The World’s Largest Open Access Agricultural & Applied Economics Digital Library

This document is discoverable and free to researchers across the

globe due to the work of AgEcon Search.

Help ensure our sustainability.

Give to AgEcon Search

AgEcon Search
http://ageconsearch.umn.edu
[email protected]

Papers downloaded from AgEcon Search may be used for non-commercial purposes and personal study only.
No other use, including posting to another Internet site, is permitted without permission from the copyright
owner (not AgEcon Search), or as allowed under the provisions of Fair Use, U.S. Copyright Act, Title 17 U.S.C.
 Staff Papers Series

Staff Paper P79-35 October 1979

A THEORY ON INFORMATION AND ITS APPLICATION

TO THE EFFECT OF LABELING ON FOOKiPRODUCTS

Richard J. Sexton

~— .—

Department of Agriculturaland AppliedEconomics

t

Unlvcmty of Minnesota
lnstitutc ot Agriculture, Fmcstry and HOIIW Economics
St PatI1.Minncsotii 55108
 A THEORY ON INFORMATION AND ITS APPLICATION*
TO THE EFFECT OF LABELING ON FOOD PRODUCTS

by
**
Richard J. Sexton

October 1979

Staff papers are published without formal review within the Department
of Agricultural and Applied Economics.

$<
A paper completed as a Plan B project paper in partial fulfillment
of the requirements for the degree of Master of Science in Agricul-
tural and Applied Economics.

A*
Research Assistant, Department of Agricultural and Applied Economics,
University of Minnesota. The author is particularly indebted to Professor
Jean Kinsey for her continuing contributions during the development of
this paper. Professor Benjamin Sexauer and Terri A. Erickson also made
many helpful comments and suggestions on earlier drafts of this work.
 TABLE OF CONTENTS

?.%&
I. INTRODUCTION .............................................. 1

II. AN ECONOMIC THEORY ON THE EFFECT OF LABELING

ON FOOD PRODUCTS .......................................... 8

A. Determination of Losses from Imperfect

Information .......................................... 8

B. Determination of an Optimum Amount of Infor-

mation Provision ...................................... 19

c. Analysis of Consumer’s Response to Nutrition

Information ........................................... 24

III. AN APPLICATION OF THE MODEL .............................. 26

A. Background ........................................... 26

B. Empirical Framework ................................... 28

c. Empirical Results ................................... 31

D. Interpretation of Results ............................. 33

IV. SOME POLICY AND EMPIRICAL RECOMMENDATIONS ................. 36

v. SUMMARY .................................................. 38

VI. APPENDIX ................................................. 40

REFERENCES ............................................... 42
 A THEORY ON INFORMATION AND ITS APPLICATION
TO THE EFFECT OF LABELING ON FOOD PRODUCTS

Richard J. Sexton

I. Introduction

It is generally understood that when a consumer purchases a product

he/she has actually chosen the product for certain attributes or character-

istics contained in the product. For instance the attributes contained in

a typical food product might be taste, energy, nutrients or convenience of

preparation. This approach was first introduced by Lancaster (1966). A

problem arises in this regard if the consumer does not have complete know-

ledge of or access to data regarding all the attributes contained in any

product. When such a situation exists it leads to the likelihood that the

consumer’s expenditure decisions may not be optimal and that budget misallo-

cation will occur.

In particular the concern over the quality of Americans’ diet and

consumers’ lack of knowledge about food products led in 1969 to the White

House Conference on Food Nutrition and Health (Final Report, 1970). One

result of this Conference was a call for some standardized system to provide

nutritional and other information to consumers. As a result rules were

promulgated by the Food and Drug Administration (FDA) in 1972 and approved

in 1973 (U.S. Department of Health, Education and Welfare, (USHEW), 19731)

requiring the now familiar standardized nutritional label for all packaged

food products which either are nutritionally fortified or which make nutritional

claims. In addition the regulation provides that any nutritional data

voluntarily provided to be in the FDA format. Beloin (1973) provides a

summary and explanation of the regulations.

 2

Since the first nutritional labels began appearing in 1973 there has

been a good deal of study concerning the benefits and effectiveness of the

existing regulations and the possibility of their expansion or alteration.

Statutory authority in the area of food labeling is split among three federal

agencies. The United States Department of Agriculture (USDA) has jurisdic-

tion over meat and poultry products with the FDA controlling all other foods.

The Federal Trade Commission (FTC) is involved in food labeling through its

responsibility for regulation of food advertising. An FTC proposal in 1974

(presently unadopted) would have mandated that television commercials for

foods required to have nutritional labeling on packaging contain nutritional

and caloric information or make a statement as to the lack of such nutri-

tional value (Bettman, 1975).

The FDA on the other hand seems to be taking a wait and see attitude

regarding the effectiveness of present regulations before proposing exten-

sions or alterations. Among the possibilities currently being considered

by FDA are alterations in the format in which the data is presented (USHEW,

19781). Suggested changes include elimination of numbers from the label and

replacing them with graphs or wedge charts or adjectives such as good, fair

or poor. Also FDA is considering expansion of regulations to require dis-

closure of data about certain specific substances in food which consumers

may want to avoid for various reasons. In this regard FDA intends to pro-

pose that sodium and potassium content be declared on all nutritional labels.

In an attempt to ascertain the effectiveness of existing regulations and to

determine consumers perceptions of possible changes FDA, FTC and USDA jointly

held public hearings on the issue of food labeling at five locations in

the U.S. from August 22 to October 29, 1978 (USHEW, 19781, 19782).
 3

Studies and research on the effects of label disclosures have not been

able to discern any particular tangible benefits accruing to consumers from

the regulations. Consequently advocates of label disclosure have been forced

to rely on criteria such as the consumer’s “right to know” to justify their

call for disclosure regulations (Daly, 1976; USHEW, 19782).

An interesting dichotomy seems to have been uncovered in studies con-

cerning consumers’ use of nutritional data. It appears when asked, a large

percentage of consumers will say they want nutritional data, they will use

such data and they would be willing to pay something extra to have such data.

However, studies pertaining to actual usage of nutritional data tend to show

that consumers often don’t understand the data in the format presented and

seldom make use of it.

In particular a survey by Lenahan, et al.,(1972) indicated nearly 100%

favored nutritional labeling, 60% would use such data and 44% would be willing

to pay more on their food bill for it. A similar national survey of 1500

persons by FDA (USHEW, 19732) produced similar results. Seventy-five percent

said they would use nutritional labeling and almost half were willing to pay

50c more per week on their food bill to have such labeling. Ironically a

repeat survey by FDA (USHEW, 1975) after the requirements had been in effect

for awhile indicated a somewhat less favorable outlook towards labeling.

Fifty-eight percent of respondents reported to have noticed nutritional

labeling on food products and from this group 57% had used the data (this

group comprises 33% of the entire sample). Only 40% were now willing to pay

an extra 50C a week for the data.

A survey of New York state households by Daly (1976) indicated 91%

were concerned about nutrition and 58% were willing to pay something extra
 4

on their food bill for it. A national survey of women conducted by the

Gallup organization for Redbook (1976) reported 91% were concerned about

nutrition, 76% were interested in having more information about nutrition,

66% had read nutritional labels “during the last several months” and 49%

were willing to pay an extra 3$ per food item to have nutritional data be

placed on all foods. A study by Liechtenstein(1974) reported 71% would use

the label data and 54% were willing to pay something extra on their grocery

bill to have it. Another national survey by the Economic Research Service

(ERS) in the U.S. Department of Agriculture (USDA, 1977) reported 54% found

nutritional data useful or extremely useful. This figure increased to 63%

in a followup phase of the survey conducted about a year after the first phase.

The results of these and other surveys in the area are quite similar.

They indicate, as mentioned earlier, that a large majority of consumers

want nutritional data and have a desire to use the data. Generally about

half indicate some willingness to pay extra to obtain the data. The problem

is that studies report that actual usage of the data by consumers is minimal,

and consumers often have difficulty understanding the data in its present

format.

Lenahan, et al.(1972) reported that only 15% of respondents understood

the labels and only 10% used them. As noted earlier FDA (USHEW, 1975) reported

one-third of consumers had used the data. The Redbook survey (1976) reported

58% found information about nutrition confusing. Similar results were reported

by Jacoby, et al (1977). They found that consumers did not utilize much of

the available data especially when brand names were available. They also

found that acquisition decreased as the number of items shopped for increased.

During long shopping trips data acquisition was estimated to be less than 10%.
 5

The authors concluded that the majority of consumers neither use nor compre-

hend nutritional data in making food purchase decisions.

In a test of the FTC proposal to require presentation of nutritional

data on television commercials Scammon (1977) compared two peanut butter

spreads (Skippy and Koogle), and nutritional data were fabricated to make

Koogle appear more nutritious than Skippy. The data were presented in the

proposed FTC format. The results showed that in spite of advertising to the

contrary a large majority of respondents thought Skippy was more nutritious.

In fact those presented with the fabricated data were more apt to conclude

that than the control group.

When respondents in the second FDA survey (USHEW, 1975) were asked if

they understood label components, most claimed they did. For instance some

reported levels of understanding are as follows: calorie 82%, RDA 64%,

protein (grams) 62%, fat 58%, cholesterol content 51% and sodium content

36%. However, direct questioning regarding nutritional knowledge often did

not bear out the claims.

Particularly distressing to supporters of nutritional disclosure is

the fact that low income groups, those with lesser education and the elderly

(those most in need of dietary improvement) are least likely to understand,

acquire and use nutritional data. Daly (1976) reported that difficulty in

understanding or utilizing data was much more acute among the poor, aged and

uneducated. Chi square tests to measure differences in response to questions

concerning ability to process and use information were significant at .05 or

less for groupings based on age, income and education. The FDA (USHEW, 1975)

reported only 25% of those over 50 had used nutritional labels, and use was

only 242 for those with low socioeconomic status. These groups also expressed
 6

much less willingness to pay extra for such data. The ERS report (USDA,

1977) indicates similar findings in this area also. Conversely the Redbook

survey (1976) indicated only minor variations in response based upon age.

Some evidence does seem to indicate that consumers are becoming more

aware of nutrition and labeling in recent years. Better Homes and Gardens

(1978) found 78.5% within a panel of its readers had read label information

in the last ten days. Redbook (1976) reported 66% had read labels during the

past several months.

Summaries of the literature on labeling have been provided by General

Mills (1978) and Babcock, et al.(1975). Costs to industry for providing

labeling have been reported by Albrecht (1978), General Foods (1975) and the

Wall Street Journal (1973). Rusoff (1978) summarizes the industry viewpoint

on labeling. Summaries of the Government’s outlook towards labeling have

been provided by FDA (USHEW, 19782) and Forbes (1978).

The understanding of factors in consumer information processing is

important in determining why consumers often fail to utilize the information

resources at their disposal. The well supported principle of concreteness

says that consumers tend only to use information that is explicitly displayed

in the stimulus and will use it only in the form in which it is displayed.

Information that has to be stored in memory, infered from stimulus or trans-

formed tends to be ignored. From this Bettman and Kakkar (1977) have con-

cluded that policy to make data available to consumers is not sufficient.

The data must be presented in a form which is readily processable.

A study by Stanley (1977) tends to verify this proposition. In the

study consumers were presented with nutritional data on cereals and then

immediately after were asked to rank the cereals on perception of nutritional

 7

content and preference. The result was a correlation = ,678 between the

nutritional ranking of the cereals and consumer choice. An R* = .945 was

obtained for amount of variance in preference explained t]y nutritional per-

ception. Quite obviously this was a case where data was presented in an

easily processable format and where it didn’t have to be stored in memory.

Bettman and Kakkar (1977) also note that consumers acquire and process

information through either of two basic strategies. Choice may be made by

processing brands of a product whereby one brand at a time is chosen and

information is gathered on several attributes, then another brand is examined,

and so on. The other strategy is processing by attribute where a particular

product attribute is examined for several brands, and then another attribute

is examined, and so on. Studies have shown that consumers will process data

in the fashion easiest given the display. Since most displays emphasize

brands, processing by that strategy is encouraged and processing of attributes

such as nutritional value is naturally decreased.

In this regard Assam and Bucklin (1973) performed a latin squares

analysis of variance experiment to test consumers preference of different

label types. They found that manufacturer’s promotional statements on the

label were just as effective at influencing consumer’s purchasing decisions

as positive nutritional data. Similar conclusions were reached by Berning

and Jacoby (1974) who found that manufacturer determined data including price,

advertising and packaging are key factors in generating interest and awareness.

Other factors become important only at later stages. One could conclude from

this that tailoring displays to emphasize nutritional attributes would en-

courage processing by that attribute.

 8

In a summary article of the work on consumer use of information aids

Monroe Friedman (1977) concluded that overload of information and lack of

public understanding of the data provided were the basic problems causing

consumer failure to effectively utilize data at their disposal.

II. An Economic Theory on the Effect of Labeling on Food Products

A. Determination of losses from imperfect information

The situation reviewed in the preceding pages is basically a problem

dealing with the nature of consumers’ demand for and usage of information,

i.e., why do consumers on the surface want more data on food labels and then

when it is provided apparently fail to make use of it. To explain this and

other problems pertaining to product information acquisition, this portion.of

the paper will develop a theoretical framework within which the benefits and

costs of providing information to consumers can be analyzed. Due to the

great variety of products on the market, attempts to develop a theory suf-

ficiently general to encompass all or most products would be a difficult and

questionable undertaking. I will restrict the framework of this analysis

to food products and potential information disclosures concerning them.

Before proceeding it is necessary to distinguish between information

and data. Information can be defined as data which increases the knowledge

of the recipient. Data is therefore potential information. Before it becomes

information it must be processed by the recipient. The concern of this paper

is with data and information about particular food products. The data in

this case is produced either voluntarily or through regulation by the food

processor. But information is produced by the consumer through a household

 9

production function of the type developed by Becker (1973) with the inputs

being new data, prior information stored in memory and time. The mere

existence of data on a food label will do nothing unless the consumer pro-

cesses it into information. Since the type of data being dealt with here

is either mandated by regulation, or if voluntarily provided, subject to

regulatory guidelines; it is assumed for purposes of this paper that all

the data provided is accurate and not misleading.

Start with a consumer’s demand curve for some food product which can

be called product ‘A’. This demand curve at any time is largely determined

by the consumer’s perception of the attributes contained by the product

(Lancaster, 1966), This evaluation is made based upon the knowledge and

information available to the consumer at the time. Some of the attributes

of the product are readily discernible simply by trying the product, for

instance the way a product tastes can be quite accurately determined in this

fashion. Knowledge about other attributes such as nutritional content cannot

be easily determined in this manner. Therefore in a state of imperfect

knowledge and incomplete information the consumer’s existing demand curve

for ‘A’ may be different from what it would be in a state of perfect infor–

mation.

Let D1 be the consumer’s initial demand curve for ‘A’ based upon

his/her subjective evaluation of the attributes of ‘A’ in period t. Now

define DT as the consumer’s true demand curve for ‘A’ based upon perfect

information. If D1 # DT, the result will be a misallocation of consumer

expenditures with corresponding changes in consumer’s surplus and losses

in welfare.
 10

To formulate this loss assume in the absence of perfect information

that the following three possibilities exist with some probability of

occurance attached to each:

(1) The consumer has overevaluated the true quantity and nature of

attributes contained in the product and D1 > D is the demand

T(1) ‘here ‘T(1)
curve which would occur with perfect information if this situation holds.

(2) The consumer has underevaluated the true quantity and nature of

attributes contained in the product and D1 < D

T(2)’ ‘here ‘T(2) ‘s ‘he
demand curve which would occur with perfect information if this situation

occurs.

(3) The consumer has correctly evaluated the attributes of the pro-

duct and the current demand curve, in this case Dl, is equal to the demand

curve, DT, which would occur with perfect information.

Only one of these situations can actually exist at any time so the

probabilities of their occurrence sum to one. Letting Xl, X2, X3 stand

respectively for the probabilities of (l), (2) and (3) occurring it can

be said that

X1+X2+X3=1.

‘ote ‘hat ‘*lY ‘ne ‘f ‘T(l)’ ‘T(2) ‘r ‘1 = ‘T actually ‘Xist at any ‘ime
as the demand curve which would be attained with perfect information.

The consumer incurs a loss in terms of welfare if either situation

(1) or (2) occur. There is, of course, no loss if (3) occurs.

 11

Figure I

‘A(t)
M

p*

..(t)
DT(l) I)l DT(2)
‘2 ‘1 ‘3

Using the above figure and assuming a constant market price for ‘A’, P+<,

the loss if (1) or (2) occurs can be formulated (Peltzman, 1973). If (1)

occurs it becomes clear that the consumer has purchased too much of ‘A’.

He/she has purchased Al units in period t but actually should have purchased

only A2 units. For the additional units, A2 Al, beyond A2 which the consumer

purchased he/she paid an amount equal to A2 EF Al, but the actual value of

these units to him/her as now revealed by DT(l) is only A2EGA1. Therefore

the consumer has incurred a welfare loss in period t on these units of A2EFA1

- A2EGAl = GEF. If (2) occurs then the consumer has purchased too little

of ‘A’ and she/he should have purchased A3 rather than Al units.

The consumer’s surplus in period t would have been P*MJ had he/she purchased

A3 units, but since only Al units were purchased the surplus on the remaining

This surplus loss is represented by the area FHJ.

‘nits’ ‘lA3’ ‘s 10st”
 12

Multiplying the loss to the consumer if (1) or (2) occur by the probability

of their occurrence (Xi) enables one to formulate an equation for expected

loss (EL) to the consumer in period t from imperfect information:

(El) = xl [1/2(EF)(FG)]+ X2 [1/2(IiF)(FJ)].

‘L(E)
Letting the respective demand curves be written as follows:

‘1: QA= fO(P)

: QA= fl(P)
‘T(l)
: QA= f2(P)
‘T(2)

the expected loss can be formulated as follows:

‘2 1 ‘1 1
(E2)
‘L(t)
= Xl [J f (P) dp+ ((A1-A2) P~’)- ~ f (p) dpl
o 0
‘3 2
+X2 [_f f (P) dp -~1 f2(P) dp - ((A3-A1) P*)].
o 0

Lquation 2 shows that the consumercan expect to incur some dollar loss from

misallocation of expenditures on good ‘A’. The consumer, therefore, has

incentives to obtain information which would reduce this EL, and shefhe

should be willing to pay for data (potential information) an amount equal

to the reduction in EL which would be obtained with more information. A

consumer demand curve for additional information about product ‘A’ can

thus be formulated based upon this willingness to pay. Let B1P (Figure 11)

be the consumer’s demand curve for additional information (I) about ‘A’.

The point where the demand curve intersects the horizontal axis, lP, is

significant because it corresponds to the level of information needed to

 13

Figure II
PI

obtain perfect information about ‘A’. To see this, note that any new infor-

mation the consumer obtains about ‘A’ can be expected to reduce EL from mis-

allocation. The larger the level of information obtained the smaller becomes

EL and the less becomes the willingness to pay for additional information.

It follows directly that the consumer’s willingness to pay for the last bit

of data that will give “perfect” information goes to zero since the loss

from misallocation has also gone towards zero. At this point 8QA/31A = 0.

Further information will not affect the level of ‘A’ purchased.

Now formulate the demand curve for information about ‘A’ as follows:

&/
= f(QlA)--
‘I

where P = the price of information or willingness to pay for information

I
Q1A = quantity of information about ‘A’.

— .— — —.—.-—
*/
— This formulation is chosen because quantity of information is usually
provided exogenously by regulation. The quantity provided thus determines
willingness to pay.
 14

The demand curve BI is based on the consumer’s willingness to pay for

P
information to reduce EL. In particular since EL = O with perfect information

it must be true that

(E3) = jp f(QIA) dQIA.

‘L(t) ~,

The expected loss is equal to the area under the information demand curve.

This means that the demand for information will be greater for those products

which have a greater EL.

Given the information demand curve, BIP, assume now that through govern-

ment regulation or voluntary producer initiative some quantity of nutritional

data is provided exogenously on the label of ‘A’.

Assume the consumer obtains 011 in additional potentially useful infor-

mation from the data. The consumer’s willingness to pay for this data is
T

‘1
(E4) f f(QIA) dQIA.
o
In period t the consumer purchases some finite quantity, Al, of ‘A’, where

E4 gives his/her willingness to pay for the 011 in potential infor-

‘1 ~ 0“
mation which now appears on each package of ‘A’. The consumer would then

be willing to pay this amount extra for a package of ‘A’ containing the

potential information. But he/she would be willing to pay this extra amount

only one time since repeated encounters with the data would provide no new

potential i.nformation. If the consumer perceives the label disclosure on the

Aith purchase of ‘A’ in period t, he/she would then be willing to pay a price

for this unit equal to

A.
11
fl f“(t’)dp -tf f(QIA) dQIA.
A o
i-1
 15

Graphically, this could be interpreted as a break in the consumer’s demand

curve at whatever unit he/she encounters the new data. It would be expected

that consumers of ‘A’ would encounter the new data at various points along

their individual demand curves during the period t. Also since consumers

may have previously acquired and processed elements of the new label data

through other means, the amount of potential new information will vary for

each consumer in that data with which a consumer is already familiar cannot

be expected to further decrease his/her EL. Further note that willingness to

pay for each unit of potential information is unique to each consumer since

it is determined by EL which will certainly vary from consumer to consumer.

Considering the total market demand, obtained by summing each individual

demand, it would follow that this curve would temporarily shift out during

any market period in which a labeling scheme had been instituted. This shift

would be a reflection of consumers’ additional willingness to pay for a pacli-

age containing the new data. As a practical matter this should be a very

transitory situation which will exist only in the period of initial exposure

to the new data and prior to evaluation of it. Once consumers evaluate the

data a new market demand curve will exist as a consequence of changes due to

evaluation in individual consumer’s demand.

The consumer’s willingness to pay for 011 in potential information is

represented by E4. The demand for information is determined by EL, and the

consumer is willing to pay a price for this data equal to the reduction in EL

that will be generated by the data. By subtracting E4 from the formulation

for EL, E2, one obtains a new EL, EL’, which will exist after evaluation of

the new information. Therefore,

 16

I?2- E4 = EL’

To determine how information reduces EL assume that upon processing

the data the consumer decides one of three things:

(1) He/she has been overevaluating the product and D1 > DT.

(2) He/she has been underevaluating the product and Dl < DT.

(3) He/she has been correctly evaluating the product and D = DT.
1
Suppose the consumer decides D1 > D~ , a new demand curve, call it D , is
o
formed where Do < D1. As long as 011 < 01 the consumer still has incom-
P
plete information. This missing information, 01 – 011, if obtained could
P
reveal to the consumer that he/she is still overevaluating the product and

or that he/she is now underevaluating the product and Do < DT or

‘O > ‘T’
that he/she is correctly evaluating the product and Do = DT. If the con-

sumer decides D > D~, he/she will reduce purchases in the present period
1
to some level A on Figure III where A2 ~ A4 < Al (A2 could be O if D
4 T(1)
was such that the consumer would not purchase any ‘A’ at P*).

Figure III

‘A
 17

It has been shown that the consumer may still be overevaluating or

underevaluating the product after processing 011 in information if 011 < 01 .

P
As such the consumer will incur losses if either circumstance occurs. The

immediate concern is with the change in the area of potential loss due to the

reevaluation. By referring to Figure III it can be shown that since the con-

sumer is now operating on Do and purchasing less ‘A’ than before> the IOSS

from overevaluation (the loss if D the perfect information demand

T(1) ‘s
curve) is less than before. This loss is now expressed by the area REV.

information demand curve, the loss

‘owever’ ‘he 10ss ‘f ‘T(2) ‘s ‘he ‘erfect
from further underevaluation, expressed by the area RSJ, becomes larger. By

letting L stand for the total area of loss the change in the total area of

loss can be expressed as follows:

A
‘1 1 ‘4 1
(E5) AL = fl f2(P) dP + j- f (P) dP - [2P’~(A1-A4) + J f (P) dP
o 0 0

+f‘4 f2 (P) dP].

o

As a result of reevaluation there will be some increase in the area of loss

in the event of underevaluation (RSHF from Figure III and some decrease in

the area of loss in the event of overevaluation (VRFG from Figure III).

These changes tend to offset each other. But depending upon the position-

ing of the particular demand curves the offset is not likely to be complete,

and there may be some increase or decrease in the total area of loss. There-

fore, the sign of 8L/31 is uncertain.

It should be noted that it is consistent within the present analysis

that situation (3) on page 2 (that D1 = DT) could still hold true. If this

were the case the consumer would now be underevaluating the product due to
 18

Lhc added information since D < D The loss i.sobtained in a straight

o 1“
forward manner by setting D1 = DT(2), which is, of course, consistent with

the analysis. It can be seen that E5 still holds true in this circumstance.

It has been shown that the change in the total area of loss from re-

evaluation due to new information is uncertain. However, it has also been

shown that if information is to be demanded and utilized by consumers, it

must be true that 21EL/~1< 0, that is, information must reduce the expected

loss. According to the development above, increased information does not

necessarily reduce the area of loss. It must be, then, that it reduces the

probability of these losses occurring. Recall from pagelO that X3 is the

probability that the consumer’s current demand curve is equal to DT. DO is

now the current demand curve, and D is based upon more information than D
o 1
was. It must be true then that the probability that DO is correct is greater

than the probability that existed for D~ being correct before the additional

information was obtained. In other words X3 has increased due to added infor-

mation. Since Xl + X2 + X3 = 1, it must be that Xl andlor X have decreased,

2
and therefore that EL has decreased. X3 can be expressed as follows:

= f(I)
‘3

where I is the quantity of information obtained about the product.

These results imply that additional information need not result in

better purchasing decisions by consumers. In fact it could result in worse

decisions. This type of phenomenon has been observed by Jacoby (1977). But

since EL has decreased one would expect on average better decisions to follow

from increased information.

 19

11 upon evaluating the information the consumer decides (2) or (3)

from page 16, the same basic results would follow. If she/he decides (2)

(that he/she has been underevaluating the product) it will lead to an in-

crease in demand to some D > D And if the consumer decides (3), D1 = DT,
o 1“
there will be no change in demand for the product since he/she still believes

D1 is correct, and since this opinion is now based upon more information the

probability is greater now that he/she is right. EL will therefore decrease.

B. Determination of an optimum amount of information provision

1. The costs of information

There are two types of costs involved in information provision.

Producers incur a cost in providing the data on the label and consumers

incur a cost in processing the data into useful information. I assume there

is a one to one potential conversion ratio between data and information,

that is, 011 in data can through processing be converted into 011 in informa-

tion by the consumer. Remember that since data is defined as potential

information the amount of data that exists on any label is subjective and

unique to each consumer. The cost to producers for providing data apparently

includes some fixed component plus some variable component which is a function

of output. General Foods (1975), the Wall Street Journal

— (1973) and Albrecht

(1978) discuss the costs incurred by industry in labeling. The fixed com-

ponent of cost arises due to expenses incurred in initially obtaining the

data and in preparing the label. The variable component occurs due to the

need for inspections and monitoring the production process to ensure no devia-

tions from the label specifications. Also it is often argued that per unit

production costs are higher due to input rigidity imposed by labeling. In

 20

other words manufacturers are unable to alter their ingredient mix in

r(,sl)(]l)s~}
L() l~ri{:~s iI clt)ill~;
sigll:lls SC) would change tllc’
lab(’1specil-ic~l-

tions. As such the cost to producers for providing label data for any food

product can be written as:

c = ~ + (3QA (cost for product ‘A’)

P

where Cp = the cost to producers for providing label data

a = fixed cost of providing label data

8 = per unit cost of providing label data.

The cost to consumers is an implicit cost of time. The more data a

consumer wishes to process into information the greater will be the cost in

time. The cost to consumers for processing data into information can be

written as follows:

cc = YT

where cc = the cost to consumers for processing information

Y = a measure of the opportunity cost of time

T = time required to process data into information.

T can be formulated as follows:

T=~O
‘I

where A = time required to process a unit of information

Q1 = quantity of information processed.

Lambda (A) can be interpreted as a function of education and prior nutritional

knowledge. The higher one’s level of education the more quickly and easily
 21

information can be processed and the lower will be A. Lambda is also a

function of the degree of clarity and conciseness with which the data is

presented. The clearer and more potentially understandable the data is,

the lower will be A. Over information or information congestion can be

interpreted as increasing A. By substituting for T one can obtain

cc = YOQ1).

2. The benefits of information

The benefit to consumers from processing information is the reduction

in expected loss (EL) from expenditure misallocation. The more information

a consumer processes the less will be EL. Since EL determines the demand

for information, PI = f(Q1,4),the demand for further information decreases.

The change in the demand for information with respect to quantity of in-

formation obtained measures the reduction in EL from obtaining information.:’

Therefore we have

~[ff(QIA) dQIA]
(E6) .— -<0
aQIA

The value attained by E6 measures the marginal benefit to the consumer for

processing information considering the present time period t only. Since

the value in E6 is negative the absolute value of the expression is taken

“Information (I) does not appear as an argument i.nequation E2,

also DT(I) and DT(2) are not observable. So although it would be desirable
to express MB directly in terms of the change in E2 it is not possible to
do SO. Therefore it must be expressed in terms of E3 as is done in E6, E6’.
 22

th
to obtain the marginal benefit in period t, MBt, to the i consumer from

processing information about ‘A’ in that period.

(E6’) MBti = la(.ff(QIA)dQIA)/aQIA\

Assuming these benefits are obtained independently by consumers, the total

benefit in period t to consumers is obtained by the summation of the in-

dividual MBti over all consumers of the product.

N
(E7) Total MBt = Z
i=l MBti

where N is the total number of consumers of the product.

Since it has been shown that willingness to pay for specific types of

product data measures expected reduction in EL from that data, an empirical

specification of the marginal benefit expression is obtainable by asking

consumers how much they would be willing to pay for some additional specific

information disclosure. Responses would serve as a proxy for the expected

reduction in EL or as the marginal benefit of obtaining the information.

of information demand and provision

). iletermin(ation

The benefits to producers from supplying label data depend upon how

consumers evaluate the information provided. If, upon processing the label

data, consumers decide they have been overevaluating the product, they will

reduce their demand for the product and the producer will incur sales losses

as well as higher costs due to labeling. The producer will derive some

positive benefit from providing label data only if evaluation of the data

leads consumers to increase demand for the product. One would expect,

therefore, that producers would not voluntarily provide additional label

 23

data on food products unless they perceived that evaluation of the data

by consumers would lead to an increased demand for the product thus en-

abling them to charge a higher price or sell a larger quantity. It could

also be noted that studies by Daly (1976) and the FDA (1975) have indicated

certain indirect effects on consumer behavior from label disclosure, These

effects, such as increased consumer confidence, could alter the demand curve

for affected products.

The optimum level of information for consumers to process on any

product occurs where the marginal benefit of processing (the reduction in

EL) is equal to the marginal cost of processing. It has been shown that

the marginal benefit curve decreases as additional information is obtained.

The marginal cost function as formulated is constant. Considering the present

time period t only, the optimum level of information to process on product A

from the ith consumer’s viewpoint occurs where

(E8) MBti = 3CC /aQIA =y~=Mc

i i

th
where MC is the marginal.cost of processing for the i consumer and which
i
can be interpreted through “YAas being equivalent to the wage rate multiplied

by the unit processing time,

Since it is likely that any reductions in loss from information process-

ing would be permanent, it becomes necessary to measure these benefits into

future time periods by employing present value considerations to obtain a

more accurate representation of the true benefits of processing information.

This can be expressed as follows:

 24

M
(1:9) PVM13i= Z —:B~–
j=t (l+r)j

th
where PVMBi = the discounted benefit stream over the life of the i

consumer.

j = time periods t, 2, 3 ..... M

M = remaining life span of the consumer from period t

r = an appropriate discount rate.

The total discounted benefit to all consumers of the product, total PVMB,

then becomes

N
(E9’) Total PVMB = Z PVMBi .
i=l

This derivation is analogous to that indicated by Nelson (1971) in his dis-

cussion of information search.

c. Analysis of consumer’s response to nutrition information

A brief summary of the theoretical framework is in order. It was

shown that in the absence of perfect information about a product the con-

sumer incurs some expected dollar loss (EL) from expenditure misallocation.

It was then demonstrated how this EL leads to a demand for information by

consumers who would rationally be willing to pay for information an amount

equal to the reduction in EL that would be generated. Finally it was shown

that both producers and consumers incur costs in providing data and informa–

tion respectively. These costs place limitations on the amount of data

voluntarily provided and the amount of information processed.

By appealing to the model it is possible to explain much of the material

outlined earlier in the review section of this paper which indicated a high
 2.5

desire by consumers for information but little actual processing or usage

of such information. In terms of the model the high desire for nutritional

and other information is an expression of the demand curve for information,

BI The low actual usage of the data provided is explained when one con-
P“
siders the costs consumers incur in processing the data into information.

Another research result outlined earlier, that certain segments of the popula-

tion (those with low incomes, poor education and the elderly) report a lower

than average usage of information, can also be explained in terms of the

model. It would be expected that on average members of these groups would

incur much higher costs of processing information in terms of A (time required

to process) than other consumers, although this may be offset somewhat by a

lower y (opportunity cost of time). In extreme cases A would be very large

for those who were not literate in English, visually handicapped or un-

familiar with nutrition and nutritional jargon. Figure IV below illustrates

this situation. A consumer with a marginal cost of processing information

of MC2 would not process any information if BI was his/her information demand
P
curve. If the costs were lower, say MCI, the consumer would process and use

some positive amount, 01 of information,

1’

Figure IV
‘IA
B‘ ~MC2

!
I
t

o 11 I QIA
P
 26

In terms of policy any programs designed to increase consumers’

nutritional literacy or to present the label data in a more easily process-

ible format will lead to consumers processing more information which will

reduce EL, a socially desirable result. A preliminary summary on the joint

hearings conducted by FDA, FTC and USDA reports senior citizens, “individuals

of low comprehension”, non-English speaking people and children were very

much in favor of proposals to present the label data in graphics form

(USHEW, 1978). In terms of the model this can be seen quite simply as an

appeal by these people to lower the cost of processing information. Another

conclusion from the hearings was that those with special dietary needs were

most willing to incur extra costs to obtain information. This would be ex-

pected since these individuals would have a higher EL and consequently a

higher demand for information.

III. An Application of the Model

A. Background

The model and theory developed in this paper are meant to be generally

applicable to all types of information and data pertaining to food products.

The type of data most frequently discussed thus far is nutritional data, but

unfortunately at the present time the author is not aware of any nutritional

data that would enable a direct application of the model. A type of product

information that can be analyzed, however, is weight information. Some foods,

specifically meats and poultry, tend to accumulate moisture in the packing

process, If the weight of this moisture is included in the weight of the

product appearing on the package label, the label weight will tend to over-

state the true amount of product contained in the package.

 27

Proposed federal regulations (USDA,19791_)

will require that on average

the drained weight of the product (the product minus accumulated moisture-–

essentially the amount of pure product) be equal to or greater than the label

weight. However, for any particular package, depending on the amount of

moisture accumulation, the label weight may be greater or less than the drained

weight. Most states allow a “dry tare” approach to be used in determining

label weight. With this method the label weight is determined by subtracting

the weight of the packaging material (the dry tare) from the weight of the

total package, and this results in the moisture being included in the label

weight. A few states (California, Washington, and Michigan in particular)

require that the label weight be based on the drained weight of the product.

In any case the actual drained weight of the product in a package may be more

or less than the weight specified on the label. Thus the consumer of any

given package may receive more or less a pure product than he/she had anticipated

based on the labeled weight. This is a problem of imperfect information quite

suited to analysis within the model developed previously.

The data to be analyzed consist of 406 packages of chicken sampled by

USDA researchers at retail outlets across the nation. Researchers obtained

the label weight for each package, then drained the moisture from each one

and reweighed it enabling them to measure the amount of moisture accumulation

in each package. The data were collected by USDA (19792)inseveral states

including some states which require drained weight labeling and some which

allow dry tare labeling. With this data it is possible to obtain an estima-

tion of loss due to imperfect weight information and derive a rough representa-

tion of the benefit to consumers from improved weight information.

 28

B. Empirical framework

For each of the 406 packages of chicken we have labeled weight (WL),

drained weight (WD) and price per labeled lb. (P~~). From each package using

P’~and WL a point in Euclidean two space somewhere on the demand surface can

be obtained. This is shown as point F on Figure V.

Figure V

Price \
per \
labele
lb.

o ~
w. Quantity (lbs.)
L

To apply the model to this data it is necessary to be able to estimate

the triangular areas of consumer loss GEF and FHJ as illustrated on Figure V

and developed earlier in Figure I. No information exists about the demand

functions of individual consumers purchasing chicken. However, assuming the

packages of chicken sampled are representative of packages of chicken sold,

a demand curve going through some point such as F can be assumed to exist as

indicated on Figure V. The demand elasticities used in this study are the

national market price elasticities of consumer demand for chicken estimated

by Brandow (1961) and George and KinS (1971),

 29

In outlining the empirical framework assume for convenience that the

consumer has perfect information about the nature of all the product’s attri-

butes except weight. In the case where WD > WL, given the above assumption,

the consumer will be underevaluating the product (possibility 2, page 10)

since there is more of the product, and hence more of the product’s attributes,

contained in the package than the consumer has been led to believe from the

information available–-namely the labeled weight. The loss to the consumer

from this underevaluation is FHJ. To estimate FHJ, HF and FJ must be found.

The procedure used to find HF is as follows:

1.0 WD - WL = WS, where WS is in this case the surplus amount of product

the consumer receives over and above the amount specified on the label. When

WD < WL, WS is the amount by which the product is deficient.

The consumer should be willing to pay some amount for WS. Assume he/she

is just willing to pay P~~for an amount up to W Therefore, the consumer is

L“
willing to pay less than P’”for amounts greater than W~ since the demand

curve is downward sloping. Employing this concept it is possible to arrive

at a formulation to measure the consumer’s additional willingness to pay ~~

labeled pound. This then is HF.

——— The reader should consult the appendix for

a more detailed exposition of the steps taken in arriving at HF

p9<w w~
1.1 HF= ~fi (2 -~w—) where Ec = the elasticity coefficient
L L from the assumed demand
relationship (see p. 28).

FJ is found as follows:

2.0 piiw = purchase price/pkg.

L
p$,~~
2.1 -—-~= effective pricellb,
‘D
 30

p$<w
2.2 +- P~~= effective change in price/lb.
D

When WT # Wn the label price per lb. and the actual or effective price per
L, u

lb. are not the same. The result is an effective price change. When

WD > WL the result is an effective price decrease as indicated in step 2.2.

A price change will always lead to a change in quantity demanded (Q) for any

product whose demand elasticity coefficient is other than O. FJ is the addi-

tional amount of product the consumer would purchase given the effective price

decrease. Solving for the change in Q will yield FJ. This is done by employ-
AQ P
ing the elasticity formula, Ec = making the appropriate substitutions
AP Q’
WL
(AP = P$’(w—- 1), P = P$’,Q = WL) and solving for AQ. When simplified the
D
following expression is obtained (the steps in arriving at FJ are also de-

lineated in more detail in the appendix):

w
2.3 FJ = AQ = EcWL(~L - 1).
D

To estimate the area of FHJ the triangle area formula is employed to

obtain:

3.0 LFHJ = l/2HF(FJ) where L = the loss from underevaluation.

FHJ

The opposite case a consumer can encounter is where WD t WL. Here

the consumer will be overevaluating the product (possibility 1, p. 10). Based

on the labeled weight she/he will think the product contains more than it

actually does. The loss from overevaluation is GEF (Figure V). The method

employed to estimate this loss is completely analogous to the technique

outlined for estimating FH.Jand hence is not delineated here. Let it suffice

to say that had the consumer known the weight deficiency, he/she would have

been willing to pay less per labeled pound for the product enabling FG to
 31

be obtained. Also the weight deficiency results in an effective price

increase which would lead to a decrease in Q enabling one to obtain EF. The

loss from overevaluation is therefore:

3.1 = l/2EF(FG) where L = the loss from overevaluation.

‘GEF GEF

c. Empirical results

In this instance the equation to estimate a loss from overevaluation,

is the same as that used to estimate a loss from underevaluation, L

‘GEF‘ I?HJ
‘
so a single equation can be employed to estimate the losses from imperfect

weight information. The equation obtained through combining and simplifying

the expressions outlined in 1.1 and 2.3 is

4.0 L = ~_‘s (EL.- 1) (2EC - ;).

WD
L

See the appendix for the algebra involved in the simplification.

Six runs of the analysis were made. The results are summarized in

Table 1. Results were obtained for the entire sample using (a) Brandow’s

elasticity estimate (Ec = -1.16) and (b) George and King’s estimate

(Ec = -.78). Results were also obtained by dividing the sample into two

parts, data from states which have drained weight labeling regulations and

data from those states which do not have such regulations. These subsamples

were each analyzed separately using both elasticity estimates. Two observa-

tions from the drained weight states subsample appeared to be outliers.

Both observations exhibited W. > Wn by a substantial amount due either to

L u

mislabeling or sampling error--although there was no evidence of such error.

If these cases resulted from mislabeling they represent legitimate consumer

losses and would be valid, although extremely influential observations. If

 ‘2 --I

Table 1

(1) (2) (3) (4) (5) (6)

Sample Numbers Numbers Loss to Total ex- Loss as a
Size of cases of cases Consumers penditure proportion
(n) (L) from sample of expenditure
;he:ew ~/ ;he:e#/
TE = P*WL LPE = L/TE
Run DLDL
— .—.— —— —

Full sample
Ec = -1.16 406 198 203 $0.3872 $1058.21 .0003659
(Brandow)

Full sample
Ec = -.78 406 198 203 0.2649 1058.21 .0002503
(Gee. & King)

Nondrained
weight states 123 7 114 0.1147 336.32 .0003410
EC = -1.16

Nondrained
weight states 123 7 114 0.0776 336.32 .0002307
EC = -,78

Drained weight
states 283 191 89 0.2725 721.89 .0003774
Ec = -1.16

Drained weight
states 2/ 281 191 87 0.1096 715.97 .0001530
Ec = -1.16--

Drained weight
states 283 191 89 0.1873 721.89 .0002594
Ec = -.78

Drained weight
states 281 191 87 0.07370 715.97 .0001029
EC = _ @
.

—
1/
–The observation in columns 2 and 3 may not total the sample size (column 1)
since in a few cases W =W
L D“
2/
‘-Influential cases dropped.
 33

the error was made in the sampling process the cases are, of course, invalid.

Since their exact cause is uncertain the results of the analysis for the

drained weight states with these cases omitted are also included in Table 1.

The first three columns of the table are self explanatory. Column 4

(loss to consumers) is the result obtained from applying equation 4.0 to

each observation and summin~ over all the included observations. Column 5

(total expenditure) is simply the amount consumers would have had to pay at

retail for the sum of the observations included in the particular run. And

Column 6 (loss as a proportion of expenditure) is obtained by dividing the

loss from each run by the total expenditure (column 5) for that run.

D. Interpretation of results

Caution needs to be exercised in interpreting the results of the appli–

cation due to the small size of the data set and some rather strong assumptions

necessary to fit this particular data set to the model. At first glance the

actual losses to consumers seem quite small, but when one considers the total

amount spent on chicken in particular and other meats and poultry in general,

these types of losses can be significant. For instance per capita chicken

consumption in the U.S. in 1978 was 47.7 lbs., the average price was $0.66/lb.

meaning that on average U.S. consumers spent about $31.72 per capita on

chicken in 1978. Taking the liberty of projecting the results from this

sample to these 1978 consumption figures, a crude estimate of the annual

loss from imperfect weight information on chicken can be obtained. Multiply-

ing the per capita expenditure figures by the approximate U.S. population

and then multiplying by the loss as a proportion of expenditure (LPE) figures

from column 6 of Table 1 gives a loss to consumers of about 2,550,000 dollars

 34

when LPE = .0003659 (full sample using Ec = -1.16). The loss is about

1,740,000 dollars when .0002503 is used as LPE (full sample usingEc. = -.78).

It should be noted that the procedure used to obtain these results

necessarily represents some departure from the model’s basic theory. Loss

from imperfect information is unique to each consumer based upon his/her

demand for the product, the information at each’s disposal, etc. The only

theoretically correct way to measure this loss is with the individual con-

sumer as was done in the model where individual demand curves led to obtaining

an EL for the individual which led to obtaining the MB of information for the

individual. Aggregation was done only in the final step where the individual

MBi’swere summed to obtain an aggregate benefit figure from information pro-

vision. In this application it was necessary to begin with an aggregate

demand curve and assume its applicability to the individual consumer. The

loss figures from the individual packages were then summed to obtain the

total loss (L) figure. Essentially the basic difference is chat aggregation

is utilized at the beginning in the application rather than only at the end

as the model mandates.

The figures obtained for dollar losses from imperfect weight informa-

tion also can be interpreted as the value to consumers of improved weight

information. Figures of this sort have obvious significance in terms of

cost-benefit measurement for programs designed to improve label information

to consumers. For convenience one can take the simple average of the two

loss figures obtained to give a single loss formulation, (2,550,000 +

1,740,000)/2 = 2,145,000. Figures of this sort could then be compared with

the costs incurred by sellers in providing new data and by consumers in

processing it to determine if there are net benefits or net costs generated

 35

by imposing the additional disclosure. In this particular example processing

costs to consumers are probably O since there would be no new information on the

label, only a more accurate representation of the product weight. So, again

assuming the loss figure $2,145,000 is somewhat valid, if processors could

provide improved weight information of the type discussed here at an annual

cost of less than 2,145,000, it can be concluded that it would be beneficial

to have them do so.

As another example it would be possible for all states to require a

drained weight labeling system as is done in some states. Taking the average

of the LPE’s obtained from the drained weight states with the influential

outlying cases dropped (rows 6 and 8 of Table 1) and multiplying by total

national expenditures gives an approximate annual loss of 890,000 dollars.

If such a system was imposed nationwide the reduction in loss would be

2,145,000 - 890,000 = 1,255,000 dollars. Therefore, if a drained weight

labeling system for chicken could be imposed nationwide at an annual cost

of less than 1,255,000 it would be socially beneficial to do so.

The results also indicate that in states which impose drained weight

labeling regulations overevaluation seems less likely to occur (note column

3 from Table 1) and when the influential cases are removed LPE becomes con-

siderably less in th(’sestates also. To test for significance of differences

between the two subsamples, the absolute values of the WS were summed for

each subsample. The mean value for w~I was .0569 lbs. for the nondrained

weight states and .0431 lbs. for the drained weight states. A t test for

significant differences between samp e deviation proportions produced a

t = 2.86, significant at a .01 level of confidence indicating that the mean

deviation be~ween ‘dDand WT was significantly less in states which had a

,
drained weight labeling systt~m.
 36

11 sllollld
also be noted that losses are greater when the larger elasticity

coeflicient is used. This result will be true in general since differentiating

the loss equation (equation 4.0) with respect to Ec gives

-w~p~c w
5.0 g-<= y— (WL- 1)
.
D

the value of which will always be positive for W~+o. The minus sign at the

front of the expression is due to Ec always being a negative term. When

>W WS is negative making the first term in the expression positive.

‘L D’
In the second term WL/WD > 1 makes that term positive also. When WL K WD,

WS is positive making the first term negative. The second term is negative

also since WIIWD

, i 1, so the signs cancel out.

[v, some i’t~l

i[y and Empirical Recommendations

The theoretical development in this paper indicates that in general

consumers can expect to incur some dollar loss in economic welfare in any

purchase situation in which they possess less than perfect information.

The more information a consumer processes the less will be the expected loss.

However, policy designed to simply put more and more product data before

consumers is not necessarily advisable. The costs to producers in generating

the data and to consumers in processing it must be considered. More attention

should be concentrated on the format of presentation in order to reduce these

costs and in turn encourage more provision and processing of data. k such

the following specific types of policy recommendations pertaining to food

labeling would seem to be in order. FDA should consider the effectiveness of

alternative presentation formats as a means of reducing processing costs.

Programs of nutritional education should be considered also as a means of

red[lcingprocwsing costs. Also consumers should be made aware of losses

 37

‘theymay incur f“romla(k of information. If they fail to recognize these

losses, they will not demand information. Finally, provision of still more

label data is not likely to be a good idea until consumers have learned how

to use the data currently at their disposal. Further provision at this time

could lead to information congestion and raise processing costs. No specific

recommendations can be made from the application of the model to the weight

data since no information is presently available on the costs involved in

providing more accurate weight data.

In terms of empirical analysis I would suggest that the following

questions need to be answered in the approximate order of their listing, As

will be noted work on some of these questions has already been done or is

c(lrrentl.y
underway.

1. What fooclshave l~utritionallabels and what percentage of total

food expenditure is on foods which provide nutritional labeling? Surprisingly

this is not known but FDA :isinvestigating the question.

2. Have nutritional labeling requirements directly or indirectly made

foods more nutritious or expensive?

3. Is there a relationship between price and nutrient content of

brands of foods?

4. Do most consumers have a desire for the type of data supplied

through labeling requirements? The model presented previously indicates

they should, and existing studies seem to indicate they do.

5. What percentage of data on labels is understood, and would different

formats of presentation increase unde.rstanding? Existing studies indicate a

large lack of understanding and much confusion regarding some of the data

currently provided. There is no substantial evidence regarding the effective-

ness of ~ll(ernativeprf~sentationformats.
 38

() . WI I;I[ {lr~~
th<’b(?ne[i~sand (:ostsassociated with speciEic types of

label disclosure pcograms? This paper developed a mechan.i.srn

for analyzing

the benefits from obtaining information and made an initial estimation of

such benefits with the weight application,

7. Has the provision of potential information on food labels altered

consumer purchasing decisions? This is a key question and evidence to date

suggests consumers will alter their buying decisions based upon this type

of information. The -——-—

Redbook survey (1976) reported 50% of women stopped

or cut back on purchases of a food because of high sugarp 39% due to high

cchol.e~rerol
,)nci
29% because of additives. However, no evidence has been

presented t~)+ow that provision of label data has significantly altered

demand for any specific food products,

8. What are the indirect costs and benefits which result from label

regulations, and how important are they? Many such effects have been postu-

lated, but discussion of them in this paper has been avoided in an attempt

to maintain some degree of conciseness.

v. Summary

The concern of this paper has been with product information in general

and nutritional information (labeling) on food products in particular. In

section I the studies cited indicated that most consumers do desire and would

be willing to pay for information of the type attainable from the standard

nutritional label which appears on many food products. These same studies,

however, indicated that often consumers were apparently not presently making

use of such data.

In section II a model was presented which developed a theoretical frame-

work for explaining and analyzing the information problem. It was shown that
 39

in the absence of perfect information about a product consumers incur some

expected dollar loss (EL) from expenditure misallocation. It was then demon-

strated that this EL leads to a demand for information by consumers who

would rationally be willing to pay for information an amount equal to the

reduction in EL that would be generated from it. Finally, it was shown that

both producers and consumers incur costs in providing data and processing

information respectively. These costs place limitations on the amount of data

voluntarily provided and the amount of information processed.

In section III the model was applied to a specific case of imperfect

information. Data indicated that the labeled weight for packages of chicken

could be either greater or less than the actual drained weight of the product.

It was shown that consumers would incur a loss in either case.

 40

VI. Apl)f
’11(1
ix

A. Computations to obtain equation 1.1.

Of interest is the consumer’s additional willingness to pay for WS.

This willingness to pay is shown by the shaded area in Figure Al. This area

Figure Al.

Price
per
labeled \
lb. \
\
p>k
p*.~p ___
R

w
L
WD Quantity (lbs.)

is found by computing the average of areas WLFPW and WLRSWD, WLFPWD = P~fWS
D
since W =W-w To find WLRSW it is necessary to solve for the change
s D L“ D
in pri((’,,AP.which is (Iqu;Il
to 1’Sfrom Figure Al. This is done by solving

A(JP
L 11(’ el[jsticit:!
f“ormula, ‘iorAF’. Solving for P gives AP = 4!!!1.
“ = APW’ Ec Q
w p*
Now substituting in WS for AQ, WL for Q and P~;for P yields AP = ~~ .
L
Wsp$c
The area W RSWD is then equal to (Pi<– –——— The average of the areas
L 13ch!,
) ‘s “
WLFPWD + WLRSW p*ws + (pi,_ W:p*/EcWL) WS
D .
is then . Factoring out P*
2 2
WLFPWD + WLRSWD w pk
Ind Ws yields ‘s
--:–- (2 - -F=). This gives the total
2
‘1
 41

willingness to pay for W In order to find how much more the consumer would
s“
be willing to pay ~labeled pound if he/she knew that the actual weight

of the product was WTJ + W~, divide,the expression through by WL giving

P$;w
s
1.1 (2 -&-)
‘F = mL—
L

where HF is the additional willingness to pay per labeled pound as indicated

in Figure V.

B. Steps to obtain equation 2.3.

It is necessary to solve the elasticity equation for AQ = EcAP ~,

where AP = P~’WL/WD- P’: (from expression 2.2)

P = P7~

Q=WL

Therefore, FJ = AQ = Ec(P’’’WWDWD
- Pi;)WL/P$:.

Factoring out P* and canceling yields

2.3 FJ = AQ = ECWL (!l

w _ 1)5
D

c. SICps to (~t)t(lin
(’qllation
4.0,

[nitially substitute equations 1.1 and 2.3 into the triangle area

formula:

p*w
‘s ‘L
L = 1/2 [+ (2 - ~~)]
[EcWL (~ - 1)].
L L D
p$cw
Employ the associative law of multiplication and combine ~ with EcWL and
L
cancel the WL’S. This gives

p>kw
L=(T ‘S ‘L
‘) Ec (2-~~) (~-l).
LD

Multiply the middle term through by Ec and cancel, which yields

p>?w. w
‘s
3.0 L = (-+ (WA - 1) (2Ec - ~).
1) L
 42

REFERENCES

J..J.,“l’heCost of Government Regulatj.onsto the Food Industry,”

A1.brecllt,
Food Technol.o~,
_._.— 29(10):61, 1975.

Assam, Edward and Louis Budklin, “Nutritional Labeling for Canned Goods:
A Study of Consumer Response,” Journal of Marketing, 37, April 1973,
pp. 32-7.

Babcock, M.J., A. Burkart H. Guthrie, A. Lictenstein and C. Thoroughgood.

Consumer Use of Nutrition Information on Food Labels. New Jersey
Agricultural Experiment Station Bulletin 344, September 1975.

Becker, Gary and R. Michael, “On the New Theory of Consumer Behavior,”
Swedish Journal of Economics, 75, December 1973, pp. 378-96.

Beloin, A., “Nutritional Labels: A Great Leap Forward,” FDA Consumer, Summer
1973, pp. 10-16.

Berning, C. and Jacob Jacoby, “Patterns of Information Acquisition in New

Product Purchases,” ——.—
Journal of Consumer Research, September 1974,
pp. 18-22.

Better Homes and Gardens. Special Food

—. Survey, 1978.

Bettman, J.R., “Issues in Designing Consumer Information Environments,”

Journal of Consumer Researc&, December 1975, pp. 169-77.
—.

————— and P. Kakkar, “Effects of Information Presentation Format on Consumer

Journal of Consumer Research,
Acquisition Strategies,” ————. ——— March 1977,
pp. 233-40.

Brandow, G.E. Interrelationshi~sAmong

.—. — Demands
—.—— for Farm Products and
Im~lications for Control—of Market Supply. Agricultural Expe~ment
Sta~ion Bulle~in 680, ~~n~s~~vania State University, 1961.

Daly, Patricia, “The Response of Consumers to Nutritional Labeling,”

Journal of Consumer——.——
———.——.— Affairs, 10:2, Winter 1976, pp. 170-8.

Friedman, Monroe, “Consumer Use of Informational Aids in Supermarkets,”

Journal
—.——..— of Consumer Affairs, 11:1, Summer 1977, pp. 78-89.
.————..

Forbes, Alan, “Nutrition Labeling from the Government Point of View>”

Food Technolo~,
—— .-——.— December 1978, pp. 37-41.
General Foods Corporation. The Im~act of Government Regulation on General
——
lJood:>,
.- 1975.

General Mills, Inc. .—...—.

A Summary_Re~ort on [J.S.Consumer Knowledge, Attitudes
.-.—.———___
and Practices About Nutrl~ion, 1977.

George, P.S. and G. A. King. ——

Consumer
.— Demand for Food Commodities in the U.S.
With Projections for 1980, Giannini Foundation Monograph 26, University
of California, Berkeley, 1971.
 43

Jacoby, Jacob, “Information Acquisition Behavior in Brand Choice Situations,”

Journal of Consumer Research, March 1977, pp. 209-16.

, R. Chestnut and W. Silberman, “Consumer Use and Comprehension of

Nutrition Information,” —.—
Journal of Consumer Research, September 1977,
pp. 119-27.

Lancaster, Kelvin, “A New Approach to Consumer Theory,” Journal of Political

Economy, 74, April 1966, pp. 132-57.

Lenahan, R.J., J,,A.Thomas, D.A. Taylor, D. L. Call and D.I. Padberg. Consumer
Reaction to Nutrition Information on Food Product Labels. Search 2,
No. 15, Col~nellUniversity Agricultural Experiment Station, 1972.

Liechtenstein,A,H., “Nutrient Labeling: An Investigation of Formats and

Consumer Attitudes~’M.S. Thesis, Pennsylvania State University, 1973.

Nelson, Phillip, “Information and Consumer Behavior,” Journal of Political

Econo~, 78, March/April 1970, pp. 311-29.

Peltzman, Sam, “An Evaluation of Consumer Protection Legislation: the 1962

Drug Amendments,” ———
Journal of Political
— Economy, 81:5, September 1973,
pp. 1049-91.

Redbook Magazine. —.——

Nutrition,
——.—— The Redbook Nutrition Report, 1976,
Rusoff, Irving, “Nutrition Labeling from the Industrial Point of View,”
Food Technolom, December 1978, pp. 33-6.
——.-—.——_..——

Scammon, Debra, “Information Load and Consumers,” Journal of Consumer

Research, December 1977, pp. 140-8.
.--——-.——-

Stanley, T.J., “Nutritional Information and Preferences for Breakfast

.Journalof
Cereals,” -.-.—— Consumer Affairs,
—-.._.—__ -.—.—— 11:2, Winter 1977, pp. 121-7.

Us. Department of Agriculture. ——..

National Food
———..~—~z Situation, June 1977,
pp. 17-25,

.--.——
.. “Solicitation of ~nl[~rmation: Net Weight Labeling,” ——
Federal
R~gister. /+4:]71,August 31, 19791,pp. 51275-7.

“ASS6?SSlTknt of Prop(~~cdNet W<i@lt labeling Regulations,” Staff

R~port, ESCS for Food Safety and Quality Service, August 19792.

Us. Department of Health, Education and Welfare. Food Label HearIi,

——
Background
_._...
___. Information,
__ ——— —._ 19781.
Food Label
—-——— Hearings, Summar~..—
_—.——_.—.—_—.- Re~ort, 19782.

——— FDA Consumer Nutritional

————-— Knowled~
—-.—-——- Survey Report
.— .— —.- II, 1975.

.——-—— ~ “Nutritional Labeling,” Federal Register, 38:13P January 19, 19731,

._._—————

FDA —_.
—._. . .—_.
..——_. Consumer Nutritional Knowle&
-.— — ——._———_.—.___— . survey
.—— Repyt.—.—1, 19732.
 44

Wall Street Journal. “Effect of Cost Increases Due to Labeling,” February 20,
1973, p. 7.

White House Conference on Food, Nutrition and Health. Final Report. U.S.
Government Printing Office, 1970.