are likely to offset the expected decline in
consumption associated with the increase in the opportunity cost of
time resulting from additional education. Similarly, though increases
in family size are expected to result in an increase in the opportunity
cost of time of the meal planner and hence a potential decline in
at-home consumption of seafood, increases in family size by definition
necessitates increased consumption in total. Hence the expected
decline in household consumption of seafood resulting from an increase
in the opportunity cost of time of the meal planner associated with an
increase in family size may be offset by increased consumption necessi
tated by an increase in family size.
The number of guest meals served from home food supplies (Z12),
found to be a significant factor by Capps (1982) in explaining expendi
ture on seafood consumed at home, was introduced into the analysis to
account for the expected increase in consumption of total seafood and
especially those seafood product forms most likely to be served when
entertaining guests. Since the less processed seafood product forms
are generally associated with a higher quality product and hence viewed
as more preferred items, increased weekly consumption of these product
forms is expected to be positively related with increases in the
number of guest meals.
Although Perry (1981) concluded that the money value of away-from-
home consumption was generally unimportant in explaining seafood
expenditures for at-home consumption, a similar variable was included

37
in the present analysis for two reasons. First, increases in the money
value of meals consumed away from home^ (Z13) may imply a lower need to
consume meals at home, ceteris paribus. The second reason for includ
ing the money value of meals consumed away from home is that a large
proportion of total seafood consumption reportedly occurs in the away-
from-home food market. Furthermore, the away-from-home trade as a
percentage of the total varies substantially from one seafood product
to another. For example, consumption of certain shellfish species such
as shrimp and lobster occurs largely in institutional and restaurant
outlets, while the proportion of other seafood products, such as canned
tuna, consumed in the away-from-home market is considerably less
(Vondruska, 1985). In general, it is expected that those seafood
product forms difficult to prepare at home, such as fresh seafood, are
most often consumed in the away-from-home market. To the extent that
consumption of seafood products away-from-home substitutes for consump
tion of similar products at home, the money value of meals consumed
away from home and at-home consumption of seafood are expected to be
negatively related.
The effect of income on consumption in general and on at-home
consumption of seafood in particular has been discussed extensively
throughout this chapter. In this study, before tax income (M) was
used as a proxy for the resources available to the household for the
^Excludes the value of snacks purchased and consumed away from
home.

38
purchase of seafood products. No distinction was made with respect
to the sources of income and their separate effects on consumption of
total seafood and specific product forms. Though increases in house
hold income have been found to be related to increases in expenditures
on total seafood consumed at home (e.g., Perry, 1981; Capps, 1982),
less evidence exists concerning the relationship between income and
at-home consumption of specific seafood product forms. For example,
the estimated income parameter associated with a specific seafood
product form may be either positive or negative depending upon whether
that product form is considered to be a normal or inferior good and may
in fact even vary among different segments of the population.
Substitutes for at-home consumption of seafood and the specific
product forms, denoted as S in equations (2.11a and 2.11b), follow from
the concept of the demand function provided in equation (2.4a). In
most cross-sectional studies of this nature, substitutes are omitted
from the analysis because prices of substitutes encountered by any
given household should be the same prices as those encountered by any
other household. In the estimation of the total seafood consumption
models no substitutes were specified. However, with respect to the
models for specific product forms, consumption of the alternative
seafood product forms were considered as appropriate substitutes.
For example, shellfish consumption by a given household was related to
consumption of finfish by that same household. Similarly, consumption
of fresh seafood was related to the consumption of the summation of
8
Though an arguement could be made to use total food expenditures
rather than income as an explanatory variable in the analysis, the
latter variable was used because it is more directly applicable for
answering policy oriented questions.

39
frozen and canned seafood. In a very strict sense, one might consider
that consumption of one seafood product is simultaneously related to
consumption of other seafood products by way of the budget constraint
(equation 2.1). However, given the very small proportion of the con
sumer food dollar being allocated to seafood purchases, the problem of
simultaneity with respect to these variables is probably negligable.
In fact, the Longwood Research Group Limited (1984) concluded that
heavy users of one category of seafood tended to be heavy users of
other types of seafood. This is in contrast to what one would expect
to find if in fact the budget constraint played a major role in
determining substitutability among alternative seafood product forms.
Econometric Models and Statistical Considerations
Econometric Models
The conceptual models developed in the previous section (equations
2.11a and 2.11b) were fully specified to include the actual variables
included in the estimated relationships. Incorporating these changes
and making a similar change in the notation yielded the following
weekly expenditure and at-home quantity consumption equations:
EXP.
i
= aQ + a^Xl + 02X2
+ . + + U
(2.12a)
Qi
= Bq + 8-^X1 + 82^
+ . + + ^2
(2.12b)
where
EXPi
= weekly household
of total seafood
expenditures on at-home consumption
and specific product forms
Qi
= weekly household at-home quantity consumed of
total seafood and specific product forms

40
X1-X3
X4-X5
X6-X8
X9-X17
X18-X19
X20
X21
X22
X23
region of the country in which household resides
XI = Northeastern region
X2 = North Central region
X3 = Southern region
Western region (base region)
degree of urbanization in which household resides
X4 = central city
X5 = suburban (metro)
nonmetro (base urbanization)
season during which household was interviewed
X6 = spring quarter
X7 = summer quarter
X8 = fall quarter
winter quarter (base season)
household life cycle stage
X9 = young single adult without children
X10 = young married adults without children
XII = young single adult with children
X12 = young married adults with children
X13 = middle aged single adult without children
X14 = middle aged married adults without children
X15 = middle aged single adult with children
X16 = middle aged married adults with children
X17 = elderly single adult
elderly married adult (base life cycle stage)
race of respondent
X18 = White
X19 = Other than White or Black
Black (base race)
household presently receiving food stamps (equals 1
if household is presently receiving food stamps,
0 otherwise)
household caught fish for own use (equals 1 if household
caught fish for own use, 0 otherwise)
meal planner employed outside the home (equals 1 if meal
planner employed outside the home, 0 otherwise)
sex of meal planner (equals 1 if meal planner is female,
0 otherwise)

41
X24 =
family size (total number living in household)
X25 =
family size squared
X26 =
number of years of schooling of meal planner
X27 =
number of guest meals served from household food
supply in previous 7 days
X28 =
dollar value of meals purchased and consumed away
from home (excluding snacks)
X29 =
before tax income (thousand dollars)
X30 =
before tax income squared
X31 =
interaction between before tax income and race
(X18 X29)
X32 =
interaction between before tax income and family size
(X24 X29)
X33 =
expenditures on (or quantity consumed) of alternative
product forms
l, a2, a3,
... 33 =
estimated coefficients associated with weekly
expenditure equations
81 Bo Bo
... ^33 =
estimated coefficients associated with weekly
quantity equations
U1 "
normally distributed random disturbance specific to
the expenditure equations
u2 =
normally distributed random disturbance specific to
the quantity consumed equations
Though most of the independent variables included in equations
(2.12a) and (2.12b) enter in a binary manner (X1-X23), family size
(X24), education level of the meal planner (X26), guest meals (X27),
money value of meals consumed away from home (X28), before tax income
(X29), and expenditures (or quantities) of alternative seafood product
forms (X33) enter the equations in a continuous manner. Among this
latter group of variables, family size and income were specified in a

42
quadratic form. Income was specified in a quadratic form for three
reasons. First, the quadratic specification of the income variable
allows for a declining/increasing marginal propensity to consume/
purchase with increased income. Second, the quadratic specification
of the income variable allows for a satiety level providing an upper
bound on quantity consumed yet at the same time does not restrict
expenditures to behave in a similar manner. Third, the quadratic
specification of the income variable is easily modelled. Though the
quadratic specification of the income variable provides no assurance
that the commodity in question is not purchased/consumed given an
income below some threshold value, as will be demonstrated shortly,
the statistical technique used in the analysis does associate a lower
income with a lower probability of purchasing/consuming the commodity
Family size was introduced into the analysis in a quadratic specifica
tion to account for possible economies to scale in the purchasing and
consumption of seafood and specific product forms associated with
increased family size.
Two interactions (X31, X32) were introduced into the analysis.
The first interaction, that between White households and the linear
income term, allows for differences in the marginal propensity to
purchase/consume total seafood and specific product forms among house
holds of different races. Similarly, the interaction between the
linear family size term and linear income term allows for differences
in the marginal propensity to purchase/consume among households of
different sizes. An argument could probably be made for the intro
duction of other interaction terms, in addition to the two specified.
However, the use of too many interaction terms would have probably

A3
resulted in severe collinearity problems among the regressors.
Therefore, those used were only those considered most appropriate.
Statistical Considerations
The model developed in the previous section can be expressed in
matrix form as follows:
y
y
t
t
where,
yt
xt
ut
xte + ut if xte + ut > o
o if xt e + ut _< o
t = 1, 2, . N
dependent variable
vector of independent variables
2
error term assumed iid N(0, o )
(2.13)
This model specification referred to as the Tobit model after its
founder is well known and used extensively in economic studies of a
cross-sectional nature (see Amemiya (1984) for a good review of the
basic model and its uses). Given the specification in (2.13), an
assumption is implicitly made that an underlying stochastic index equal
to Xtg + U is observed only when strictly positive. In other words,
yt will only be positive given a value of XtB + Ut greater than zero.
Otherwise, yt will equal zero. For example, assume two households with
identical attributes with the exception of income. Furthermore, assume
that the household with the higher income consumed seafood while the
household with the lower income did not consume seafood. This would
imply that the first household with the higher income had exceeded that
threshold level required to consume seafood (i.e., Xt8 + Ut > 0), while

44
the second household with the lower income had not crossed that
threshold level (i.e., XtB + Ut < 0). Factors such as those in Xt
probably influence at-home consumption of seafood and thus the Tobit
procedure is appropriate for this analysis.
As shown by Greene (1981), OLS estimates of (2.13) are both biased
and inconsistent due to the non-normality of the expected error terms.
Thus, some estimated procedure other than that of OLS must be used if
unbiased or at least consistent parameter estimates are to be obtained.
Since the original work by Tobin (1958), several methods of estimating
equation (2.13) which assure consistent estimates of the true
parameters have been developed and used (see Amemiya (1984) for a
discussion of the different methods). Since the different methods are
widely known and should in all cases provide the same parameter esti
mates assuming a unique maximum, the different approaches to estimating
(2.13) will not be discussed.
Though the uses and methods of estimation of the Tobit model are
well known and documented, less well known is the amount and types of
information that can be obtained from the Tobit estimates. The types
of information that the Tobit model provide are discussed here and
used extensively in the next two chapters. The uses of the Tobit
model, first presented by McDonald and Moffitt (1980), are the basis
for the ensuing discussion with some modifications to their work added
towards the end of the discussion. Those equations which will be
modified are assigned the letter (a) after the numerical numbering.
The modified equations are assigned the letter (b).
The unconditional expected value of the dependent variable in
equation (2.13) was shown by Tobin (1958) to equal

45
E(y)
= XBF(Z) + of(Z)
where
Z
= n/o
f(Z)
9
= unit normal density function
F(Z)
= cumulative normal distribution function
(2.14a)
The unconditional expected value of the dependent variable represents
the expected value of the dependent variable associated with all
observations. Furthermore, as shown by Amemiya (1973), the conditional
expected value of the dependent variable for observations above the
limit (i.e., positive observations), y* is given by
E(y*) = E(y|y > 0)
= XS + E(u|y > 0) (2.15a)
= X3 + of(Z)/F(Z)
The relationship between the unconditional expected value of the
dependent variable (expected value associated with all observations)
and that of the conditional expected value of the dependent variable
(expected value associated with positive observations) is given as
follows:
E(y) = E(y*) F(Z)
(2.16a)
Defined as
10
Defined as
_1_ -Z2/2
2 it 8
B'X
L 2
o 1 -Z 12
- e

46
Thus, the unconditional expected value of the dependent variable is
equal to the conditional expected value of the dependent variable
adjusted by the probability of observing a positive value of the
dependent variable, F(Z). Differentiating this relationship with
respect to an exogenous variable, X, gives the effect of a change in
the dependent variable resulting from a change in X.
3E(y)/3X. = F(Z) (3E(y*)/3X) + E(y*) (3F(Z)/3Xi) (2.17a)
Furthermore, it can be shown that the two partial derivatives on the
right-hand side of equation (2.17a) are equal to
3F(Z)/3X. = i{Z)?>/o (2.18a)
and
3E(y*)/3Xi = + (a/F(Z)) 3f(Z)/3Xi (2.19a)
- (of(Z)/F(Z)2) 3F(Z)/3Xi
which upon reduction yields
3E(y*)/3X = 6 [1 Zf(Z)/F(Z) f(Z)2/F(Z)2J (2.20a)
i i
Finally, note that after substitution of (2.20a) and (2.18a) into
equation (2.17a) and upon rearrangement of the terms, one arrives at
the following expression:
SE^/SXi = F(Z)3i (2.21a)
a much simplier expression than that of equation (2.13).
For purposes of this study, a modification of the above equations
was necessary. Notably, the expressions given above are based on the
assumption that all of the variables in the vector Xt are continuous

47
in nature. However, several of the variables evaluated in the present
study were binary in nature (X1-X23). The main consequence of such
a specification pertains to the evaluation of X3 given a binary
variable. Specifically, the probability of observing a positive
value of the dependent variable now becomes conditional on the binary
variable being evaluated. Thus, the value of the standard normal,
(Xg|Xi)/a, and hence the value of the cumulative normal density
function F(Z |X^) and the value of the unit normal density function
f(Z |X-^) all become dependent on the binary variable being evaluated.
Hence, equations (2.14a) through (2.21a) need to be adjusted accord
ingly when discussing binary variables. These adjustments are
E(y|x) =
(X6|X) FiZjX^ + ofiZjXi)
(2.14b)
E(y*|X) =
(XS|X.) + af (zi|xi)/F(zi| x^
(2.15b)
E(y |X) =
E(y*|X) F(ZilXi)
(2.16b)
8E(y|Xi)/8Xi =
F(ZilXi) OEirHx^/aXi)
+ E(y*|Xi) (3F(Zi|xi)3Xi)
(2.17b)
9F(Zi|Xi)/3Xi =
f (Zi|Xi) 3i/ 0
(2.18b)
=
Si + (a/F(Zi|Xi)) afiZilXii/aXi
- (afZilX^/FZilXi)2) BFCZilX^/SXi
(2.19b)
3E(y*|Xi)/3Xi =
B.[l (Z i |X i) f (Z i|X i)/F(Z i|Xi)
- f(Zi|Xi)2/F(Zi|Xi)2J
(2.20b)
3E(yfX.)/3X. =
FiZilxpSi
(2.21b)
Technically, it would be preferable to use the concept of a
limit rather than a derivative in evaluating equations (2.17b) through
(2.21b) due to the discrete nature of Xi in each of the equations.

48
For evaluation of the continuous variables in the model, use of equa
tions (2.14a) through (2.21a) is valid.
The total change in the dependent variable y given a change is
as specified in equation (2.17a) can be broken into two components.
The first component [F(Z)( 3E(y-55")represents the change in the
expected value of the dependent variable if above the limit (positive)
weighted by the probability of being above the limit (positive). The
second component [Eiy^KBFiZi/sX^) ] represents the change in the prob
ability of being above the limit (positive) weighted by the expected
value of the dependent variable if above. The breakdown of the total
change of the dependent variable into these two components is probably
very consistent with the actions of consumers in the market and is thus
useful in studying the consumption patterns of households.
It is interesting to note the simularity between the Tobit
estimates and OLS estimates. From equation (2.21a) it can be observed
that the effect of a change in X^ on the dependent variable y is equal
to only when F(Z) is equal to one. This is in fact what one would
expect since as F(Z) approaches one, OLS estimates should be obtained.
Multiplying equation (2.17a) by X^/E(y) gives the Tobit
elasticity, n^, which equals
[ F( ^
(2.22)
i
Substituting equation (2.16a) for E(y) and making the appropriate
reductions provides the following specification of the elasticity of y
with respect to Xi:

49
, 9Xi E(y*). 3F(Z)
ni E(y*) ^ 9X )+ F(Z) ( 9Xi *
(2.23)
The total elasticity is comprised of two components. The first compo
nent measure the conditional elasticity associated with the nonlimit
observations. The second term measures the elasticity of the prob
ability of participation associated with a change in X^.
Data Source and Considerations
The 1977-78 Nationwide Food Consumption Survey (NFCS) provides the
data used in the analysis. This is the most recent of the household
food consumption surveys periodically conducted under the auspices of
the United States Department of Agriculture. The survey encompassed
approximately 15,000 households throughout the 48 continguous states
and contained detailed information on characteristics for each
household. In addition, the survey contained detailed information
pertaining to expenditures on and the corresponding quantities of a
continum of foods consumed at home (measured at the level at which the
foods came into the kitchen) by each of the households surveyed. The
survey was conducted over a one year period (April 1977 through March
1978) and was stratified according to a variety of factors including
season, region, and urbanization in an attempt to have the sample
represent the universe of households in the continental United States
as accurately as possible. Though information on 14,930 households
was provided on the original NFCS data tapes distributed through the
Department of Commerce's National Technical Information Service, only
10,689 observations of the original 14,930 were retained for the
current analysis. Of the 4,241 deleted observations, approximately 92

50
percent were deleted due to the household's refusal to report annual
income. The remaining 8 percent were deleted due to the omission of
other relevant information. In light of the fact that households
reporting incomplete income information often provide poor or incom
plete expenditure information (Buse, 1979), it was deemed appropriate
to delete these observations. While omission of those observations for
which information is missing could potentially lead to sampling bias
(Maddala, 1977), a comparison of those households not reporting income
with those reporting income indicates that this was not a serious
12
problem in the present study.
Among those households not reporting income, 50.2 percent of the
households consumed seafood at home during the one week interview
period compared to 50.8 percent among those households reporting
income. Average weekly consumption among those households consuming
seafood and not reporting income equalled 1.94 pounds compared to 1.91
pounds among those households reporting income.

CHAPTER III
TOTAL SEAFOOD ANALYSIS
The discussion of the results associated with the total seafood
models is given in two sections. First, descriptive statistics
comparing/contrasting consumers and nonconsumers of seafood as
established by the survey data are presented and briefly discussed.
In the second section, the results of the Tobit analysis associated
with the total seafood models are presented and discussed.
Descriptive Statistics Associated with
Total Seafood Analysis
Descriptive statistics of the 10,689 households included in the
analysis are presented in Table 3-1. Though the information presented
in this table is of a descriptive nature without any attempt to
separate partial effects, the information does provide an overall
comparison of those households which consumed seafood during the one
week survey period versus those households which did not.
The statistics provided in Table 3-1 are assigned to one of four
categories. The first category, labelled total sample, gives the mean
values of all variables used in the analysis. For example, as
indicated in the table, 24.8 percent of the households in the analysis
resided in the Northeast region (XI). The second category, labelled
nonlimit observations, provides the mean values of all variables for
those households consuming seafood at home. For example, of those
51

Table 3-1. Descriptive statistics of variables in total seafood models
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Mean (percent)
Region:
Northeastern (XI)
0.248
0.299
0.195
0.612
North Central (X2)
0.241
0.213
0.270
0.449
Southern (X3)
0.342
0.305
0.379
0.453
Western (base)
0.169
0.183
0.156
0.550
Urbanization:
Central City (X4)
0.305
0.330
0.280
0.550
Suburban (X5)
0.355
0.374
0.335
0.535
Nonmetro (base)
0.340
0.296
0.385
0.442
Season:
Spring (X6)
0.238
0.233
0.243
0.497
Summer (X7)
0.232
0.234
0.229
0.512
Fall (X8)
0.268
0.259
0.278
0.491
Winter (base)
0.262
0.274
0.250
0.531

Table 3-1. Continued
(1)
(2)
(3)
(4)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Mean (percent)3
Household life cycle:
Young single w/o children (X9)
0.055
0.045
0.065
0.416
Young married w/o children (X10)
0.058
0.054
0.061
0.473
Young single with children (Xll)
0.035
0.035
0.034
0.508
Young married with children (X12)
0.151
0.158
0.143
0.532
Middle aged single w/o children (X13)
0.075
0.063
0.088
0.427
Middle aged married w/o children (X14)
0.115
0.119
0.111
0.526
Middle aged single with children (X15)
0.056
0.062
0.050
0.562
Middle aged married with children (X16)
0.261
0.299
0.223
0.582
Elderly single (X17)
0.099
0.073
0.126
0.375
Elderly married (base)
0.095
0.092
0.099
0.492
Race of respondent:
White (X18)
0.852
0.831
0.874
0.495
Other (X19)
0.030
0.035
0.024
0.592
Black (base)
0.118
0.134
0.102
0.577

Table 3-1. Continued
(1)
(2)
(3)
(4)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Mean
(percent)
a
Receives food stamps:
Yes (X20)
No (base)
Caught fish for own use:
Yes (X21)
No (base)
Employment of meal planner:
Yes {122)
No (base)
Sex of meal planner:
Female (X23)
Male (base)
0.074
0.926
0.077
0.923
0.234
0.766
0.264
0.736
0.465
0.535
0.454
0.546
0.908
0.092
0.932
0.068
0.071
0.929
0.529
0.506
0.202
0.798
0.573
0.488
0.476
0.524
0.496
0.518
0.882
0.118
0.521
0.370

Table 3-1. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Family size:
Total number in household (X24)
2.946
Actual mean
3.153
values
2.733
Total number squared (X25)
11.457
12.803
10.067

Education of meal planner:
Years (X26)
11.732
11.941
11.517
Guest meals:
Number of meals (X27)
1.141
1.256
1.021
Meals away from home:
Dollars (X28)
11.658
11.605
11.714

Income before taxes:
Thousand dollars (X29)
14.109
15.080
13.107
Thousand dollars squared (X30)
324.590
357.590
290.516

y (d b
cM a ^
V
Ln
Ln

Table 3-1. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Interaction terms:
Income and race (X31)
12.745
Actual mean
13.394
values
12.075
Income and family size
(X32)
46.628
52.460
40.601

Weekly expenditures and
EXP
quantity:
1.487
2.927
0.000
Q
0.971
1.911
0.000

Number of households:
10,689
5,430
5,259

Percent of households:
100
50.8
49.2

The data provided in this table associated with the binary variables (X1-X23) should be interpreted as
representing proportions rather than percentages. To obtain percentages, multiply data by 100.

57
f
^fa
s'
households consuming seafood at home, 29.9 percent resided in the
Northeast region. The third category, labelled limit observations,
provides the mean value of all variables for those households reporting
* ^
a zero level of at-home consumption of seafood during the one week
survey period. On this basis, the information contained in Table 3-1
suggests that of those households reporting no at-home seafood consump
tion 19.5 percent resided in the Northeast region. The final category
in Table 3-1, labelled proportion consuming, gives the value of the
proportion of households consuming seafood during the survey period
associated with each of the binary variables. For example, 61.2
percent of the households residing in the Northeast region of the
United States consumed seafood during the interview period.
Approximately one half (50.8 percent) of the 10,689 households
included in the analysis consumed seafood at home during the survey
period. Among households consuming seafood at home, average expendi
tures and consumption of seafood equalled $2.93 and 1.91 pounds,
respectively. For the total sample, average at-home weekly consumption
of seafood was 0.97 pounds valued at $1.49 or approximately one-half
the volume and value estimated for consumers only. Placed on a yearly £>JL
basis, at-home consumption of seafood by an average household thus &''
1/7
P
/
equals just over 50 pounds, or about 18 pounds per capita assuming an
( average of 2.75 members per household. This value equals about one-
half of the approximately 35 pounds (round weight) annual reported per
capita consumption of seafood in the United States during the 1975-77
period.
With respect to region, households in the Northeastern region of
the United States (XI) had a higher probability of consuming seafood at

58
home than did those households in either the Northcentral region (X2),
Southern region (X3), or Western region (base). Over 61 percent of the
Northeastern households consumed seafood at home compared with 44.9
percent, 45.3 percent, and 55 percent of the households residing in the
Northcentral, Southern, and Western regions, respectively.
Little variation in the proportion of households consuming seafood
at home was observed across seasons. Though a slightly larger propor
tion of households consumed seafood at home during the summer (X7) and
winter (base) quarters than in either the spring (X6) or fall (X8)
quarters, the differences would probably not be significant if a
statistical test were to be conducted.
An examination of the different family life cycles (X9-X17)
revealed that households with children were more likely to consume
seafood at home than those households without children.^ On average
(weighted), 55.7 percent of the households with children consumed
seafood compared to 45.3 of those househoolds without children.
Similarly, at-home seafood consumption tended to increase in prob
ability with increased family size (X24) which generally reflects
increased number of children.
The race of the household respondent (X18, X19) appears to be an
important consideration in determining the probability of at-home
seafood consumption. White households (X18) had a significantly lower
observed probability of at-home seafood consumption than that of either
Black households (base) or households of other ethnic origins (X19).
M
4
It was assumed (though not verified) that elderly households
(X17, base) had no children living at home.

59
Households who caught fish for their own use (X21) had a much
higher probability of consuming seafood at home than did households not
reporting fish catches (base). Among those households who caught fish
for their own use, 57.3 percent reported consuming fish at home during
the interview period. This figure compares to only 48.8 percent among
those households who did not catch fish.
The employment status of the meal planner (X22) and the sex of
the meal planner (X23) had the anticipated effects on consumption of
seafood at home. Employment of the meal planner leads to a slight
decrease in probability of at-home seafood consumption. A female meal
planner (X23), on the other hand, appears to greatly enhance the
expectancy of that household consuming seafood at home.
With respect to the continuous variables included in the models
(X24, X26, X27, X28, X29), differences in the mean values between
consumers and nonconsumers for a given variable should indicate
increased (decreased) probability of at-home seafood consumption with
respect to that variable. For example, the average family size (X24)
of seafood consumers (3.153) greatly exceeds that of nonconsumers
(2.733). Similarly, the mean values for consumers with respect to
education (X26), number of guest meals (X27), and annual income before
taxes (X29) exceed the mean values for nonconsumers. Mean values for
continuous explanatory variables in the analysis are larger for con
sumers than nonconsumers with the exception of expenditures on meals
away from home. Large differences between the mean values of the non
consumers and consumers are especially apparent with respect to family
size (X24), number of guest meals (X27), and annual income before taxes
(X29). Thus, one would expect the probability of at-home seafood

60
consumption to increase significantly with changes in the value of
these variables.
Regression Estimates of Total Seafood Analyses
Notwithstanding the general usefulness of the descriptive statis
tics just presented, there are several inherent weaknesses associated
with these types of statistics. The foremost weakness associated with
these types of statistics is that they do not control for confounded
effects among different variables. Thus, one cannot separate the
effect of one exogenous variable from that of another when examining
changes in the dependent variable. For example, the relatively low
probability of seafood consumption among households without children
when compared to those with children may be due to some factor such as
larger expenditures on meals consumed away from home among the former
group of households. The Tobit regression parameters presented in this
section can be considered as partial effects in that the confounded
effects among exogenous variables have been controlled for.
J
Results of the Tobit analysis relating to total at-home seafood
2
consumption are presented in Tables 3-2 and 3-3. The first column
gives a listing of the variables used in the analysis. The second
column in each table gives the Tobit parameter estimates associated
with each of the exogenous variables. The asymptotic t-values asso
ciated with the parameter estimates are presented in the third column.
The relatively large sample size employed in this study should assure
that the asymptotic t-values are representative of the true values.
2
The Tobit model used for this analysis was developed by the Rand
Corporation and is referred to as LIMDEP. Documentation of the model
is given by Phelps (1972).

Table 3-2. Summary statistics for Tobit analysis of weekly household expenditures on total seafood0
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
3E(EXP)
aXi
t\)
Expected change
among consuming units
3E(EXP*) F(Z-¡)
9Xi
(1)
(2)
(3)
(4)
(5)
(6)
Region:
Northeastern (X1
0.8147
4.903
(. 5469^)
0.4456
0.1738
North Central (X2)
-0.9383
-5.412
.4061
-0.3810
-0.1216
Southern (X3)
-0.3578
-2.146
.4523
-0.1618
-0.0547
Western (base)


.4812


Urbanization:
Central City (X4)
1.0843
7.507
.5198
0.5636
0.2104
Suburban (X5)
0.4888
3.628
.4717
0.2306
0.0802
Nonmetro (base)


.4322


Season:
Spring (X6)
0.0154
0.101
.4771
0.0073
0.0026
Summer (X7)
0.1428
0.937
.4873
0.0696
0.0248
Fall (X8)
-0.1414
-0.959
.4644
-0.0657
-0.0226
Winter (base)


.4759



Table 3-2. Continued
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
3E(EXP)b
SX
Expected change
among consuming units
3E(EXP#) F(Zi)
9Xi
(1)
(2)
(3)
(4)
(5)
(6)
Household life cycle:
Young single w/o
children (X9)
-0.6320
-1.803
.4605
-0.2910
-0.0996
Young married w/o
children (X10)
-0.6892
-2.292
.4559
-0.3142
-0.1068
Young single with
children (X11)
-1.2180
-3.235
.4137
-0.5039
-0.1604
Young married with
children (X12)
-1.0171
-3.853
.4297
-0.4370
-0.1428
Middle aged single
w/o children (X13)
-0.5456
-1.758
.4674
-0.2550
-0.0884
Middle aged married
w/o children (X14)
0.1697
0.698
.5252
0.0891
0.0336
Middle aged single
with children (X15)
-0.0579
-0.188
.5069
-0.0293
-0.0108
Middle aged married
with children (X16)
-0.3747
-1.437
.4812
-0.1567
-0.0637
Elderly single (X17)
-0.5676
-1.976
.4656
-0.2643
-0.0911
Elderly married (base)


.5116


Table 3-2. Continued
Category
Parameter Asymptotic
estimates t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in
3E(EXP)b
SXi
Expected change
among consuming units
9E(EXP'::~) ?{Z)
3Xi
(1) (2) (3)
Race of respondent:
White (X18) -1.2718 -4.635
Other (X19) -0.3428 -0.988
Black (base)
Receives food stamps:
Yes (X20) 0.2103 0.870
No (base)
Caught fish for own use:
Yes (X21) 1.3842 10.723
No (base)
Employment of meal planner:
-0.964
(4) (5) (6)
.4504 -0.8553 -0.2883
.6051 -0.2074 -0.0883
.6061
.4915 0.1034 0.0371
.4745
.5580 0.7724 0.3084
.4464
.4705 -0.0568
.4803
Yes (X22)
No (base)
-0.1207
-0.0199

Table 3-2. Continued
Category
Parameter
estimates
a.
i
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
9E(EXP)b
BXi
Expected change
among consuming units
9E(EXP"-) .F(Zi)
9Xi
(1)
Sex of meal planner:
(2)
(3)
(A)
(5)
(6)
Female (X23)
0.5679
2.425
.4790
0.2720
0.0958
Male (base)
Family size:
Total number in
.4417
household (X24)
Total number
0.4689
2.721
.4759
0.2062
0.0722
squared (X25)
Education of meal planner:
-0.0169
-1.036
Years (X26)
Guest meals:
0.0858
3.982
.4759
0.0408
0.0143
Number of meals (X27)
0.1649
7.831
.4759
0.0785
0.0275

Table 3-2. Continued
Category
Parameter
estimates
a.
i
Asymptotic
t-ratio
Expected
probability
F(Zi)
Expected total
change resulting
from change in
3 E(EXP)b
3X.
Expected change
among consuming units
3E(EXP*) F(Z.)
9Xi
(1)
(2)
(3)
(4)
(5)
(6)
Meals away from home:
Dollars (X28)
-0.0120
-3.598
.4759
-0.0057
-0.0020
Income before taxes:
Thousand dollars (X29)
Thousand dollars
0.0817
3.572
.4759
0.02518
0.0088
squared (X30)
-0.0003
-2.619
Interaction terms:
Income and race (X31)
Income and family
-0.0416
-2.086



size (X32)
0.0047
1.577




Table 3-2. Continued
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
8E(EXP)b
8X
Expected change
among consuming units
9E(EXP*) F(Z<)
aXi
(1)
(2)
(3)
(A)
(5)
(6)
Other seafood:
Dollars (X33)
NAC
NA
NA
NA
NA
Constant:
a0
-3.2808
-5.902



a
The value of X$ at
the means of all X^'
s is equal
to -0.29854; o
= 4.93351.
^The effects of the interaction and/or squared terms have been accounted for in the construction of the
linear terms associated with those variables.
c
Not applicable.

Table 3-3. Summary statistics for Tobit analysis of weekly household quantity consumption of total seafood
Category
Parameter
estimates
ei
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
3E(Q)b
3X.
Expected change
among consuming units
3E(Q*) F(Z )
ax.
(1)
(2)
(3)
(4)
(5)
(6)
Region:
Northeastern (XI)
0.5888
4.382
.5010
0.2950
0.1073
North Central (X2)
-0.5883
-5.197
.3842
-0.2260
-0.0688
Southern (X3)
-0.0060
-0.044
.4414
-0.0026
-0.0009
Western (base)


.4420


Urbanization:
Central City (X4)
0.7393
6.343
.4807
0.3553
0.1256
Suburban (X5)
0.4234
3.895
.4490
0.1901
0.0639
Nonmetro (base)


.4071


Season:
Spring (6)
0.1739
1.417
.4535
0.0789
0.0267
Summer (7)
0.2623
2.132
.4623
0.1213
0.0416
Fall (X8)
0.0215
0.181
.4383
-0.0094
0.0031
Winter (X9)


.4361



Table 3-3. Continued
Category
Parameter
estimates
Bi
Asymptotic
t-ratio
Expected
probability
F(Zt)
Expected total
change resulting
from change in Xj
3E(Q)b
3Xi
Expected change
among consuming units
3E(Q*) F(Z^)
aXi
(1)
(2)
(3)
(4)
(5)
(6)
Household life cycle:
Young single w/o
children (X9)
-0.5569
-1.967
.4255
-0.2370
-0.0770
Young married w/o
children (X10)
-0.5615
-2.312
.4251
-0.2387
-0.0774
Young single with
children (Xll)
-0.9970
-3.289
.3825
-0.3814
-0.1157
Young married with
children (X12)
-0.7776
-3.653
.4038
-0.3140
-0.0988
Middle aged single
w/o children (X13)
-0.4500
-1.797
.4630
-0.2084
-0.0862
Middle aged married
w/o children (X14)
0.1682
0.858
.4980
0.8038
0.0303
Middle aged single
with children (X15)
-0.0317
-0.128
.4780
-0.0152
-0.0054
Middle aged married
with children (X16)
-0.2881
-1.369
.4522
-0.1303
-0.0440
Elderly single (X17)
-0.4718
-2.036
.4777
-0.2254
Elderly married (base)


.4826
___

Table 3-3. Continued
Category
Parameter
estimates
*i
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X.-
3E(Q)b 1
aXi
Expected change
among consuming units
3E(Q*) F(Z)
ax
(1)
(2)
(3)
(A)
(5)
(6)
Race of respondent:
White (X18)
-1.3990
-6.344
.4181
-0.7838
-0.2516
Other (X19)
-0.7943
-2.839
.5263
-0.4180
-0.1579
Black (base)


.6051


Receives food stamps:
Yes (X20)
0.2263
1.164
.4677
0.1058
0.0366
No (base)


.4457


Caught fish for own use:
Yes (X21)
1.2961
12.451
.5426
0.7033
0.2724
No (base)


.4129


Employment of meal planner:
Yes (X22)
-0.0252
-0.249
.4455
-0.0112
-0.0037
No (base)


.4480



Table 3-3. Continued
Category
Parameter
estimates
^i
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
9E(Q)b
3Xi
Expected change
among consuming units
3E(Q*)
3Xi
F(Z.)
(1)
Sex of meal planner:
(2)
(3)
(4)
(5)
(6)
Female (X23)
0.4437
2.344
.4499
0.1996
0.0672
Male (base)
Family size:
Total number in
.4060
household (X24)
Total number
0.3759
2.710
.4469
0.1579
0.0530
squared (X25)
Education of meal planner:
-0.0117
-0.896
Years (X26)
Guest meals:
0.0662
3.809
.4469
0.0269
0.0099
Number of meals (X27)
0.1201
7.069
.4469
0.0037
0.0180

Table 3-3. Continued
Expected total
Parameter Asymptotic
Expected
change resulting
Expected change
Category
estimates t-ratio
probability
from change in X.¡
among consuming units
ei
F(Zt)
3E(Q)b
3E(0*) FCZi)
3X
3Xt
(1) (2)
Meals away from home:
Dollars (X28) -0.0131
Income before taxes:
Thousand dollars (X29) 0.0487
Thousand dollars
squared (X30) -0.0002
Interaction terms:
Income and race (X31) -0.0295
Income and family
size (X32) 0.0034
(3)
(4)
(5)
-4.837
.4469
-0.0059
2.637
-2.146
.4469
0.0129
-1.839
1.411


(6)
-0.0020
0.0043
1.411

Table 3-3. Continued
Category
Parameter
estimates
*1
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in Xn-
9E(Q)b
Expected change
among consuming units
9E(Q*) F(Z.)
axt
(1)
(2)
(3)
(4)
(5)
(6)
Other seafood:
Pounds (X33)
NAC
NA
NA
NA
NA
Percent of households:
*0
-2.5837
-5.772



aThe value of Xg at the
means of all X^'
's is equal
to -0.52941; a
= 3.96192.
bThe effects of the interaction and/or squared terms have been accounted for in the construction of the
linear g^ terms associated with those variables.
Not applicable.

73
In column 4 of each table, the vlue of the cumulative normal
distribution function associated with each variable is presented.
The value of this function varies with each of the discrete variables
due to variations in E(X3|X). For discrete variables in the models,
the values provided for the cumulative normal distribution function are
interpreted as the expected probability of observing a positive level
of the dependent variable given the occurrence of X^, holding all non-
mutually exclusive variables at their mean levels. Mutually exclusive
variables are set equal to zero. For example, to determine the
expected probability of seafood consumption by average Northeastern
households, the value of XI is set equal to one, the values for X2 and
X3 are set equal to zero, and the values for all remaining variables
are set at their respective means. For continuous variables, the value
provided for the cumulative normal distribution function represents the
probability of observing a positive level of the dependent variable
given the mean values for all exogenous variables. Of course, the
value of the cumulative normal distribution function varies with
changes in the level of exogenous variables^
Multiplication of the appropriate parameter estimates, given by
those values in column 2 by their respective expected probabilities of
occurrence (column 4) provides the unconditional or total expected
change in the dependent variable due to a change in X^. These esti
mates are given in column 5 of Tables 3-2 and 3-3. The unconditional
or total effect of a change in expenditures or quantity consumed with
respect to a change in the independent variable X^ can be decomposed
into two parts. The first part represents the change in the value of
expenditures or quantity consumed among existing consumers weighted by

74
the probability of being a consumer. The second part represents the
change in the probability of being a consumer weighted by the expected
value of expenditure or quantity consumed among consuming households.
The values for the first component, the conditional expected change in
expenditures or quantity consumed resulting from a change in as
derived in equations (2.20a) and (2.20b), are provided in column 6.
Thus, by definition the values for the second component of the total
or unconditional change can be obtained by subtracting the values in
column 6 from those in column 5.
The parameter estimates associated with the at-home total seafood
expenditure and consumption equations appear satisfactory and reason- ^
able based on several criteria. First, the estimates, with few
exceptions, conform to either theoretical expectations and/or results
of previous research. Second, the relative magnitudes of the estimated
(
probabilities of observing a positive level of seafood consumption
associated with each of the categories of binary variables are for the
most part in agreement with the observed probabilities given in column
5 of Table 3-1. Finally, a high proportion of the parameter estimates,
for a cross-sectional study of this nature, were statistically signifi
cant (at a 10 percent significant level) in explaining weekly household
consumption of seafood at home. More detail is given to these factors
in the discussion of the individual explanatory variables.
Before providing an in depth discussion of the results, a few
general findings are discussed here. First, with few exceptions, the
results associated with the expenditure model were found to be con
sistent with the results pertaining to the quantity consumed model.
Second, as indicated by the results of both models, the change among

75
consuming households resulting from a change in consistently
averaged from 30-40 percent of the total change. This implies that
approximately 35 percent of the change in at-home consumption of
seafood with respect to a change in X-¡^ is due to increased/decreased
consumption of seafood by those households currently consuming seafood
as opposed to entry or exit among households. Thus, approximately 65
percent of the total change is attributable to entry/exit into (from)
a fy*
the at-home seafood market by households. This finding has significant )
implications to the seafood industry and its support groups who would
r
like to know the relative merits and associated costs of increasing
at-home seafood consumption by enticing new consumers into the
at-home seafood market as opposed to increasing consumption among
currently consuming households.
Region
Total weekly expenditures on seafood consumed at home were esti
mated to be highest among the households residing in the Northeastern
region of the United States (XI) (Table 3-2) which is consistent with
the results presented by Perry (1981) and Capps (1982). Similarly,
total weekly quantity of seafood consumed by households residing in the
Northeastern region was estimated to be higher than that for other
regions (Table 3-3). Based on the information provided in column 5 of
Table 3-2, the total expected expenditures on seafood consumed at home
for a household in the Northeastern region were estimated to exceed
those for a household in the base region (West) by $0.45. Similarly,
expenditures among Northeastern households were estimated to exceed
expenditures among North Central households (X2) and Southern

households (X3) by $0.83 and $0.61, respectively. The information
provided in column 5 of Table 3-3 suggests the total expected weekly
consumption of seafood by households in the Northeastern region
exceeded that of households in the North Central region, Southern
region, and Western region by 0.52 pounds, 0.30 pounds, and 0.30
pounds, respectively, ceteris paribus.
The relatively high estimates associated with at-home seafood
expenditures for a household in the Northeastern region compared to
households in other regions of the United States are the result of two
factors. First, the estimated probabilities of a household having
positive weekly consumption expenditures and quantities, given in
column 4 of Tables 3-2 and 3-3, exceed those associated with any other)
region. Second, the expected expenditures among consuming households
residing in the Northeast exceeded those of households in other
regions, as noted in the last column of Tables 3-2 and 3-3.
Given the estimated differences in at-home seafood expenditures
among households residing in the different regions, establishing
probable causes for these differences may be beneficial to the seafood
industry and its support groups. Traditionally, the Northeast region
has had an extensive fishing industry. This factor, in conjunction
with the close proximity of most of the Northeastern States to the
ocean, has resulted in a relatively steady supply of fresh seafood to
the households in this region. Transporting fresh seafood products
from coastal states to the inland states, such as many of those located
in the North Central region, is risky and expensive. O'Rourke (1977)
categorizes the U.S. seafood marketing system into two distinct
segments. The first segment, defined by O'Rourke (p. 239) as the

77
coastal fisheries "... are exploited by small, ill-equipped or part-
time fishermen, delivering their product to dockside warehouses for
fresh distribution to communities within a 50-mile radius." O'Rourke
further claims that the New England fisheries deliver much of their
specialty catches in this manner. Because of this factor "... large
areas of the continental U.S. have below average consumption of many
fish and shellfish species." The second segment which may account
for two-thirds of the U.S. consumption of seafood is comprised of
". . large, highly capitalized canners or prepackers selling
standardized, breaded, and heavily promoted products through nationwide
retail chains or institutional outlets." The relative unavailability
and expense of certain fresher seafood products in the North Central
region probably explains to some extent the relatively low estimates
of at-home seafood expenditures and consumption in this region compared
to the other regions of the United States. Furthermore, one would
expect regional differences to be relatively large for fresh seafood
products and somewhat less for the more processed seafood products sold
either frozen or canned. The validity of this hypothesis is examined
in the next chapter when results pertaining to the seafood product
forms are analyzed.
Given the differences and probable causes for these differences as
relating to at-home consumption of seafood, what are their implications
to the seafood industry and its support group? Most obvious and
probably the most important from a policy standpoint lies in the
jrC
realization that the North Central region provides a relatively large
and untapped source by which to increase national at-home seafood
consumption. To this extent, it may be beneficial to the seafood
/

78
industry and the support groups to look for ways of reducing costs and
preserving the freshness of seafood when transporting fresh seafood to
the inland regions of the country. Unfamiliarity with many seafood
products among North Central households may also help to explain the
differences in the at-home seafood consumption patterns across regions.
Thus, promotion aimed at familiarizing these households with the dif
ferent available products may prove useful in the short run. In the
long run, as the regional structure of the population shifts due to
increased population mobility and the transportation of fresh seafood
products becomes more economically feasible due to improved methods of
preserving the freshness of seafood, regional differences in seafood
consumption will probably decline naturally. For example, between 1950
and 1980, the proportion of the U.S. population living in the Northeast
and Midwest sections of the United States declined by 17 percent and 12
percent, respectively, while the proportion of the U.S. population
residing in the South and West increased by 7 percent and 44 percent,
respectively (United States Department of Commerce, Bureau of the
Census, 1984). Additional declines in the Northeast and Midwest
sections of the country are expected until at least the year 2000.
With these demographic shifts in population should come an exchange of
knowledge among households concerning different types of seafood and
methods of preparation which may eventually lead to the disappearance
of regional differences in at-home seafood consumption patterns.
Urbanization
Households in the central city (X4) were estimated to have higher
weekly consumption of seafood at home than those households in either

79
suburban areas (X5) or nonmetro areas (base), ceteris paribus.
Similarly, households in suburban areas were estimated to have higher
weekly at-home consumption of seafood than those households in nonmetro
areas, ceteris paribus. Central city households had total expected
weekly expenditures (quantity consumed) equal to $0,564 (0.355 pounds)
in excess of those households in nonmetro areas. Households in
suburban areas had total expected weekly expenditures (quantities)
equal to only $0,231 (0.190 pounds) greater than households in nonmetro
areas.
Since proximity to the coast was a major factor leading to the
development of many of the larger cities in the United States, house
holds in the larger cities may have greater access to a larger variety
and quality of seafood than those households in either suburban or
nonmetro areas. For example, New York City, Boston, and New Orleans
have major fish markets acting as central locations from which distri
bution of seafood products to other localities is coordinated. Due to
a decline in the accessibility of moderate cost quality seafood as one
moves away from the distribution centers, one would expect that the
probability of a household purchasing and hence consuming seafood to
decline in relation to distance from the distribution center. As noted
in column 4 of Tables 3-2 and 3-3, the estimated probabilities of
observing a positive level of expenditures and consumption of seafood
for a household residing in a central city area exceed those of a
household residing in a suburban or nonmetro area. These estimated
probabilities are consistent with the observed probabilities of
observing a consuming household given in column 4 of Table 3-1.

80
Season
At-home quantity of seafood consumed was estimated to be greatest
in the summer quarter (X7), followed by the spring quarter (X6), fall
quarter (X8), and winter quarter (base). Weekly expenditures, however,
were found to be much more constant across quarters with no statistical
differences noted for any season, as judged by the nonsignificance of
the asymptotic t-values associated with the parameter estimates (column
3 of Table 3-2). A quantity change not reflected by a corresponding
expenditure change suggests that price must also be changing.
During the one year period in which the survey data used in this
analysis were collected (April 1977-March 1978), the price for seafood
consumed at home as measured by the Consumer Price Index increased by
7.6 percent (United States Department of Agriculture, 1983). Given the
7.6 percent increase in the price of seafood consumed at home, the
question arises as to why the estimated weekly expenditures on seafood
consumed at home did not show a similar increase. In fact, weekly
expenditures were estimated to be lower (though only marginally) in the
later half of the survey year than in the first half. A possible
explanation put forward to answer this question is that households may
have reacted to the increase in the price of seafood consumed at home
by reducing quantities purchased leaving weekly household expenditures
on seafood unchanged. This would imply the quantity elasticity with
respect to price must equal approximately unity.
.rA

Household Life Cycle
The household life cycle category (X9-X17) appears to be very
useful in explaining weekly household at-home seafood consumption as
judged by the number of statistically significant parameter estimates.
Household life cycle estimates suggest that the composition of the
household, independent of size, explains both expenditures and quanti
ties of seafood consumed at home in a rather systematic and logical
manner. Furthermore, the manner in which consumption of seafood can be
explained via the household life cycle is as expected, given the
current understanding of the at-home seafood market. For example,
there appears to be a general tendency for increased consumption of
seafood associated with the maturing of the household. Households with
the household head less than 35 years of age (X9-X12) consistently
consumed less seafood than households in more mature life cycle
categories. Households with the household head from 35 through
64 years of age (X13-X16) generally consumed less seafood than house
holds comprised of an elderly married couple (base). Given that the
difference in household size between that of an elderly individual
(X17) and that of an elderly married couple (base) has been accounted
for by the variables representing household size (X24 and X25), the
estimated difference in at-home seafood consumption between these two
groups of households may represent differences in eating habits. For
example, elderly individuals may not wish to cook only for themselves,
especially those items requiring any amount of preparation time. This
would preclude them from consuming all but canned seafood which takes a
minimal amount of preparation before being suitable for consumption.

82
Overall, households comprised of young adults with children (Xll-
X12) had the lowest at-home seafood consumption of any of the life
cycle categories, ceteris paribus. A representative household
comprised of a young single adult with children (Xll) was estimated to
consume on average 0.144 fewer pounds of seafood at home than that of a
household comprised of a young single adult without children (X9) with
lower corresponding expenditures equal to about $0.21, ceteris paribus.
Analogously, a household comprised of a young married couple with
children (X12) was estimated to consume 0.0753 fewer pounds of seafood
with corresponding expenditures of about $0.12 less than that of a
household comprised of a young married couple without children, ceteris
paribus. Though these differences may appear small it should be kept
in mind that the analysis is based on a one week period. Extrapolating
to a one year period, a household comprised of a young single adult
with children is expected to consume almost 7.5 fewer pounds of seafood
at home than a young single adult without children, ceteris paribus.
With respect to the life cycle categories pertaining to middle
aged heads of households (X13-X16) the results do not appear to provide
any systematic trends. Households comprised of a single middle aged
adult without children (X13) had lower consumption and corresponding
expenditures than did those households comprised of a single middle
aged adult with children (X15). On the other hand, households con
sisting of a middle aged couple without children (X14) tended to have
higher consumption of seafood than did those households consisting of
a middle aged couple with children.
The results pertaining to the household life cycle can be used
to design and implement a seafood promotion/marketing strategy.
(S'. I

83
For example, the relatively low estimates of at home seafood consump
tion among those households categorized in the younger life cycle
categories establish the premise that seafood promotion/marketing
targeted towards this segment of the population may provide the seafood
industry and its support groups with greater net returns than that of
targeting households categorized in more mature stages of their life
cycle. Of course, the validity of this premise depends on the relative
costs associated with promoting seafood to households in the different
life cycles relative to the returns per dollar expended. However,
before targeting this group of households the reasons why this group of
households exhibits a relatively low level of seafood consumption needs
to be addressed. A couple of reasons can be offered to help explain
these results. First, the meal planner in the "younger" households
probably tends to be somewhat less experienced at preparing meals than
the meal planner in more mature households. Due to this factor, these
**- IBMJ.. > l*njr ,-!
meal planners are more likely to avoid cooking a meal which involves
any much of preparation. Seafood, especially fresh seafood, has a
reputation for being difficult to properly prepare. Thus, seafood may
not be prepared and consumed as often by "younger" households as would
be expected among more mature households.
A second explanation is specific only to those younger households
with children (X11-X12). This segment of the population exhibited the
lowest weekly seafood consumption among the different categories in
the household life cycle. The children in this household group will
generally be of a lower average age than in households categorized in
the more mature life cycles. Hence, these children will place a
larger burden on the meal planner's time than would older children in

84
the more mature household life cycles. A related factor pertains to
the hesitency among parents of serving seafood to younger children out
of fear that the bones in the fish may injure the children or that the
children will have problems eating certain types of seafood (such as
shellfish). All of these reasons suggest a marketing strategy aimed at
promoting highly processed/ready-to-eat types of seafood products to
this segment of the population. Additionally, one would expect to find
fresh seafood consumption, which requires the most preparation time and
with which bones are most frequently associated, to be lower among
younger households, especially those with children, than among other
households. However, for frozen and canned seafood for which prepara
tion time is minimal and bones are generally not a problem, one would
expect to observe little differences in consumption patterns among the
households categorized in the different life cycles. This hypothesis
will be examined in greater detail in the following chapter.
The characteristics and composition of the American household is
in a continual state of transition. Knowledge of this transition in
conjunction with the information provided by the results pertaining to
the household life cycle category can further aid the seafood industry
and its support groups in the planning stage of a long term marketing
strategy.
The first factor the seafood industry and its support groups
may wish to consider when planning a long term seafood marketing
strategy is the changing age structure of the American household.
Table 3-4 provides some statistics on the age distribution of
household heads for selected years. The statistics suggest that
younger households (those with household heads less than 35 years of

Table 3-4. Percentage distribution of household head by age for selected years
Age of house-
Year
hold head
1960
1965
1970
1975
1980
1982
< 35 years
25.4
25.5
Percent
25.3 29.2
31.0
30.4
35-64 years
60.9
60.5
55.2
50.8
48.5
48.8
;> 65 years
13.7
14.0
19.5
20.1
20.5
20.7
SOURCE: United States Department of Commerce, Bureau of the Census (various issues).

86
age) and older households (those with household heads greater than 65
years of age) have become increasingly important segments of the popu
lation over the past two decades at the expense of the middle aged
households. The results of the Tobit analysis suggest, however, that
at-home seafood consumption among younger households (X9-X12) and
households comprised of an elderly individual (X17) tends to be among
the lowest of any household life cycle category. The seafood industry
and its support groups thus may want to consider these age structure
changes when planning a seafood marketing strategy.
A second factor the seafood industry and its support groups may
want to consider when planning a long term seafood marketing strategy
is the growing proportion of households in the United States that have
no children. In 1960, 43.1 percent of all households in the United
States had no children of their own under 18 years of age. By 1982,
this proportion had increased to 49.2 percent (United States
Department of Commerce, Bureau of the Census, 1984). The results of
the Tobit analysis suggest that, at least among certain age groups,
weekly at home seafood consumption differs depending upon whether
children are present in the household, certis paribus.
A final factor the seafood industry and its support groups may
wish to consider when planning a long term promotional strategy is the
growing proportion of single adult with children households in the
United States. In 1970, 84.9 percent of all children under 18 years
of age were living with both parents. By 1982, the proportion had
fallen to 75 percent (United States Department of Commerce, Bureau of
the Census, 1984). A single parent with children can be expected to
have more constraints on his/her time than would be the case if both

87
parents were present in the household. Thus, households in this group
should have a higher demand for seafood products for which little
preparation time is required relative to households in which both
parents are present.
Race
Black households (base) had significantly higher at-home seafood
expenditures and quantities consumed than did White households (X18).
Similarly, Black households were estimated to consume greater quanti
ties of seafood than households of "Other" ethnic origins (X19) but
their weekly expenditures were not significantly different given the
insignificant t-value in column 3 of Table 3-2. Total weekly at-home
seafood consumption by a typical White household was estimated to be
0.78 pounds less than that of a similar Black household while expendi
tures by that same White household were estimated to be $0.86 less than
that of a similar Black household. Among households of some "Other"
ethnic origin, weekly at-home consumption of seafood was estimated to
be 0.42 pounds less than that of a similar Black household while
expenditures by this group were only $0.21 less than that of a similar
Black household, certis paribus.
The estimated probability of consuming seafood at home was
substantially lower among White households than among either Black
households or households of "Other" ethnic origins. Among White house
holds, the estimated probability of consuming seafood at home was 0.42,
compared to 0.61 among Black households, and 0.53 among households of
"Other" ethnic origins (column 4, Table 3-3). The large differences
associated with the probability of consumption among the different

88
races probably reflect cultural factors leading to differences in
tastes and preferences.
The proportion of Black households to that of White households has
been gradually trending upwards. In 1970, Black families represented
9.5 percent of all families in the United States. By 1982, the propor
tion of Black families to that of the total had increased to 10.5
percent (United States Department of Commerce, Bureau of the Census,
various issues). This represents more than a 10 percent increase in
the proportion of Black families in just over a decade. Households of
Spanish origin, though representing a relatively small proportion of
the total number of households in the United States, also represent an
increasingly important component of the population. Representing
approximately 3.9 percent of the total number of families in the United
States in 1970, the proportion of households in the United States of
Spanish origin has increased almost 40 percent in just over a decade )
and represented 5.4 percent of the total number of families in 1982
(United States Department of Commerce, Bureau of the Census, various
issues). As the statistics tend to highlight, though the proportion of
Black households and those of "Other" ethnic origins still represents
small proportion of the total number of households in the United
States, they do represent an increasingly important component of the
population. As such, their needs and wants in terms of seafood for
at-home consumption may want to be considered by the seafood industry
and its support groups when developing a long term marketing strategy.
Persons of Spanish origin may be of any race.
A/1 ¡ IfiT
t\
i
\

89
Household Receives Food Stamps
Households receiving food stamps (X20) were estimated to have
total at-home consumption of seafood equal to 0.11 pounds valued at
$0.10 in excess of those households not receiving food stamps.
Caution, however, needs to be exercised when discussing the importance
of food stamps to at-home seafood consumption since receiving food
stamps did not have a statistically significant effect on at-home
consumption of seafood in either the expenditure model or the quantity
model.
Though not documented, it is a commonly held belief among seafood
dealers that food stamps are an important factor in a households's
decision to purchase and consume seafood. The results of the Tobit
analysis, however, tend to refute this idea. Since eligibility to
collect food stamps is related to household income, it is likely that
Black and elderly households are major recipients of food stamps. In
fact, about 30 percent of all food stamp recipients were Black in 1982
(United States Department of Commerce, Bureau of the Census, 1985).
As discussed earlier, Black households and households consisting of an
elderly couple tend to consume more seafood at home than that of the
"average" household. The combination of these two events may have
resulted in the association of food stamps with seafood purchases.
Furthermore, receiving food stamps may influence the time of month
that households collecting food stamps purchase seafood. If receiving
food stamps tends to result in these households purchasing seafood
during a narrow span of time, seafood dealers may incorrectly associate

90
this with an overall increase in seafood purchases among this group of
consumers.
Household Caught Fish for Own Use
As expected, households which caught fish (X21) consumed a greater
quantity of seafood than did their counterparts. Similarly, weekly
expenditures on seafood consumed at home were higher among households
4
who caught fish than among those who did not. As evidenced by the
estimated probabilities given in column 4 of Table 3-3, having caught
fish was second only to race among the discrete variable in determining
whether a household consumed seafood.
Employment of the Meal Planner
Employment of the meal planner (X22) was not statistically
significant in explaining weekly expenditures or quantities of seafood
consumed at home. The estimated signs of the respective coefficients,
however, were negative as expected given the increased opportunity of
the meal planner's time when employed. In the next chapter, considera
tion will be given to the effect of employment by the meal planner on
consumption of those products which require the most preparation time.
Sex of the Meal Planner
The sex of the meal planner was significantly related to at-home
consumption of seafood, ceteris paribus. Households with female meal
4
If the fish consumed during the one week interview period
consisted of that which had been caught rather than purchased, the
price assigned to that product was based on the average retail price of
a comparable product in that region and season.

91
planners (X23) consumed greater quantities of seafood and had greater
expenditures than households with male meal planners. This may reflect
a female's familiarity with cooking procedures, etc.
Household Size
The household size was found to be an important determinant of
at-home seafood consumption which is consistent with related studies
(e.g., Capps, 1982; Salathe, 1979). The positive linear household size
coefficient (X24) and the estimated negative coefficient associated
with household size squared (X25) imply that weekly expenditures and
quantity of seafood consumed increase with increases in household size
but at a declining rate. Because of the nonlinear specification of the
household size component in the models it is useful to evaluate the
effect of household size on seafood consumption at various levels of
household size. The more important effects are presented in Table 3-5.
As the information in Table 3-5 suggests, within the relevant
range very little economies to size are exhibited in either seafood
expenditures or quantities consumed. This is not surprising given the
small magnitude of the estimated parameter associated with the squared
term of household size (X25) compared to the linear term (X24) in the
two models. The information provided in the table indicates that total
weekly seafood expenditures decline with the addition of a fourth
member while the total quantity consumed declines with the addition of
a fifth household member. Within household sizes generally encountered
in the data each additional household member resulted in increased
consumption of 0.158 pounds per week with related expenditures
increasing about $0.26.

Table 3-5. Estimated effects of changes in household size on weekly expenditures and at-home
seafood consumption3
<
Component
Number of
people
residing in
household
1
2
3
4
5
Avg.
Weekly expenditures ($):
Total expected change in
consumption 9E(EXP)/9X.£
.201
.206
.207
.205
.199
.206
Expected change attributable
to consuming households
[aEEXP^/aXj F(Z)
.062
.068
.072
.075
.077
.072
Expected probability of
consumption F(Z^)
.397
.436
.472
.507
.538
.476
Change in expected probability
of consumption 9F(Z)/9X.£
.040
.038
.035
.033
.030
.035
Weekly quantity consumed (lbs):
Total expected change in
consumption 3E(Q)/9X
.149
.155
.158
.159
.158
.158
Expected change attributable
to consuming households
[9E(Q*)/9X] F(Z)
.044
.049
.053
.056
.058
.053
Expected probability of
consumption F(Z^)
.369
.407
.443
.478
.510
.447
Change in expected probability
of consumption 9F(Z^)/9X
.038
.037
.036
.034
.031
.035
kO
ho
aEvaluated at an income level of $15,080 thousand.

93
Expected seafood expenditures among consuming households
consistently increased though at a declining rate with each additional
household member (Table 3-5). This suggests that among consuming
households each additional household member resulted in a small decline
in both expenditures and quantities consumed for household members.
These declines may relate to price discounts associated with larger
purchases and less waste per household member with increases in
household size. j
The expected probability of observing a positive level of both
expenditures and quantity consumed increased with each additional
household member. The observed probabilities, presented in Table 3-1,
bear out the fact that households consuming seafood on a weekly basis
were in fact approximately 13 percent larger than those households not
consuming seafood. Though the expected probability of consuming
seafood increased with each additional household member, it did so at
a declining rate.
It is useful to address why the expected probability of consuming
seafood at home increased with household size, ceteris paribus. One
hypothesis put foreward to answer this question relates to the expected
increase in variety of foods consumed associated with increases in
household size. Assuming each member of any given household has an
individual preference function which contributes to a household
consumption function, increases in family size suggests an increased
probability that at least one member prefers seafood. This preference
will then translate to an increased probability that the household will
consume seafood.

94
As presented in equation (2.23) an elasticity associated with
Tobit analysis can be broken into two components where the first term
reflects the percentage change in consumption among consuming units
due to a change in while the second component reflects the
elasticity of the change in the probability of consuming associated
with the changes in X^. Approximately 60 percent of the estimated
household size expenditure and quantity elasticities reflect an
increased probability of consuming seafood associated with increases
in household size while the remaining portion reflects increased
expenditures and consumption among those households already consuming
seafood as noted in the following two elasticity estimates:^
NEXp = 0.1635 + 0.2450 = 0.4085
Nq = 0.1957 + 0.2834 = 0.4791
Subtracting the quantity elasticity with respect to household size from
the expenditure elasticity with respect to household size provides a
measure of the quality elasticity which in this example was estimated
to equal -0.0706. The estimated negative quality elasticity indicates
that households tend to purchase relatively less expensive seafood
items with increases in household size.
Household size remained relatively stable from 1950 to 1960,
declining from an average of 3.37 to 3.33. However, since 1960, the
average household size has declined significantly reflecting both a
decline in birth rates and an increase in the number of single person
households. Between 1960 and 1982, the average household size in the
^Evaluated at the mean values of all variables.

95
United States declined more than 18 percent to 2.72 individuals (United
States Department of Commerce, Bureau of the Census, various issues).
Thus, holding all other factors constant, the decline in household size
between 1960 and 1982 would have resulted in about a 7 percent decline
in expenditures and approximately an 8 percent decline in weekly
quantity of seafood consumed at home. Of course, other factors have
not remained constant. The declining family size, for instance,
reflects the gradual change in household composition. Increases in the
proportion of younger and older households, as discussed earlier, at
the expense of middle aged households will yield a smaller household
size. All of these factors must be viewed simultaneously when con
sidering expected changes in at-home seafood consumption throughout
the nation.
The declining household size has another important implication
that the seafood marketing sector may wish to consider when conducting
and planning a long term marketing program. Primarily, the seafood
industry and its support groups should recognize and react accordingly
to the fact that consumers of seafood products are going to desire
smaller portions of seafood due to the decline in family size. /
Education of Meal Planner
Seafood consumption was estimated to be positively related to the
education level of the meal planner (X26). Each additional year of
education of the meal planner increased the total household expendi
tures on seafood consumed at home by just over $0.04 per week, ceteris
paribus. Similarly, the total household at-home consumption of seafood
was estimated to increase by almost 0.03 pounds per week with each

96
additional year of education of the meal planner, ceteris paribus.
Though increased expenditures and quantities of seafood consumed
associated with additional schooling may appear relatively small, it
should be kept in mind that the analysis is conducted on a weekly
basis. Extrapolating to a yearly basis, each additional year of educa
tion is expected to increase household consumption of seafood by 1.56
pounds or about 0.53 pounds per household member. This figure is
relatively large when compared to the annual U.S. per capita consump
tion of seafood equalling 12.8 pounds (edible weight) as discussed in
Chapter I. The estimated positive relationship between education and
at-home seafood consumption may reflect increased awareness of the meal
planner for a balanced and nutritious diet. This relationship is
economically important given the increase in the level of education for
the population over the past two decades. In 1960, the median number of
school years completed by all individuals 25 years or older equalled
10.6. In 1982, the median equalled 12.6 years, or almost 20 percent
more than in 1960 (United States Department of Commerce, Bureau of the
Census, 1984).
Number of Guest Meals
Based on the positive relationship of both quantities consumed
and expenditures to the number of guest meals served (X27), it must
be concluded that households entertaining guests apparently often
serve seafood. Two hypotheses, or a combination of the two, can be
forwarded in an effort to explain the statistically significant posi
tive relation between seafood consumption and the number of guest
meals. First, increasing the number of guest meals by definition

97
increases the total amount of food served from home supplies which
consists of a variety of items including seafood. Thus, increasing
the number of meals served should effectively increase the amount of
seafood consumed. Second, at least certain types of seafood are
considered delicacy items (e.g., shrimp, lobster) to be served on
special occasions such as when entertaining guests.
Expenditures on Meals Consumed Away from Home
Given the propensity by the typical American household to consume
seafood in the away-from-home market, it is especially important to
examine what is happening in the away-from-home consumption market.
Haidacher et al. (1982) provide a good synopsis of the changing
patterns in the away-from-home market based upon a comparison of the
Spring 1965 United States Department of Agriculture Household Food
Consumption Survey with the Spring 1977 Nationwide Food Consumption
Survey. Some of their results are provided in Table 3-6.
As reported by Haidacher et al. (1982), away-from-home consumption
has expanded among every income group, each household size, and among
both Black and nonblack households. Overall, the percentage of meals
consumed away from home increased approximately 57 percent between the
spring of 1965 and the spring of 1977 from 9.8 to 15.4 (Table 3-6).
Among lunch and dinner meals when seafood is most likely to be ordered
and consumed, the percentage number of meals consumed away from home
increased 41 percent and 74 percent, respectively. The data provided
in Table 3-6 also suggest that increases in income are positively
related to the percentage of meals consumed away from home. However,
with respect to household size, no apparent pattern exists between the

Table 3-6. Percent of meals
characteristics,
eaten away from
spring 1965 and
home, by type of
spring 1977
meal and selected household
Meal type
All
Household
characteristics
Breakfast
Lunch
meals
Supper
1965 1977
1965 1977
1965 1977 1965 1977
All households:
3.7
7.2
18.9
26.7
6.9
12.0
9.8
15.4
Income quintile:
1 (lowest)
2.1
7.3
15.5
26.4
3.2
7.3
7.0
13.7
2
3.2
6.4
17.1
23.2
5.7
11.0
8.7
13.7
3
4.1
6.2
18.5
24.3
7.1
12.9
9.9
14.7
4
5.0
7.8
20.6
30.5
8.5
14.3
11.4
17.8
5 (highest)
5.2
8.6
26.0
31.0
12.0
18.3
14.4
19.4
Household size:
1 member
3.4
5.8
17.1
19.8
11.6
14.8
10.7
13.5
2 member
3.6
6.0
15.9
19.4
8.4
13.1
9.3
13.0
3 member
4.6
9.4
20.1
29.5
8.0
15.2
10.9
18.3
4 member
5.0
8.2
22.2
28.4
8.0
12.2
11.7
16.4
5 member
3.6
5.5
20.3
29.1
6.7
9.4
10.2
14.8
6 member
2.6
6.8
17.0
29.7
4.1
8.5
7.9
15.1
Race:
Black
3.0
9.0
17.3
29.5
2.6
7.9
7.6
15.4
White
3.8
6.9
19.2
26.3
7.5
12.5
10.2
15.4
SOURCE: Haidacher et al. (1982, p. 55).

99
percentage of meals consumed away from home and the size of the
household. Perhaps the most important information contained in Table
3-6 is that associated with the percentage of meals consumed away from
home according to race. Among Black households, the percentage of
lunch and dinner meals consumed away from home increased 70 and 204
percent, respectively, compared to only 37 and 66 percent,
respectively, among nonblack households.
Given the propensity for away-from-home seafood consumption com
pared to at-home consumption of seafood, it is important to recognize
the emerging patterns. Just as important, however, is the recognition
that the away-from-home market is likely to expand even more in the
long run, largely at the expense of the at-home market. A growing body
of research suggests that the total expenditure/income elasticity for
food consumed away from home is approximately twice that of the at-home
consumption market (e.g., Eastwood and Craven, 1981; Haidacher et al.,
1982). Second, though the estimated price elasticity associated with
food consumed at home is apparently more inelastic than that associated
with food consumed away from home, the importance of this factor in
stabilizing at-home food consumption will probably be negated by
changes in the socioeconomic structure of the population. For example,
increases in the education level and the proportion of females entering
the work force, and a decline in household size are all believed to
lead to increased expenditures on and the number of meals consumed
away from home (see Prochaska and Schrimper (1973) and Redman (1980)
for a discussion of those factors which determine away-from-home
consumption).

100
The results of the analysis indicate a rather strong negative
relationship between expenditures on meals away from home and weekly
at home consumption of seafood which is in contrast to the results
reported by Perry (1981). A $10 increase in expenditures on meals
consumed away from home was estimated to reduce at-home consumption of
seafood by $0,057 and 0.059 pounds, respectively. This relationship is
especially important considering the growing away-from-home consumption
market and one which must be considered in any attempt to increase the
at-home demand for seafood.
Income before Taxes
Income was found to be an important determinant of seafood consumption
as has been the case in most related studies (e.g., Capps, 1982;
Salathe, 1979; Haidacher et al., 1982; Perry, 1981). The statistically
significant positive linear effect (X29) and the negative coefficient
estimated for income squared (X30) imply a positive but declining
marginal propensity to purchase and consume seafood at home with
increasing income. The statistically significant parameter estimate
associated with the interaction term between income and race (X31)
suggests a different marginal propensity of at-home seafood consump
tion among households of different races. The negative estimate of
this term indicates that white households have a lower marginal
propensity to purchase and consume seafood at home at all levels of
income than nonwhite households, ceteris paribus. The positive
estimate of the parameter associated with the interaction between
family size and income (X32), though statistically insignificant,

101
indicates propensity to increase at-home seafood consumption with
increases in family size, ceteris paribus.
The nonlinearity specification of the income component in the
expenditure and quantity models makes it important to evaluate the
effect of income on weekly expenditures and quantities of seafood
consumed at various levels of income. The more important effects are
presented in Table 3-7.
The information given in Table 3-7 indicates a declining, albeit
small, marginal propensity to consume seafood at home with increases in
household income. For example, at an income level of $5,000, the total
expected change in weekly expenditures with respect to a $1,000 change
in income equalled $0,026. At an annual household income level equal
to $25,000, a $1,000 change in income was predicted to change the total
expected weekly expenditures by $0.0242, or about 93 percent of the
estimated change at an annual income level of $5,000.
About 25 to 35 percent of the change in at-home seafood consump
tion was estimated to reflect changes among those households already in
the market in terms of either increases or decreases in weekly expendi
tures and quantities consumed. The remaining 65 to 75 percent of the
change in at-home consumption of seafood, therefore, reflects changes
in the probability of market participation, either entry or exit
weighted by expected expenditures or quantities consumed.
The positive estimates of the total change in seafood consumption
and change in seafood consumption among participating households at
various levels of income as specified in Table 3-7 indicates that the
level of income required to reach a saturation level of seafood
consumption was far in excess of that reported by most households in

Table 3-7.
Estimated effects of changes in before tax income on weekly expenditures and quantities of
seafood consumed3
Component
Before
tax income
5
10
15
20
25
Avg.
Weekly expenditures ($):
Total expected change in
consumption 8E(EXP)/3X^
.0260
.0260
.0258
.0254
.0248
.0252
Expected change attributable
to consuming households
[3E(EXP*)/3X1] F(Z)
.0070
.0088
.0090
.0092
.0093
.0088
Expected probability of
consumption F(Z^)
.4347
.4578
.4797
.5005
.5201
.4759
Change in expected probability
of consumption 3F(Z)/3X^
.0048
.0046
.0043
.0041
.0039
.0043
Weekly quantity consumed (lbs):
Total expected change in
consumption 3E(Q)/3X
.0137
.0134
.0130
.0125
.0119
.0129
Expected change attributable
to consuming households
[3E(Q*)/aXi] F(Zi)
.0044
.0044
.0044
.0043
.0042
.0043
Expected probability of
consumption F(Z^)
.4149
.4352
.4501
.4641
.4771
.4469
Change in expected probability
of consumption 3F(Z^)/3X
.0032
.0031
.0029
.0027
.0025
.0029
a
Evaluated at a family size (X24)
equal to 3.153
and a
proportion of
White households
(X18)
equal to
0.813.
102

103
the 1977-78 survey. The information in Table 3-7 provides two reasons
why the saturation level of at-home consumption of seafood is expected
to occur only at very high levels of income. First, the change in
consumption among participating households declines very slowly with
increases in income at least within the relevant range. Second, the
expected probability of consuming seafood increases with income, though
at a declining rate, throughout the range of income reported by most
households in the 1977-78 survey.
Following the specification of the Tobit elasticity, given in
equation (2.23), the elasticity of expenditures on seafood consumed at
home with respect to annual before tax income equals^
nEXP = 0.0953+0.1436 = 0.2389
Similarly, the quantity elasticity of seafood consumed at home with
respect to before tax income equals^
Nq = 0.0496 + 0.1318 = 0.1814
The estimate of the weekly at-home seafood expenditure elasticity with
respect to income, 0.2389, is well within the range of estimates given
in previous studies, the results of which are summarized in Table 3-8.
The current estimate of the at-home seafood expenditure elasticity thus
adds to the growing amount of research that indicates that at-home
consumption of seafood is very unresponsive to changes in income.
^Evaluated at the means of all variables.
^Evaluated at the means of all variables.

104
Table 3-8. Estimates of at-home seafood expenditure elasticities
with respect to income3
Study
Expenditure
elasticity
Method of estimation
Capps (1982)
0.1651
0LS nonlimit observations
Salathe (1979
0.3568b
0LS all observations
0.2407
Perry (1981)
0.0609c
Tobit
0.2040
Haidacher et al. (1982)
0.16
0LS all observations
All studies with the exception of Haidacher et al. were based on the
1973-74 household consumption survey. The Haidacher et al. study was
based on the 1977-78 household consumption survey.
bThe first estimate given for Salathe's study was based on data from
June 1972 to June 1973 while the second estimate was based on data
from July 1973 to July 1974.
Q
The two estimates associated with Perry's study gives the range among
the different regions.

105
Assuming that the quality elasticity of at-home consumption of
seafood is positive, one would expect the quantity elasticity to be
somewhat less than the expenditure elasticity. The estimate of the
quantity elasticity, 0.1814, was in fact less than the estimated
expenditure elasticity which equalled 0.2389. This translates to a
quality elasticity for seafood consumed at home equal to 0.0515. Thus,
an increase in consumption resulting from an increase in income is
expected to result in a greater increase in expenditures than quantity
consumed; the difference measuring a demand for quality and/or
services.
Household income, measured in 1982 dollars, for selected years and
among different races is given in Table 3-9. As indicated from the
data in the table, after growing steadily throughout the 1950s and
1960s, the real household income stagnated during the 1970s and even
decreased in the early 1980s. The decline in growth in real household
income during the 1970s compared to the previous ten year period and
the actual decline in real household income during the early 1980s has
probably affected the growth in per capita consumption of commercial
fish and shellfish (Figure 1-1) and especially at-home seafood
consumption. Black households, estimated to have a higher propensity
to consume seafood at home with respect to income than did White
households, experienced a 13 percent decline in real income between
1970 and 1982 compared to only a 3 percent decline among White
households. Given the growing proportion of Black households in the
United States, the decline in real income among this group poses an
obstacle in any attempt to increase at-home consumption of seafood.

106
Table 3-
-9. Median family income
selected years
in constant (1982) dollars
for
Year
Race of Household
All
White
Black
1950
13,308
13,813
7,494
1955
15,926
16,629
9,170
1960
18,317
19,018
10,528
1965
21,283
22,183
12,216
1970
24,528
25,445
15,608
1975
24,664
25,589
15,744
1980
24,626
25,658
14,846
1982
23,433
24,603
13,598
SOURCE:
United States Department
of Commerce, Bureau of the
Census
(1984).

107
Given the relatively low estimate of the at-home seafood consump
tion elasticity with respect to income, even large increases in real
household income such as that of the 34 percent increase experienced
during the decade of the 1960s, will not have an overbearing effect
on at-home consumption of seafood. For instance, an increase in
real household income during the decade of the 1980s equal to that
experienced during the decade of the 1960s was estimated to result in
only a 6.2 percent increase in the quantity of seafood consumed at home
with related expenditures increasing about 8.1 percent.
Outlook for Increasing At-Home Demand
for Seafood and Implications
The outlook for increased at-home seafood consumption over the
next several years does not appear promising without significant
advances by the seafood industry and its support groups in terms of
more effective marketing and promotional efforts. For example, generic
advertising on seafood is extremely small when compared to most other
O
food sectors. Furthermore, nongeneric advertising on seafood in major
media outlets (excluding newspapers) declined 10 percent between 1977
and 1982 compared to a 55 percent increase in nongeneric advertising
associated with meat and a 167 percent increase in nongeneric advertis
ing associated with poultry (Anonymous, 1984). These factors alone
would tend to indicate a disadvantage to the seafood industry vis-a-vis
other food sectors; all competing for a limited household food budget.
Furthermore, as discussed throughout the analysis, movement in the
g
Generic advertising on seafood averaged $85,000 annually during
1981-82 compared to $4.2 million on red meats and almost $27 million on
milk and other dairy products (Morrison and Armbruster, 1983).

108
level of many of those factors explaining at-home consumption of
seafood is in a direction not compatible with that of maintaining long
run growth in the at-home seafood market.
Summarizing some of the discussion presented throughout this
chapter, any or all of the following conditions may be expected to
result in declining.at-home seafood consumption, ceteris paribus:
(1) an increase in the proportion of households with a younger head,
(2) an increase in the proportion of households in the United States
comprised of an elderly individual, (3) an increase in the proportion
of households in the United States in which the meal planner is
employed, (4) males becoming more involved in the planning and prepara
tion of meals, (5) a decline in the average family size, and (6) an
increase in expenditures on meals consumed away from home. As indi
cated throughout this chapter, most if not all the above conditions are
currently taking place in the United States.
Offsetting those factors which are expected to result in declining
consumption of seafood at home, seafood consumption at home can be
expected to increase given any or all of the following conditions,
certis paribus: (1) an increase in the proportion of households in the
United States comprised of an elderly couple. (2) an increase in the
proportion of nonwhite households in the United States, (3) an increase
in the education level of the meal planner, and (4) an increase in
household income. As previously indicated all these conditions have
been occurring in the United States over the past two decades.
Though several factors suggest that at-home consumption of seafood
will increase very slowly if not acutally decline while other factors
suggest that at-home consumption of seafood will increase, the one

109
overriding factor that will probably determine the future status of
demand for seafood for at-home consumption is that of the change in the
market for food consumed away from home. Given the propensity to
consume seafood in the away-from-home market, increases in away-from-
home consumption can be expected to have a rather strong negative
influence on at-home consumption of seafood. Evidence suggests that
away-from-home consumption will continue to take a larger portion of
the consumers income in future years. In fact, many of those factors
found in this study to increase at-home consumption of seafood also
increase total away-from-home consumption of food. Redman (1980)
investigated those factors determining expenditure on meals away from
home and concluded that increased family income and a college educa
tion of the woman head of household, both of which were estimated to
increase at-home consumption of seafood also contribute to increased
expenditures on meals away from home. Additionally, Redman concluded
that a decline in family size, which the U.S. population is currently
undergoing, also contributes to increased expenditures on meals away
from home. Decreases in expenditure on meals away from home were
found among Black households and with increased age of the woman. As
concluded in the current study, at-home seafood consumption was rela
tively high among Black households and elderly couples, though consump
tion was much lower among elderly individuals many of which are women,
certis paribus.
As expenditures on meals consumed away from home increases,
seafood consumed away from home as a percentage of total consumption
will increase proportionately. Thus, the cyclical nature associated
with the demand for seafood as discussed in Chapter I will likely

110
continue as the general economy of the United States and more
particularly restaurant trade oscillates with changes in real income.
An effort at promoting increased at-home consumption of seafood is
probably in the long run the only effective means of dampening this
cyclical demand.

CHAPTER IV
SPECIFIC SEAFOOD PRODUCT FORM ANALYSIS
Introduction
In this chapter, a discussion of the parameter estimates
associated with the disaggregated Tobit seafood models is presented.
Given the voluminous amount of information associated with the specific
product form models, discussion centers on noting and explaining
differences in the signs and significance of the estimated parameters
among the seafood product consumption models and with respect to the
total consumption models. This is in contrast to the in depth discus
sion given to each parameter estimate in the previous chapter. All
parameter estimates are, however, presented in Appendix B.
Some descriptive statistics associated with the data used in
estimating the specific seafood product form models (fresh, frozen,
canned, finfish, and shellfish) are presented in Table 4-1. Among the
alternative types of processing (fresh, frozen, canned), canned seafood
is consumed at home by the most households and frozen seafood by the
least households (Table 4-1). In fact, canned seafood is served by
more households (31.69 percent) than fresh and frozen seafood combined
(29.63 percent). However, referring back to Figure 1-1, per capita
consumption of canned seafood has equalled only about 60 percent of
that of fresh and frozen in recent years (edible basis). Thus, it must
be concluded that consumption of fresh and frozen seafood products is
111

Table 4-1. Descriptive statistics of data used in seafood product form models
Variable
Product category
Product
category
Fresh
Frozen
Canned
Finfish
Shellfish
Number of households
10,689
10,689
10,689
10,689
10,689
Number of consuming households
1,780
1,387
3,387
5,065
815
Percent of households consuming
16.65
12.98
31.69
47.39
7.62
Average expenditure among
consuming households ($)
4.243
2.582
1.296
2.485
4.055
Average consumption among
consuming households (lbs)
3.097
1.600
0.729
1.716
2.071
Average dollar/pound among
consuming households
1.370
1.614
1.778
1.448
1.958
112

113
preferred in the away-from-home food market relative to consumption of
canned seafood products.
Average consumption of canned seafood among consuming households
was found to be approximately 46 percent that of frozen seafood and 24
percent that of fresh seafood (Table 4-1). These differences probably
reflect differences in the amount of waste associated with the dif
ferent product forms at retail. For example, most if not all waste has
been removed by the processor before canning seafood. Frozen seafood
is generally filleted before sold and hence there is no bone or head
waste. Fresh seafood, however, is often sold on a whole weight basis
and hence there is often head, bone, shell, etc., waste associated with
the product. This waste is generally acknowledged to account from
about one-half to two-thirds of the raw product form.
Price variation among product forms reflects differences in waste
as well as the cost of processing the product. Canned seafood, which
requires the most processing and which has the largest round weight to
sales weight conversion, sold for an average of $1.78 per pound (Table
4-1). This compares to a sales price of $1.61 per pound for frozen
seafood and $1.37 per pound for fresh seafood.
The percentage of households consuming finfish (47.39 percent)
compared to those consuming shellfish (7.62 percent) suggests at-home
consumption of finfish was overwhelmingly preferred to at-home con
sumption of shellfish. The observed difference in the number of
households consuming finfish at home compared to the number consuming
shellfish at home is probably related to at least two factors. First,
meal planners probably perceive finfish products as easier to prepare

114
than shellfish products. Second, shellfish is considerably more
expensive than finfish (Table 4-1).
Comparisons of Consumption Parameters
Information concerning the signs and statistical significance of
the estimated parameters associated with the specific product form and
total seafood expenditure models is presented in Table 4-2. Since the
signs and statistical significance associated with the estimated
quantity models are generally in agreement with those of the expendi
ture models they are not presented here. Summarizing the results in
this manner allows for an easy and logical comparison of the estimated
models. The complete descriptive and summary statistics associated
with each of the specific seafood product form models are presented in
Appendix B.
Region
Overall, household expenditures on most seafood product forms were
highest among households residing in the Northeastern region of the
United States (XI) and lowest among households residing in the North
Central region (X2). This difference holds true for expenditures on
fresh and canned seafood. However, expenditures on frozen seafood
showed a reversal of that of the other two processed product forms with
respect to Northeastern and North Central regions. Weekly household
expenditures on frozen seafood were estimated to be greatest in the
North Central region of the United States and lowest in the North
eastern region.

Table 4-2. Signs of estimated parameters
models3
associated with
variables
included
in Tobit
seafood ex
penditure
Seafood
product
form
Variable
Total
Fresh
Frozen
Canned
Finfish
Shellfish
Region:
Northeastern (XI)
+*
+*
_*
+*
+*
+
North Central (X2)
'/c
+
.it.
Southern (X3)
-
-
-V.
+
Urbanization:
Central City (X4)
+*
+*
+
+*
+*
+*
Suburban (X5)
+*
+*
+
+*
+*
+
Season:
Spring (X6)
+
+
+
-
+
-
Summer (X7)
+
+*
-
+
~
Fall (X8)
+

>C
+
Household life cycle:
Young single w/o children
(X9)
.JJ.
-
+*
-
Young married w/o children
(X10)
_*
-
+
+
Young single with children
(Xll)
-V.
_-;s-
+
-
Young married with children (X12)
_*
-
+
_
+*
Middle aged single w/o children (X13)
"
+
115

Table 4-2. Continued
Variable
Seafood
product
form
Total
Fresh
Frozen
Canned
Finfish
Shellfish
Household life cycle (Continued):
Middle aged married w/o children (X14)
+
-
+
+
-
+*
Middle aged single with children (X15)
-
-
+*
-
+*
Middle aged married with children (X16)
-
_*
-
+
_*
+*
Elderly single (X17)
_*
-
-
-
-
Race of respondent:
White (X18)
_*
+
+*

Other (X19)
-
_
-
+*
+
Receives food stamps:
Yes (X20)
+
+
-
-
+
+
Caught fish for own use:
Yes (X21)
+*
+*
+*
+
+*
+*
Employment of meal planner:
Yes (X22)
-
-
-
-
-
116

Table 4-2. Continued
Variable
Seafood
product form
Total
Fresh
Frozen
Canned
Finfish
Shellfish
Sex of meal planner:
Female (X23)
+*
+
+
+*
+*
-
Family size:
Total number in household (X24)
+*
_
+*
+*
+*
_*
Total number squared (X25)
-
+*
+*
Education of meal planner:
Years (X26)
+*
+
+*
+*
+*
+*
Guest meals:
Number of meals (X27)
+*
+*
+*
+*
+*
+*
Meals away from home:
Dollars (X28)
+
Income before taxes:
Thousand dollars (X29)
+*
+*
+*

+*
+*
Thousand dollars squared (X30)

117

Table 4-2. Continued
Variable
Seafood
product
form
Total
Fresh
Frozen
Canned
Finfish
Shellfish
Interaction terms:
Income and race (X31)
_
Income and family size (X32)
Other seafood:
+
+
+*
+
+
Dollars (X33)
Constant:
NAC
-
+
+*
+*
o
Parameter estimates, t-values, and associated information are provided in Appendix B.
b
+ indicates that parameter estimates was positive; indicates that parameter estimates was negative;
* indicates that parameter estimates was statistically significant at the 90 percent level (t value >
1.654).
c
Not applicable.
118

119
The relatively high expenditures on fresh seafood by households
living in the Northeastern region of the United States is consistent
with the steady supply of fresh edible seafood products harvested and
sold in the Northeastern region. Similarly, the relatively high
expenditures on frozen seafood by households residing in the North
Central region may be explained by the relative unavailability of
fresh seafood in that region and hence the need for substitute frozen
products. Apparently, households in the Northeastern region can sub
stitute fresh seafood for frozen seafood which explains the relatively
low expenditures on frozen seafood in that region. The relatively high
expenditures on canned seafood in the Northeastern region compared to
the other regions of the United States was, however, unanticipated
given the availability of this product form in all regions of the
country at expected comparable prices. A possible higher demand for
at-home consumption of seafood in total in the Northeastern region due
to differences in tastes and preferences may help to explain the higher
expenditures on both fresh and canned seafood products in that region
compared to other regions of the United States.
The results of the analysis indicated that expenditures on both
finfish and shellfish were highest in the Northeastern region of the
United States and lowest in the North Central region. These differ
ences in expenditures between the two regions probably reflect differ
ences in availability of fresh quality seafood as well as differences
in consumer preference between the two regions.
The preceding discussion should be considered by the seafood
industry and its support groups when considering what products and
product forms to provide in different regions of the country.

120
For example, the Gulf and South Atlantic Fisheries Development Founda
tion has coordinated the effort of several groups aimed at expanding
demand for many of the underutilized species landed in the Southeastern
United States which enjoy only regional acceptance. Much of the effort
has gone to developing markets for these products in the Midwestern
section of the United States. (See Cato and Prochaska (1981) for a
discussion of the program.) Given the familiarity with frozen seafood
products in this section of the country, promoting these underutilized
species in frozen form appears logical.
Urbanization
Households residing in central city (X4) and suburban areas (X5)
had higher estimated expenditures on all seafood product types and
forms than did households residing in nonmetro areas (base), control
ling for regional differences, income differences, etc. Furthermore,
with the exception of frozen seafood purchases, households in central
city areas consistently had higher expenditures on all seafood product
forms than did households residing in suburban areas, ceteris paribus.
The proximity of many of the larger cities in the United States to
major fishing ports would explain, in part, the higher expenditures on
fresh seafood consumed at home among households residing in central
city areas in the United States. However, proximity to the coast does
not explain the relatively high expenditures on canned seafood products
by central city households since canned seafood products can probably
be transported to the inland areas of the country at a minimal cost.
Therefore, some other factor must account for the estimated differences
in weekly household expenditures on canned seafood products and to some

121
extent, possibly the other product forms as well. Households residing
in nonmetro areas of the country are often farming families or live in
farming communities. As such, generations of households have been
raised on meat products and introduction of seafood products may not
have been initiated to an extent necessary to establish a strong
at-home demand for these products among these households.
Season
Generally, season of the year was estimated to be statistically
insignificant in explaining weekly household expenditure on at-home
consumption of seafood. Weekly expenditures on fresh seafood products
consumed at home were found to be highest in the summer months (July,
August, September, 1977), while expenditures on frozen seafood products
were highest in the spring months (April, May, June, 1977), and
expenditures on canned seafood products were estimated to be highest in
the winter months (January, February, March, 1978), ceteris paribus.
Household Life Cycle
The household life cycle category (X9-X17, base) was very useful
in explaining weekly expenditures on at-home consumption of the
specific seafood product forms. For example, expenditures on fresh
seafood tended to increase with the maturing of the household, with
elderly households generally having the greatest expenditures on fresh
seafood products (Table 4-2). Expenditures on canned seafood consumed
at home, however, were lowest among elderly households (X17, base),
ceteris paribus. Expenditures on frozen seafood products attributable
to differences in the stages of the household life cycle were

122
comparable to fresh seafood expenditures with the exception that house
holds consisting of an elderly individual (X17) had lower weekly
expenditures than several of the middle-aged life cycle categories,
ceteris paribus.
Finfish and shellfish expenditure patterns differed significantly
with life cycle stage, ceteris paribus. Households in the younger
stages of their life cycles (X9-X12) generally had lower finfish
expenditures than either middle aged (X10-X16) or elderly households
(X17, base). Elderly couples (base) had the highest weekly expendi
tures on finfish. However, weekly expenditures on shellfish products
were relatively low among elderly households (X17, base) when compared
to most of the middle-aged households (X13-X16) and even some of the
younger household life cycle categories (X9-X12).
Overall, the estimated parameters relating household life cycle
category to specific seafood product forms conform to theoretical
expectations. Expenditures on fresh seafoods by younger and middle-
aged households were relatively low when compared to elderly
households. The fear of bones associated with nonprocessed seafood,
especially fresh, is hypothesized to explain the relatively low
expenditures among households with young children. A lack of cooking
and/or preparation expertise may be another reason for the low
consumption. Expenditures on canned seafood products which generally
have no bones and are typically recognized as requiring little prepara
tion and cooking skills were higher among younger and middle-aged
households with and without children than among elderly households,
ceteris paribus. Similarly, weekly expenditures on finfish for at-home
consumption among households with young children (XI1, X12) tended to

123
be relatively low when compared to other households, ceteris paribus.
Though expenditures on finfish products by households with young
children tended to be relatively low, a similar trend was not noticed
with shellfish expenditures. It is likely that the specification of
the life cycle category was overly refined to account for differences
in shellfish consumption among households in different life cycles
given the relatively few positive observations for this product form.
Race
Substantial differences in preferences for the alternative seafood
product forms exist among households of different races. Black house
holds (base) made significantly higher expenditures on fresh seafood
consumed at home than either White households (X18) or households of
other ethnic origins (X19), ceteris paribus. Alternatively, expendi
tures on canned seafood were significantly higher among White
households and households of "Other" ethnic origins than among Black
households. The difference in expenditures on frozen seafood consumed
at home was not statistically significant among White, Black, and
"Other" households.
Statistically significant differences in expenditure patterns
were estimated among households of different races in the purchasing
of finfish products but not shellfish products. Estimated expenditures
on finfish products consumed at home were highest among Black house
holds and lowest among White households, ceteris paribus.

124
Food Stamps
In no instance did the fact that a household purchased or received
food stamps (X20) significantly affect expenditures on the different
categories of seafood consumed at home. Expenditures on fresh seafood
products, finfish, and shellfish were positively related with food
stamps while expenditures on frozen and canned seafood products were
negatively related to food stamps.
Fish Caught for Own Use
Households who caught fish for their own use (X21) consistently
made greater expenditures on the specific seafood product forms than
did those households who did not catch fish. Furthermore, with the
exception of expenditures on canned seafood consumed at home, having
caught fish for home use was estimated to be statistically significant
in explaining expenditures on the specific seafood product forms.
The statistical insignificance of having caught fish for home use
associated with weekly expenditure on canned seafood consumed at home
is not unexpected given the nature of the product. Canning fish is
not generally attempted by individual households. Hence, households
who catch fish are likely to consume it fresh or freeze it for later
consumption.
Employment of the Meal Planner
Weekly expenditures on the various seafood products consumed at
home were consistently lower among those households in which the meal
planner was employed (X22) than among those where the meal planner

125
was unemployed. However, only fresh seafood expenditures were related
to employment of the meal planner at a statistically significant level.
This observation is consistent with the rationale previously discussed
for including this variable. The cost to the meal planner in terms of
the amount of time required to purchase and prepare fresh seafood is
expected to be greater than that required for other seafood product
forms and thus the employment of the meal planner was expected to have
its greatest effect on consumption of fresh seafood.
Sex of the Meal Planner
A female meal planner (X23) (as opposed to a male planner) was a
statistically significant factor in explaining weekly expenditures on
canned seafood products, finfish products, and total seafood consumed
at home. Expenditures on fresh seafood and frozen seafood were also
higher among households with a female meal planner, though not at a
statistically significant level.
Family Size
Weekly expenditures on the various seafood products consumed at
home were very responsive to changes in family size (X24, X25, X32).
Both the linear and quadratic parameters describing expenditures on the
specific product forms were statistically significant for all but fresh
seafood products. The interaction between family size and income (X32)
was statistically significant only for canned seafood products.
The estimated positive linear term (X24) together with the esti
mated negative quadratic term (X25) indicates initial increasing (at a
declining rate) expenditures on frozen seafood products, canned seafood

126
products, and finfish products with increases in family size. On the
other hand, a negative linear term combined with a positive quadratic
term was estimated for expenditures on both fresh seafood products and
shellfish products. These results suggested that expenditures on fresh
seafood products and shellfish products consumed at home declined, at
least within initial ranges, with increases in family size. Given the
increase in the opportunity cost of time of the meal planner associated
with increases in family size, these results are not totally
unexpected. Convenience in seafood products is most often associated
with certain frozen products, such as fish sticks, and canned products.
Furthermore, the majority of these products are made from finfish
rather than shellfish. As family size increaes, it is logical to
assume that the meal planner becomes more dependent on these conve
nience products.
The estimated expenditure elasticities with respect to family
size for the various seafood products are presented in Table 4-3.
Expenditures on at-home consumption of frozen seafood products, canned
seafood products, finfish seafood products, and total seafood products
have positive elasticities with respect to family size while expendi
tures on at-home consumption of fresh seafood products and shellfish
seafood products exhibit negative elasticities with respect to family
size. Furthermore, the total elasticities (comprised of the elasticity
of participation and the conditional elasticity) ranged from a low of
-0.73 associated with that of expenditures on shellfish to a high of
0.59 associated with that of expenditures on canned seafood products.
The above discussion and analysis suggest convenience in terms of
purchasing, preparation, and consumption of the various seafood product

Table 4-3. Estimated weekly expenditure elasticities with respect to family size for specific
seafood product forms3
Seafood
product form
Elasticity
among consuming
households
(conditional elasticity)
Elasticity associated
with entry/exit of
households (elasticity
of participation)
Total
elasticity
Fresh
-0.0201
-0.0866
-0.1067
Frozen
0.1180
0.4457
0.5637
Canned
0.1725
0.4167
0.5892
Finfish
0.2151
0.3473
0.5624
Shellfish
-0.1028
-0.6313
-0.7341
Total
0.1635
0.2450
0.4085
3 Evaluated at the means of all variables.
127

128
categories influence, at least in part, expenditure elasticities with
respect to family size. In a study conducted by Gillespie and Houston
(1975) consumers rated shellfish more difficult to prepare than
finfish. Hence, as would be expected, the expenditure elasticity with
respect to family size for shellfish products (-0.734) was much lower
then that for finfish products consumed at home (0.562). In the same
light, the estimated weekly expenditure elasticity associated with
fresh seafood (-0.107) was substantially lower than that associated
with either frozen seafood products (0.564) or canned seafood products
(0.589); the latter two categories requiring much less expertise and
time in preparation than that required for fresh seafood products.
A breakdown of the total elasticities into the change in consump
tion among consuming households and the change in consumption resulting
from entry/exit among households is also presented in Table 4-3. As
indicated by the information contained in the table, most of the total
elasticities reflects changes in the number of consuming households
(participation effect) as opposed to increased (decreased) expenditures
among consuming households (conditional effect). Since the conditional
elasticity measures increased (decreased) consumption among consuming
households, one would expect to find it to be a larger percentage of
the total, the greater the percentage of the households consuming the
product. For example, if all households consumed a given product then
the conditional elasticity would by definition equal the total since
there could be no entry from new consumers. As a percentage of the
total elasticities, the conditional elasticities tended to be higher
among those products most often consumed at home. For example, the
conditional elasticities associated with fresh and frozen seafood

129
consumption equalled about one-fifth of their respective total elas
ticities compared to slightly less than one-third for that associated
with canned seafood consumption (Table 4-3).
Education of Meal Planner
The relationship estimated between education (X26) and expendi
tures on at-home consumption of the various seafood products examined
in this study was positive in all instances and statistically signifi
cant with the exception of expenditures on fresh seafood products
(Table 4-2). The elasticities of expenditures on at-home consumption
of the various seafood product types and forms with respect to educa
tion of the meal planner were estimated to be 0.156 for fresh seafood
products, 0.606 for frozen seafood products, 0.405 for canned seafood
products, 0.205 for finfish products, and 1.249 for shellfish products;
compared to 0.552 for total seafood. Thus, increases in the level of
education of the meal planner is expected to have pronounced effects on
expenditures on at-home consumption of the seafood product forms. The
statistically insignificant education parameter estimate and relatively
low expenditures elasticity estimate associated with fresh seafood may
be the result of a confounded effect between an increase in the
opportunity cost of time of the meal planner which prevents him/her
from preparing fresh seafood and an increased desire for a nutritional
meal associated with additional education.
Number of Guest Meals
Increases in the number of guest meals (X27) were estimated to be
positively related to weekly expenditures on at-home consumption of

130
all seafood product categories examined in this study (Table 4-2).
The estimated positive relationship between the number of guest meals
and weekly expenditures on at-home consumption of the various seafood
product categories may be for one or two reasons. First, some seafood
products are generally considered as specialty products to be served on
special occasions, such as when entertaining guests. Second, increas
ing the number of guest meals by definition increases the total amount
of all food consumed, some of which is likely to be seafood.
Expenditures on Meals Away from Home
Expenditures on at-home consumption of all individual seafood
products with the exception of shellfish products, were estimated to be
negatively related to expenditures on meals away from home. Thus, as
expenditures on food consumed away from home continues to take a larger
portion of the household's food dollar, expenditures on the various
seafood products consumed at home are expected to decline, ceteris
paribus. This is consistant with the fact that a larger proportion of
total seafood consumption takes place in the away-from-home market.
However, it must be considered that all other factors determining
at-home seafood consumption are not remaining constant.
Income before Taxes
Consistent with estimates provided in the total seafood expendi
ture analysis, the estimated parameters associated with the linear
income term (X29) and squared income (X30) were positive and negative,
respectively, for all but one of the specific seafood product form
expenditure models (Table 4-2). Weekly expenditures on canned seafood,

131
the one exception, were negatively related to income in both the linear
and squared income terms. A look at the parameter estimates associated
with the interaction of income and race (X31) reveals that though
statistical significance is noted only in the total, all parameter
estimates were negative in sign.
The expenditure and quantity income elasticities for the specific
product forms and in total are provided in Table 4-4. Among specific
product forms distinguished by level/type of processing, consumption of
fresh seafood had the highest income elasticity estimates (NgXp =
0.467, Nq = 0.413) while at-home consumption of canned seafood
exhibited the lowest income elasticity estimates (Ng^p = 0.192, Nq =
0.098). Consumption of shellfish had a significantly higher income
elasticity estimates (NgXp = 0.543, Nq = 0.929) than did finfish
(Ngxp = 0.148, Nq = 0.1163).
Table 4-4 also provides information on the decomposition of the
total elasticities into their respective components; increased
(decreased) consumption among consuming households due to an increase
(decrease) in income and increases (decreases) in consumption reflect
ing increases (decreases) in the number of participating households.
For those products in which the number of consuming households was
relatively small (e.g., shellfish) the proportion of the total elas
ticity reflecting changes in the number of participating households
was relatively large compared to those products in which the number
of consuming households was relatively large (e.g., finfish). For
example, about 80 percent of the total estimated income elasticity
associated with shellfish products consumed at home reflected changes
in the number of participating households compared to 60 percent

Table 4-4. Estimated weekly expenditure, quantity and quality elasticities with respect to
before tax income for specific and total seafood product forms3
Category
Elasticity
among existing
households
(conditional elasticity)
Elasticity associated
with entry/exit of Total
households (elasticity elasticity
of participation)
Expenditures:
Fresh
Frozen
Canned
Finfish
Shellfish
Total
Quantities:
Fresh
Frozen
Canned
Finfish
Shellfish
Total
0.0899
0.0629
0.0561
0.0558
0.0958
0.3771
0.2403
0.1361
0.0927
0.4476
0.4670
0.3032
0.1922
0.1485
0.5434
0.0953
0.1436
0.2389
0.0795
0.0516
0.0310
0.0419
0.2059
0.3334
0.1994
0.0667
0.0744
0.7228
0.4129
0.2510
0.0977
0.1163
0.9287
0.0496
0.1318
0.1814
132

Table 4-4. Continued
Category
Elasticity
among existing
households
(conditional elasticity)
Elasticity associated
with entry/exit of
households (elasticity
of participation)
Total
elasticity
Qualities:
Fresh

0.0541
Frozen


0.0522
Canned


0.0945
Finfish


0.0322
Shellfish


-0.3933
Total


0.0575
a
Evaluated at the means of all variables.
133

134
for finfish. The estimated differences in the proportion of the total
elasticities accounted for by changes in the number of participating
households among the specific product categories is of importance for
marketing strategies. The number of consuming households is inversely
proportional to the number of potential consuming households. Hence,
for seafood products such as shellfish, one would expect to find a
relatively large market expansion through entry of new consumers as
income increases.
The estimated income elasticities presented in Table 4-4 should be
of considerable value to the seafood industry and its support groups.
As discussed in Chapter I, growth in per capita consumption of canned
seafood from 1960 through 1983 averaged only about a quarter of the
growth in per capita consumption of fresh and frozen seafood even
though about 56 percent of the seafood advertising in major media
outlets is specific to canned tuna and canned salmon (Anonymous, 1984).
At least part of the reason for the relatively small growth in per
capita consumption of canned seafood may be the result of a low income
elasticity associated with this product. A comparison of the quantity
elasticities with respect to income for the specific seafood product
forms suggests the income elasticity for canned seafood consumed at home
is only about a third of that for frozen seafood consumed at home and a
fourth of that for fresh seafood consumed at home. Thus, continued
promotion of canned seafood products may be important to assure a
stable though not necessarily increasing market for these products.
This is especially true given the fact that demand of canned seafood in
the expanding away-from-home market appears to be somewhat limited.

135
Though the estimated income elasticities for at-home consumption
of canned seafood were the lowest among the specific product forms, the
quality elasticity for canned seafood consumed at home (0.0945) was the
largest among the specific product forms (Table 4-4). This may reflect
the possibility that the higher priced and better quality fresh and
frozen seafood products tend to be predominately associated with the
away-from-home consumption market while most types of canned seafood
are generally considered as products for at-home consumption.
Other Seafood Expenditures
Expenditures on shellfish were significantly related to finfish
expenditures (X33) in a positive manner and vice versa-*- (Table 4-2).
This relationship parallels the work conducted by The Longwoods
Research Group Limited (1984) who concluded that heavy (frequent) users
of one category of seafood were also heavy (frequent) users of other
categories of seafood. Similarly, purchases of canned seafood were
positively related to expenditures on other seafood, though not at a
statistically significant level. Expenditures on fresh and frozen
seafood were both negatively related to expenditures on other seafood
which is in contrast with the results offered by The Longwoods Research
Group Limited.
"'Caution should' be exercised when examining the parameter estimate
associated with other seafood expenditures (quantities) in the dif
ferent models due to a possible simultanous bias associated with these
variables as discussed in Chapter II.

136
Outlook for Increasing At-Home Demand for
Specific Product Forms and Implications
Much of what was discussed in the last section of the previous
chapter extends to the discussion in this section. As mentioned,
generic seafood advertising of seafood is minimal when compared to
generic advertising of either meat or dairy products. As also men
tioned, the majority of nongeneric seafood advertising is directed at
increasing sales of only canned salmon and canned tuna. These factors
alone indicate a disadvantage in long run growth in at-home consumption
of most seafood products compared to meat, poultry, and dairy products.
Though it is difficult to predict future at-home consumption due
to the simultaneous shifting of all factors that determine consumption
of the specific product forms, a couple of general comments concerning
expected future demand for these products can be made based on the
preceding analysis. In terms of at-home consumption of the specific
product forms, the potential for long run growth appears greater in the
fresh and frozen seafood product markets than in the canned seafood
market for several reasons. First, the estimated income elasticities
of at-home quantity consumption of fresh seafood (0.413) and frozen
seafood (0.251) were about three to four times the size of the
estimated income elasticity of at-home consumption of canned seafood
(0.098), even though the majority of the nongeneric advertising dollars
goes towards the promotion of canned seafood products. Thus, increas
ing household real income should promote more long term growth in fresh
and frozen seafood consumption than canned seafood consumption. Second,
the growing proportion of Black households in the United States should
benefit growth in consumption of fresh seafood products since as the

137
results of this chapter suggest, Black households prefer consumption
of fresh seafood to that of canned seafood. Third, the analysis
indicates that the current decline in family size will be especially
conducive to growth in the market for at-home consumption of fresh
seafood and equally disadvantageous to long run growth in the market
for at-home consumption of canned seafood. Finally, the increasing
mobility of the U.S. population together with expected improvements in
terms of packaging and transporting fresh and frozen seafood products
to the inland portions of the country should prove conducive to
increasing sales of these products.
Though the above mentioned factors suggest long run growth in
at-home consumption of fresh and frozen seafood products and a possible
stagnation in the already modest growth of canned seafood consumption,
the seafood industry and its support groups need to be aware of certain
factors that may impede growth of at-home consumption of fresh and
frozen seafood. The main factor which is likely to impede long run
growth in at-home consumption of especially fresh seafood but also
frozen seafood relates to an increasing cost of the meal planner's
time. An increase in the education level of the meal planner, an
increase in the number of meal planners employed outside the home, and
a decrease in average family size are all believed to be related to an
increase in away-from-home food consumption. Thus, though increases
in education and decreases in family size were shown to be positively
related to at-home consumption of fresh seafood in this analysis, it
must be realized that they will also result in an increase in the
number of meals consumed away from home. Given the substantial
proportion of seafood consumed away from home, especially fresh and

138
frozen, the seafood industry and its support groups must develop new
fresh and frozen seafood products that require minimal preparation and
cooking time.

CHAPTER V
CONCLUSIONS AND IMPLICATIONS FOR FURTHER RESEARCH
Conclusions
This study was conducted to enhance the current understanding of
socioeconomic and demographic factors believed to influence at-home
consumption of total seafood and specific seafood products (fresh,
frozen, canned, finfish, and shellfish). A better understanding of the
factors determining at-home consumption of seafood also provides
information which can be used in examining the away-from-home seafood
market, seafood import demand, and ultimately the total demand for
seafood in the United States.
In addition to the "traditional" estimates associated with
regression analysis, the Tobit procedure used in this analysis provided
a method for decomposing the change in the level of at-home seafood
consumption resulting from a change in any exogenous variable into two
components. The first component measured the change in consumption
resulting from increased (decreased) consumption among existing
(consuming) households. The second component measured the change in
total consumption related to an increase (decrease) in the number of
participating households. For the total seafood consumption models,
it was estimated that approximately 35 percent of the change in
at-home seafood consumption resulting from a change in an exogenous
variable results from increased/decreased consumption among
139

140
consuming households. The remaining 65 percent of the total change
is attributable to entry into or exit from the at-home seafood market
by households. Among product forms distinguished by type of process
ing, the proportion of total change in consumption (resulting from a
change in an exogenous variable) attributable to increased/decreased
consumption among consuming households ranged from about 20 percent for
either fresh or frozen seafood to about 30 percent for canned seafood.
This breakdown of the total change into its two components should
assist the seafood industry and its support groups in making a compari
son of the benefits of attempting to increase at-home seafood consump
tion by encouraging increased consumption among consuming households as
opposed to increasing at-home seafood consumption by increasing the
number of consuming households.
The variables used in the analysis can be loosely grouped into
four categories. The first category includes variables consisting of
demographical and seasonal demand shifters. The second category
includes those variables determining the family structure. The third
category includes those variables that constitute the social and ethnic
structure of the household. The final category includes those vari
ables relating to economic considerations of the household.
Within the first category, variables denoting region, urbaniza
tion, and season were used to explain at-home seafood consumption.
The two demographic variables, region and urbanization, were of con
siderable value in explaining at-home seafood consumption patterns.
Northeastern households consumed significantly higher amounts of total,
fresh, canned, and finfish seafood at-home than did households residing
in other regions of the United States. Similarly, households residing

141
in center city areas consumed more seafood in total and more of all
product forms with the exception of frozen. Season was of significance
in explaining seafood consumption only in isolated instances.
Within the second category, variables denoting family size and the
composition of the household were used to explain at-home seafood
consumption. At-home consumption of seafood in total and for frozen,
canned, and finfish were found to be positively related to initial
increases in family size while at-home consumption of fresh seafood
and shellfish were negatively related to increases in family size.
To measure household composition, a set of variables reflecting the
life cycle of the 'typical' household was included in the analysis.
The results pertaining to this set of variables indicate that the
composition of the household, independent of size, is very important
in explaining at-home seafood consumption patterns in total and for
specific product forms. For example, households in the younger stages
of their life cycle, especially those with children, avoided consump
tion of fresh and finfish seafood products and total seafood relative
to households in more mature stages of their life cycle. Elderly
households, on the other hand, consumed the least canned seafood of
any life cycle stage, ceteris parabus.
Within the third category, variables denoting the race of the
household, the employment status of the meal planner, the sex of the
meal planner, the education of the meal planner, expenditures on meals
consumed away-from-home, the number of guest meals, and the household
catching fish were used to explain seafood consumption. Race of house
hold was used in the analysis to account for variations in tastes and
preferences among households of different races which in turn would

142
lead to differences in at-home seafood consumption. The results
suggested that Black households had higher at-home consumption of total
seafood and fresh seafood while White households had higher consumption
of canned seafood, ceteris paribus. Similarly, Black households had
considerably higher consumption of finfish than did White households,
while little difference in shellfish consumption between Black and
White households was evident.
The employment status of the meal planner, the sex of the meal
planner and the education of the meal planner loosely describe some of
the social attributes of the meal planner. Employment of the meal
planner was hypothesized to increase his/her cost of time and hence
result in a decline in at-home consumption of those products which
require relatively more preparation time. Though the results supported
this hypothesis, statistical significance was noted only for fresh
seafood consumption. Households with female meal planners compared to
those with male meal planners consumed more seafood in total and of
each product form with the exception of shellfish. With respect to
education of the meal planner, the results of the analysis indicated
that total seafood consumption and consumption of all product forms,
with the exception of fresh, were positively related to increased
education at a statistically significant level. Though consumption
of fresh seafood was positively related to increased education, the
relationship was not statistically significant.
Expenditure on meals consumed away from home, the number of guest
meals, and catching fish often represent a form of social entertainment
for members of the household. The results of the analysis indicated
that total seafood consumption and consumption of all product forms,

143
with the exception of shellfish, were negatively related to increased
expenditures on meals away from home at a statistically significant
level. Total seafood consumption and consumption of all product forms
were positively related to an increased number of guest meals at a
statistically significant level. Finally, households who caught fish
for their own use had higher at-home consumption of total seafood and
all specific product forms than did households who did not catch fish.
Within the final category, variables denoting whether the house
hold received food stamps and household income were used to explain
at-home seafood consumption. Receiving food stamps was found to have
little effect on at-home seafood consumption. Income, on the other
hand, was an important factor in determining at-home seafood
consumption. The estimated elasticities of at-home seafood consumption
with respect to income were positive and inelastic for all specific
seafood product categories and in total. Among product forms differ
entiated by level of processing (fresh, frozen, and canned), fresh
seafood exhibited the highest income elasticity while at-home con
sumption of canned seafood exhibited the smallest estimated income
elasticity. Similarly, at-home consumption of shellfish exhibited a
much higher income elasticity estimate than that found for finfish.
Implications for Further Research
A natural and needed extension of this study would be a parallel
study of the away-from-home seafood consumption market. Data collected
recently by the Market Research Corporation under contract with the
National Marine Fisheries Service could provide the basis for the
needed research. A comparison of the at-home and away-from-home

144
seafood markets would provide information which could be used to
study the interrelationships between the two markets and help answer
questions which continually arise regarding various policy issues.
A second extension of this study would be a parallel study of
at-home and away-from-home consumption of specific species such as
shrimp, groundfish, etc. An analysis of this nature was omitted from
this study because of the relatively few observations in the data base
used pertaining to any specific species within socioeconomic and demo
graphic classes. However, since the data collected by the Marketing
Research Corporation includes away-from-home seafood consumption, this
data base is probably adequate for an analysis of this type.
With respect to the present study, there are at least two
research options available to improve the understanding of at-home
seafood consumption. First, additional variables and/or a change in
the specification of those variables used in the current analysis may
prove beneficial. For example, religion has often been hypothesized
to influence seafood consumption but was not noted in the data base in
the current study. Similarly, additional interaction terms might be
tested. Estimation and inclusion of a wage rate of the meal planner
similar to that discussed by Prochaska and Schrimper (1973) may avoid
some of the confounding effects associated with family size and educa
tion that were observed in this study.
A second option that may prove beneficial when investigating
at-home seafood consumption relates to the statistical technique used
in the present study. With the Tobit procedure the question of whether
to consume and the level of consumption are considered simultaneously.
In reality, however, these questions may be addressed independently

145
by the consumer. In the first step, the consumer may decide whether
to consume a given good and in the second step decides the level of
consumption. A two step statistical procedure that addresses these
two questions independently may, therefore, have some advantages over
the Tobit procedure.
A final area of study which might prove to be extremely fruitful
to the seafood industry and its support groups involves examining the
relative merits of increasing at-home seafood consumption by encourag
ing entrance into the at-home seafood market by nonconsuraing households
as opposed to encouraging increased consumption among consuming
households. Such a study would have to examine the relative costs
associated with these two alternatives.

APPENDICES

APPENDIX A
DEFINITIONS OF SELECTED VARIABLES

Region (X1-X3)
Those areas of the 48 conterminous states as defined by the United
States Department of Commerce for the 1970 Census of Population. The
four census regions and their states are
Northeast (XI)Connecticut, Maine, Massachusetts,
New Hampshire, New Jersey, New York, Pennsylvania, Rhode
Island, Vermont;
North Central (X2)Illinois, Indiana, Iowa, Kansas,
Michhigan, Minnesota, Missouri, Nebraska, North Dakota,
Ohio, South Dakota, Wisconsin;
South (X3)Alabama, Arkansas, Delaware, District of
Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland,
Mississippi, North Carolina, Oklahoma, South Carolina,
Tennessee, Texas, Virginia, West Virginia;
West (base)Arizona, Arizona, Colorado, California, Idaho,
Montana, Nevada, New Mexico, Oregon, Utah, Washington,
Wyoming.
Urbanization (X4-X5)
Distinction of households in central city, suburban, and non
metropolitan areas was based on the standard metropolitan statistical
area (SMSA) as defined by the United States Department of Commerce in
the 1970 Census of Population.
All urbanizationsComposite of central city, suburban, and
nonmetropolitan households appropriately weighted.
Central city (X4)Population of 50,000 or more and main or
core city within SMSA.
Suburban (X5)Generally within the boundaries of SMSA but
not within legal limits of central city SMSA.
Nonmetro (base)All U.S. areas not within SMSA.
148

149
Season (X6-X8)
Surveyed seasons of the year are
Spring (X6)Months of April, May and June, 1977.
Summer (X7)Months of July, August, and September, 1977.
Fall (X8)Months of October, November, and December, 1977.
Winter (base)Months of January, February, and March, 1978.
Household Life Cycle Stage
Young households with or without children (X9-X12)head of
household is less than 35 years old.
Middle aged households with or without children (X13-X16)
head of household is greater than or equal to 35 years of
age but less than 65 years of age.
Older households (X17-base)head of households is 65 years
of age or greater.
Seafood Expenditures and Quantity Consumed
The seafood expenditure and quantity data used in this analysis
includes food commodities with the first four digit codes equalling
4521 or 4522 in the 1977-78 National Food Consumption Survey data
tapes. As such, food items which included a seafood product as only
one component of the total item were excluded from the analysis.
Seafood quantities were reported in the form brought into the kitchen.

APPENDIX B
DISAGGREGATED SEAFOOD STATISTICS

151
Table B-l. Descriptive statistics of variables in fresh seafood models
(1)
(2)
(3)
(A)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Mean (percent)3
Region:
Northeastern (XI)
0.248
0.303
0.237
0.203
North Central (X2)
0.241
0.156
0.258
0.108
Southern (X3)
0.342
0.366
0.337
0.178
Western (base)
0.169
0.175
0.168
0.172
Urbanization:
Central City (X4)
0.305
0.397
0.287
0.217
Suburban (X5)
0.355
0.333
0.359
0.156
Nonmetro (base)
0.340
0.270
0.354
0.132
Season:
Spring (X6)
0.238
0.235
0.238
0.164
Summer (X7)
0.232
0.262
0.226
0.188
Fall (X8)
0.268
0.254
0.271
0.158
Winter (base)
0.262
0.249
0.265
0.158

Table B-l. Continued
(1)
(2)
(3)
(A)
Category
Consumers &
Consumers
Nonconsumers
Proportion
nonconsumers
(nonlimit
(limit
of category
(total sample)
observations)
observations)
consuming
Mean (percent)a
Household life cycle:
Young single w/o children (X9)
0.055
0.036
0.059
0.109
Young married w/o children (X10)
0.058
0.053
0.059
0.152
Young single with children (Xll)
0.035
0.037
0.034
0.176
Young married with children (X12)
0.151
0.110
0.159
0.121
Middle aged single w/o children (X13)
0.075
0.080
0.074
0.178
Middle aged married w/o children (X14)
0.115
0.143
0.110
0.207
Middle aged single with children (X15)
0.056
0.064
0.055
0.190
Middle aged married with children (X16)
0.261
0.269
0.260
0.172
Elderly single (X17)
0.099
0.096
0.100
0.162
Elderly married (base)
0.095
0.112
0.090
0.196
Race of respondent:
White (X18)
0.852
0.713
0.880
0.139
Other (X19)
0.030
0.040
0.027
0.222
Black (base)
0.118
0.247
0.093
0.349
152

Table B-l. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Receives food stamps:
Yes (X20)
No (base)
Caught fish for own use:
Yes (X21)
No (base)
Employment of meal planner:
Yes (X22)
No (base)
Sex of meal planner:
Female (X23)
Male (base)
Mean (percent)3
0.074 0.095 0.070 0.214
0.926 0.905 0.930 0.163
0.234 0.298 0.221 0.212
0.766 0.702 0.779 0.153
0.465 0.429 0.472 0.154
0.535 0.571 0.528 0.178
0.908 0.928 0.904 0.170
0.092 0.072 0.096 0.130
153

Table B-l. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(A)
Proportion
of category
consuming
Actual mean
values
Family size:
Total number in household (X24)
2.946
2.963
2.943
Total number squared (X25)
11.457
11.672
11.414

Education of meal planner:
Years (X26)
11.732
11.559
11.767

Guest meals:
Number of meals (X27)
1.141
1.333
1.102

Meals away from home:
Dollars (X28)
11.658
10.641
11.862

Income before taxes:
Thousand dollars (X29)
14.109
14.572
14.017
Thousand dollars squared (X30)
324.590
355.264
318.461

154

Table B-l. Continued
Category
(1)
Consumers &
nonconsumers
(total sample)
(2)
Consumers
(nonlimit
observations)
(3)
Nonconsumers
(limit
observations)
(4)
Proportion
of category
consuming
Actual mean
values
Interaction terms:
Income and race (X31)
12.745
11.756
12.943
Income and family size (X32)
46.628
47.275
46.498

Other seafood:
Dollars (X33)
0.780
0.783
0.780
Pounds (X33)
0.455
0.425
0.461

Weekly expenditures and quantity:
EXP
0.707
4.243
0.000
Q
0.516
3.097
0.000

Number of households:
10,689
1,780
8,909

Percent of households:
100
16.65
83.35

The data provided in this table associated with the binary variables (X1-X23) should be interpreted as
representing proportions rather than percentages. To obtain percentages, multiply data by 100.
155

Table B-2. Summary statistics for Tobit analysis of weekly household expenditures on fresh seafood3
Category
Parameter Asymptotic
estimates t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X.
9E(EXP)b
Expected change
among consuming units
3E(EXP-) F(Z.)
3X.
(1)
(2)
(3)
(4)
(5)
(6)
Region:
Northeastern (XI)
1.1633
3.035
.2389
0.2779
0.0658
North Central (X2)
-2.6341
-6.209
.1271
-0.3348
-0.0616
Southern (X3)
-0.2033
-0.524
.1949
-0.0396
-0.0086
Western (base)


.2005


Urbanization:
Central City (X4)
1.9322
5.718
.2266
0.4378
0.1011
Suburban (X5)
0.7099
2.181
.1867
0.1325
0.0281
Nonmetro (base)


.1660


Season:
Spring (X)
0.5887
1.639
.1977
0.1164
0.0254
Summer (X7)
1.2707
3.583
.2206
0.2803
0.0641
Fall (X8)
0.1812
0.517
.1867
0.0338
0.0072
Winter (base)


.1788


156

Table B-2. Continued
Category
Parameter
estimates
a.
i
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X^
9E(EXP)b
9X
Expected change
among consuming units
3E(EXP*) F(Z-i)
9Xi
(1)
(2)
(3)
(A)
(5)
(6)
Household life cycle:
Young single w/o
children (X9)
-3.1139
-3.706
.1423
-0.4431
-0.0843
Young married w/o
children (X10)
-1.7503
-2.512
.1814
-0.3175
-0.0662
Young single with
children (X11)
-3.0440
-3.518
.1446
-0.4402
-0.0843
Young married with
children (X12)
-2.9030
-4.717
.1469
-0.4265
-0.0827
Middle aged single
w/o children (X13)
-1.5097
-2.156
.1894
-0.2859
-0.0609
Middle aged married
w/o children (X14)
-0.1662
-0.307
.2327
-0.0387
-0.0090
Middle aged single
with children (X15)
-1.2426
-1.767
.1977
-0.2457
-0.0535
Middle aged married
with children (X16)
-1.1957
-2.023
.1977
-0.2364
-0.0515
Elderly single (X17)
-0.8230
-1.280
.2090
-0.1720
-0.0383
Elderly married (base)


.2389

___
157

Table B-2. Continued
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X-^
9E(EXP)b
9X.
Expected change
among consuming units
9E(EXP*) F(Z-¡ )
9X.
(1)
(2)
(3)
(4)
(5)
(6)
Race of respondent:
White (X18)
-5.2222
-8.982
.1736
-1.0623
-0.2185
Other (X19)
-2.4925
-3.329
.2676
-0.667
-0.1663
Black (base)


.3707

Receives food stamps:
Yes (X20)
0.3583
0.660
.2061
0.0738
0.0164
No (base)


.1949


Caught fish for own use:
Yes (X21)
3.1191
10.557
.2743
0.8556
0.2167
No (base)


.1685


Employment of meal planner:
Yes (X22)
-0.6360
-2.154
.1841
-0.1171
-0.0248
No (base)


.2061

158

Table B-2. Continued
Expected total
Category
Parameter
Asymptotic
Expected
change resulting
Expected change
estimates
t-ratio
probability
from change in X.
3E(EXP)b
3X.
i
among consuming units
a
i
F(Z.)
9E(EXP*) F(Z.)
3X.
i
(1)
(2)
(3)
(4)
(5)
(6)
Sex of meal planner:
Female (X23)
0.8101
1.476
.1977
0.1602
0.0330
Male (base)


.1736

Family size:
Total number in
household (X24)
Total number
-0.3922
-1.018
.1949
-0.0258
-0.0056
squared (X25)
0.0603
1.714

Education of meal planner:
Years (X26)
0.0483
0.985
. 1949
0.0094
0.0020
Guest meals:
Number of meals (X27)
0.2500
5.424
.1949
0.0487
0.0106
Meals away from home:
Dollars (X28)
-0.0156
-2.042
.1949
-0.0030
-0.0007
159

Table B-2. Continued
Category
Parameter
estimates
ai
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in X^
9E(EXP)b
SXi
Expected change
among consuming units
9ECEXP*) F(Z-i)
3X
(1)
Income before taxes:
(2)
(3)
(4)
(5)
(6)
Thousand dollars (X29)
Thousand dollars
squared (X30)
Interaction terms:
0.1926
-0.0006
3.958
-1.963
.1949
0.0234
0.0051
Income and race (X31)
Income and family
-0.0615
-1.505



size (X32)
Other seafood:
-0.0067
-0.961
Dollars (X33)
Constant:
-0.0787
-1.130
.1949
-0.0153
-0.0033
a0
-6.9439
-5.575



The value of Xg at the means of all X^'
s is equal
to -7.43939; a
= 8.68865.
^The effects of the interaction and/or squared terms have been accounted for in the construction of the
associated linear terms.
160

Table B-3. Summary statistics for Tobit analysis of weekly household quantity consumption of fresh seafood'
Category
Parameter
estimates
h
Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in Xi
3E(Q)b
3Xi
Expected change
among consuming units
3E(Q*) F(Z .)
axi 1
(1)
(2)
(3)
(A)
(5)
(6)
Region:
Northeastern (XI)
1.0323
3.153
.2260
0.2339
0.0540
North Central (X2)
-2.0036
-5.539
.1230
-0.2464
-0.0444
Southern (X3)
0.2296
0.696
.1949
0.0447
0.0097
Western (base)


.1867


Urbanization:
Central City (X4)
1.5590
5.437
.2177
0.3393
0.0770
Suburban (X5)
0.6699
2.427
.1841
0.1233
0.0261
Nonmetro (base)


.1611


Season:
Spring (X6)
0.6A65
2.117
.1922
0.1243
0.0268
Summer (X7)
1.2407
4.118
.2177
0.2701
0.0613
Fall (X8)
0.3270
1.098
.1814
0.0593
0.0124
Winter (base)


.1711


161

Table B-3. Continued
Category
Parameter
estimates
6i
Asymptotic
t-ratio
Expected
probability
F(Zi)
Expected total
change resulting
from change in X^
3E(Q)b
aXi
Expected change
among consuming units
3E(Q*) F(Z<)
3X
(1)
(2)
(3)
(A)
(5)
(6)
Household life cycle:
Young single w/o
children (X9)
-2.6944
-3.770
.1357
-0.3656
-0.0687
Young married w/o
children (X10)
-1.5748
-2.660
.1736
-0.2734
-0.0563
Young single with
children (X11)
-2.7134
-3.706
.1357
-0.3682
-0.0692
Young married with
children (X12)
-2.5502
-4.889
.1401
-0.3573
-0.0679
Middle aged single
w/o children (X13)
-1.2786
-2.156
.1841
-0.2354
-0.0498
Middle aged married
w/o children (X14)
-0.1010
-0.220
.2296
-0.0232
-0.0054
Middle aged single
with children (X15)
-1.0427
-1.753
.1922
-0.2004
-0.0433
Middle aged married
with children (X16)
-1.0483
-2.094
.1922
-0.2015
-0.0436
Elderly single (X17)
-0.7158
-1.314
.2033
-0.1455
-0.0321
Elderly married (base)




162

Table B-3. Continued
Category
Parameter
estimates
h
Asymptotic
t-ratio
Expected
probability
F(Z.)
Expected total
change resulting
from change in X.
8E(Q)b
aXi
Expected change
among consuming units
3E(Q*) F(Z.)
3X
(1)
(2)
(3)
(A)
(5)
(6)
Race of respondent:
White (X18)
-4.7697
-9.723
.1635
-0.8916
-0.1790
Other (X19)
-2.5689
-4.031
.2514
-0.6458
-0.1565
Black (base)


.3745


Receives food stamps:
Yes (X20)
0.3883
0.848
.2033
0.0789
0.0174
No (base)


.1894

Caught fish for own use:
Yes (X21)
3.0059
12.022
.2776
0.8344
0.1387
No (base)


.1587

Employment of meal planner:
Yes (X22)
-0.4019
-1.605
.1814
-0.0729
-0.0153
No (base)


.1977


163

Table B-3. Continued
Category
Parameter
estimates

Asymptotic
t-ratio
Expected
probability
F(Z)
Expected total
change resulting
from change in
8E(Q)b
9*i
Expected change
among consuming uni