...2 L:
..C. .12..

 

 

 

 

3 1293 01770 4275

”55’s IllllmllfllillHlIHHllIlllHll illllllllfllflllllllllHlfl

This is to certify that the

dissertation entitled

Consumer Demand for Ecolabeled Apples

presented by

Jeffrey Raymond Blend

has been accepted towards fulfillment
of the requirements for

Ph.D. mgmem Agricultural Economics

 

 

fiw

Major professor
DateW/ ’6 /‘7 7 3/

MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771

 

LIBRARY

Mlchlgan State
University

 

 

 

 

PLACE IN REFURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

8gp. H Q7.

 

 

a?
(:99
69

~ 6’

 

T9

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/” W14

CONSUMER DEMAND FOR ECOLABELED APPLES
By

Jeffrey Raymond Blend

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Agricultural Economics

1998

ABSTRACT
CONSUMER DEMAND FOR ECOLABELED APPLES
By

Jeffrey Raymond Blend

An ecolabel is a voluntary claim that a product meets environmental
standards. It offers consumers the chance to lower the harmful impacts from
their consumption by buying products which are manufactured and grown in
ways that cause less environmental damage. Because ecolabeling is new, it is
uncertain whether consumers would buy such products and at what price.

This paper estimates average household demand and marginal
willingness to pay for ecolabeled apples. It tests several hypotheses about
variables which affect consumer demand for ecolabeled foods including the
comprehensiveness of the ecolabeled claim, the proof offered that it is honest
and various personal factors such as income and familiarity with the claim.

A theoretical model is developed which explains how ecolabeling
would affect a consumer’s typical apple purchasing choice. The model
asserts that the consumer faces a tradeoff between their utility from apple

consumption and their disutility from the environmental damage caused by its

production.

Data fi‘om a phone survey of U.S. households is used to estimate
consumer demand for regular (i.e. unlabeled) apples and ecolabeled apples in
markets with and withOut ecolabeling. Because ecolabeled apples are new, a
market scenario is created in which respondents were presented with prices
and descriptions of regular and ecolabeled apples and asked which they
would buy and how much they would buy in a single shopping trip. A Cragg
Double Hurdle model and a Tobit model are used to estimate demand for
each type of apple. The Cragg model appears to be the more appropriate
model for estimating demand for a single shopping trip.

The results indicate significant consumer demand for ecolabeled apples
and a willingness to pay a price premium for them. This result holds
regardless of how environmental claims are worded or whether proof of their
validity is stated in the label. Price is a very important determinant in
consumers’ decisions of whether to purchase regular and ecolabeled apples.
Respondents bought ecolabeled apples primarily for environmental concern
and health attributes. Respondents with more education and from larger

households bought more ecolabeled apples.

Copyright by
Jeffrey Raymond Blend
1 998

To my parents and my wife Becky

ACKNOWLEDGMENTS

Research for this paper was provided by USDA, ERS Cooperative
Agreement No. 43-3AEL-6-80060 and the Michigan Agricultural Experiment
Station.

I am deeply grateful to my parents for all their love, support and
contribution towards my current and future success in the world. They
instilled in me a good work ethic, solid morals and the ability to persist in the
face of challenging obstacles.

I thank my wife Becky for her love and patience. Her support greatly
contributed to the quality of this work and to my time in graduate school.

I am deeply grateful to my mentor and major advisor, Eileen 0. van
Ravenswaay. Her assistance, dedication and encouragement throughout my
graduate experience were extraordinary. She knew when to compliment my
work and when to urge me beyond the limits I had previously set. She spent
a great deal of time and effort developing my professional skills and boosting
my confidence in what I can accomplish. I have been blessed to have such a
dedicated advisor.

I would like to thank Jeff Wooldridge for his extensive help with the

statistical aspects of this paper. I thank my other two committee members,
John Hoehn and Patricia Norris for their time and genuine interest in my
work.

I extend my thanks to Takashi Yamano for his generous and valuable
assistance with the econometrics software used in this research. I thank
Frank Lupi and Scott Swinton for their support and interest. I thank Chris
Edge and David Jones for their help and great fiiendship as well as Lorie
Srivastava, Karin Stefl‘ens and Hirokatsu Asano. I am grateful to all of the
other researchers not mentioned who contributed in any way to this
dissertation.

My deepest respect and thanks to the art of Yoga and its invaluable
contribution to my life. I thank my Yoga instructors Ruth Fisk, Rob
Eschbach and Amar Gupta. Without my practice, this experience would have

been much more difficult and not nearly as valuable or insightful.

vii

TABLE OF CONTENTS

LIST OF TABLES ......................................... . ix
LIST OF FIGURES .......................................... xi
Introduction ................................................. 1
Theory ..................................................... 5
Stn'vey Design .............................................. 17
Sample Selection ............................................ 29
Model Specification ......................................... 31
Results ................................................... 37
Conclusions ................................................ 50
TABLES .................................................. 59
APPENDD( A-DEVELOPMENT OF ECOLABEL DESCRIPTIONS AND
PRICE ............................................... 89
APPENDD( B-SAMPLE SELECTION ........................... 97
APPENDD( C-SURVEY INSTRUMENT ........................ 99
BIBLIOGRAPHY .......................................... 1 l l

LIST OF TABLES

Table 1-Names of Ecolabels in the Questionnaire ................... 59
Table 2-Number of Respondents for Each Ecolabel . . . . . . . . . . . . . . . . . 59
Table 3-Household Composition ................................ 60
Table 4-Age of the Respondents ................................ 61
Table S-Educational Attainment of Respondents .................... 62
Table 6-Gender of Respondents ................................ 63
Table 7 -Household Income .................................... 64
Table 8-Q: Where do you usually purchase fi'esh apples? (Open ended) 65
Table 9-Q: When you buy apples, how often do you buy organic apples? 65
Table lO-Q: How much have you seen, heard, or read about Integrated Pest
Management or 1PM? ................................... 66
Table ll-Variable names, definitions, means and standard deviations . . . . 67

Table lZ-Percent respondents purchasing regular and ecolabeled apples when
both are available ...................................... 69

Table l3-Percent respondents purchasing regular and ecolabeled apples when
the price premium on ecolabeled apples is $_/lb. ............. 69

Table l4-Cragg Double Hurdle for ecolabeled apples in a market with
ecolabeling ........................................... 70

ix

Table 15-Cragg Double Hurdle for regular apples in a market with
ecolabeling ........................................... 72

Table l6-Cragg Double Hurdle for regular apples in a market without
ecolabeling ........................................... 74

Table 17 -Tobit coefficients and calculated marginal effects for ecolabeled
apples ............................................... 76

Table 18-Tobit coeflicients and calculated marginal effects for regular apples
in a market with both types of apples ........................ 78

Table l9-Tobit coefficients and calculated marginal effects for regular apples
in a market without ecolabeled apples. ...................... 80

Table 20-Calculated Tobit elasticities of prices, income and significant
variables for all three types of apples ........................ 81

Table 2l-Q: Why did you choose to buy the ecolabeled apples? (Open ended)
.................................................... 82

Table 22-Q: Why did you choose not to buy the ecolabeled apples? (Open-
ended) ............................................... 83

Table 23-Comparison of results between Tobit, OLS and SUR for ecolabeled
apples in a market with ecolabeling ......................... 84

Table 24-Comparison of results between Tobit, OLS and SUR for regular
apples in a market with ecolabeling ......................... 86

LIST OF FIGURES

Figure l-Consumer buys both types of apples in a single shopping trip . . . 12
Figure 2-Consumer buys ecolabeled apples only in a single shopping trip . 13

Figure 3-‘I'he Survey Versions .................................. 24

Introduction‘

Ecolabeling is a new phenomenon in agriculture. There are standards
for many other safety and quality attributes of agricultural products, but
environmental standards are just beginning to emerge. There are many
unanswered questions about what standards consumers want and at what
price.

An ecolabel is a voluntary claim that a product meets environmental
standards. It signals to consumers that a particular product has caused less
environmental impact than others in the same category. The purpose of an
ecolabel is to offer consumers the chance to lower the harmful impacts fi'om
consumption by changing the way products are manufactured and grown. On
agricultural products, ecolabels serve to promote innovative farming

techniques (e. g. Integrated Pest Management (IPM), cover crops) that result

 

1This section is adapted fiom Eileen 0. van Ravenswaay and J efi‘rey R.
Blend, “Using Ecolabeling to Encourage Adoption of Innovative
Environmental Technologies in Agriculture,” Flexible Incentives to Promote
the Adoption of Environmental Technologies in Agriculture, Frank Casey,
Andrew Schmitz, Scott Swinton and David Zilberrnan (Eds), Kluwer
Publishers, forthcoming, and Jeffrey R. Blend and Eileen van Ravenswaay,
“Consumer Demand for Ecolabeled Apples: Survey Methods and Descriptive
Results,” 1998, Department of Agricultural Economics Staff Paper No. 98-
20, Michigan State University, East Lansing, MI.

in higher environmental standards than what is already required by law.

Ecolabels are much like a seal of approval. They are awarded by a .
public or private nonprofit organization that establishes environmental
standards for product categories and certifies that products meet those
standards. Thus, an ecolabel is like a seal of approval because it is a signal of
high standards as well as a signal that products meet standards.

The increasing number of ecolabels in this country and others is
evidence of their growing popularity worldwide. More than 20 countries and
the Em'opean Community have adopted public ecolabeling programs to
encourage the development of manufacturing processes and products with
less environmental impact (U .8. EPA 1993a, 1993b, 1993c, 1994). In the
U.S., there are several private nonprofit ecolabeling programs (e.g., Green
Seal and Scientific Certification Systems) and voluntary environmental
labeling programs supported by the government (e.g., EPA Energy Star
Program).

The premise behind ecolabels is that environmental damage from
production and consumption of a good affects the consumers of that good.
Economists typically assume that environmental damage associated with

production or consumption affects some third party, not the producers or

2

consumers in the market. In the case of ecolabels, however, it is no longer
appropriate to analyze the problem as a producer extemality. The consumer
now faces a tradeofl‘ because her own consumption is the ultimate cause of
the environmental damage she suffers. Without an ecolabel, the consumer
has only one way she can reduce the damage, namely, to consume less. With
an ecolabel, she has the option of buying from a seller who has reduced the
damage for her.

Marketing research has found substantial consumer interest in green
products in the U.S. The most relevant marketing study comes from the Food
Marketing Institute (FMI, 1997) which identified consumer segments very
interested in buying green goods and paying more for them.2 There is little
research, however, on the potential demand for ecolabeled products.

This dissertation estimates potential consumer demand for ecolabeled
apples. The main research question is whether consumers would buy
ecolabeled apples and how much they would pay for them if they were
available at their normal place of purchase. Apple demand with and without

ecolabeling is estimated to measure the effect of introducing ecolabeled

 

2One quarter of their sample was identified as being “very interested”
in pm'chasing green products and 8% were identified as being “very
interested” in paying a 10% price premium.

3

apples. The results have implications for producers and retailers interested in
growing and selling ecolabeled foods. In particular, they gain a better idea of
whether a suflicient market exists for their products.

This dissertation examines whether demand and marginal willingness
to pay depend on how the ecolabel is worded. Two important dimensions of
that wording are the comprehensiveness of the ecolabel’s environmental
claim and the form of proof given to support the claim. The results of this
inquiry will give ecolabelers an indication of whether certain ecolabels would
perform more effectively than others.

The dissertation examines personal factors that affect ecolabeled apple
demand. Such factors include income, household size, education, one’s
primary motivation for purchasing the ecolabeled apples and one’s familiarity
with the label claim. This information is important to marketers because it
pinpoints those segments of the population most likely to buy ecolabeled food
items and provides potential selling points to use in marketing the ecolabels.

Finally, this dissertation tests the validity of several aspects of survey
instrument design. This is done by examining whether apple purchase
location, the existence of organic apples and the fact that some surveys were

given after apple season affect demand. Such information may help those

4

who construct similar surveys.

Data from a phone survey of U.S. households is used to estimate three
difi‘erent apple demands: (1) regular (i.e. unlabeled) apples in a market
without ecolabeling, (2) regular apples in a market with both regular and
ecolabeled apples and (3) ecolabeled apples in a market with both regular
and ecolabeled apples. Cragg Double Hurdle and Tobit models are estimated
to determine the factors that affect demand for each type of apple.

TheoryJ

In standard economic theory, uncompensated environmental damage is
usually treated as a negative production or consumption extemality. In other
words, the environmental damage is assumed to be home by a party external
to the product market who is unable to seek compensation from the market’s
participants. Consequently, the damage is not accounted for by consumers or
producers in the product market and not reflected in the equilibrium price and

quantity.

 

3This section is adapted from Eileen 0. van Ravenswaay and Jeffrey R.
Blend, “Using Ecolabeling to Encourage Adoption of Innovative
Environmental Technologies in Agriculture,” WW
tho Adoption of Environmental Technologies in Ago'culture, Frank Casey,
Andrew Schmitz, Scott Swinton and David Zilberrnan (Eds), Kluwer
Publishers, forthcoming.

The premise behind ecolabels is that some of the uncompensated
disutility fi'om the environmental damage associated with production or
consumption of the product is experienced by the market’s consumers. For
example, suppose production or consumption of a product results in wastes
that harm the environment. If the consumer knows about this environmental
harm and believes that she will suffer fi'om it without any compensation, she
will experience disutility from consumption. Thus, she faces a tradeoff
between the marginal utility she derives from additional consumption and the
marginal disutility she derives fi'om additional uncompensated environmental
damage.

Why might consumers be aware of the environmental damage from
their own consumption? There is much greater appreciation today of the
interdependency of ecosystems elements. Environmental science shows us
that what we once regarded as separate, independent elements of the
ecosphere are in fact highly interdependent. Thus, a change in one element
may have many indirect effects that are only now coming to be appreciated
and, thus, are not accounted for in the current set of property rights governing
goods traded in markets. Also, there has been extensive media coverage of

environmental problems resulting fi'om the production and consumption of

6

goods in this society. Thus, consumers are learning that money is not the
only thing they sacrifice to acquire goods.

Assume a typical apple market in which the consumer’s preferred
varieties are available. The consumer derives utility from apple taste and
nutritional value. Assume that the production of the apples causes
environmental damage. Any damage of which the consumer is aware causes
her disutility.4 Thus, she faces a tradeoff between the marginal increase in
utility she derives from the food attributes of an additional unit of
consumption and the marginal decrease in utility from uncompensated
environmental damage. Assuming all other food attributes remain unchanged
and that all production results in some type of uncompensated environmental
damage to the consumer, the consumer’s problem can be expressed as:

(1) Maximize U(Y,X,E(X))
s.t. PyY + PXX=M
where U is a quasi-concave utility function, X is quantity of apples

purchased, Y is the quantity of all other goods purchased (none of which

 

‘Because people are not omniscient, they will probably not be aware of
all the damage caused by apple production and thus will not sufl'er disutility
from that which they do not know. Of course, this assumes that the damage
they do not know about does not affect them.

.7

affect consumer health or safety, environmental quality or the amount
purchased of X), E is environmental quality that affects the consumer, PX is
the price of apples, Py is the price of Y and M is income. The efl‘ect of E on
utility is strictly positive (aU/dE>O). Since apple production uses resources
and creates waste, we assume that the effect of X on E is negative
(6E!6X<0). Because the earth has a regenerative capacity, however, this
expression is rewritten as dE/dXsO to reflect the possibility that apples may
be grown in a sustainable way that does not cause any environmental damage
in the long run. Thus, the marginal utility of X may be positive or negative
depending on the relative magnitude of the direct (6U/6X>0) and indirect
(dU/dE-dE/dXsO) effects of X on utility.

This model can be used to capture the effects of environmental damage
caused during either production or consumption of a good. Since we are
interested in the potential effect on producer adoption of environmental

technologies, we only consider the producer-related damage here.5

 

5Many of the consumer extemalities associated with a product involve
the use of complements such as waste disposal services, energy use, and
water use. This can be incorporated into the specification of the production
function for environmental quality, E. The main result is that the producer
will refonnulate the product if the difference in marginal cost is less than or
equal to the difference in marginal willingness to pay.

8

Suppose that some producers adopt an innovative environmental
technology that does not change any of the food attributes of X (i.e., no ‘
changes in apple safety or quality) but reduces the amount of environmental
damage affecting the consumer per unit of consumption. Suppose they
develop an ecolabel to advertise the environmental claim to consumers. Let
X’ be the quantity purchased of this ecolabeled version of X. The efi‘ect of
X’ on E is non-positive (BE/6X50) and E(X)<E(X’). Because X and X’
difl’er only in their environmental attributes, it is assumed that they are
substitute goods.

Assume for the moment that the firrn’s claim about the relationship
between X’ and E compared to that between X and E is truthful, and that the
consumer is aware of and fully understands it. Assume that the
environmental technology raises the marginal costs of production. Because
marginal costs are higher, X’ is sold at a higher price Px’. The consumer’s
problem becomes:

(2) Maximize U(Y,X,X’,E(X,X’))
s.t. PYY + PXX + PX’X’ =M
Consumers do not purchase negative amounts of goods, so X,X',Y20.

Since by assumption dU/dX = dU/BX’, the first order conditions imply

9

that if

(3) Px’- PX > {dU/dE(6E/6X’ - dE/dXfiA ,

where A is the Lagrangian multiplier, the consumer will not purchase any X’.6
In other words, if the marginal cost of making a truthful environmental claim
is greater than its marginal value to consumers, no ecolabeled apples will be
purchased. The first order conditions also imply that if

(4) Px’- Px < {aU/6E(aE/6X’ - dE/dX)”. ,

the consumer will not purchase any X. If the marginal cost of making a
truthful environmental claim is less than its marginal value to consumers, only
ecolabeled apples will be purchased if apples are purchased. No regular
apples will be purchased. Finally, if

(5) Px’- Px = {BU/6E(dE/dX’ - dE/dX)”. ,

the consumer will purchase either all of X, all of X’ or some of both apples if
they purchase apples. Because the marginal cost of the claim is equal its

marginal value to consumers, they are indifferent between the two types of

 

‘IfX and E are additively separable and corner solutions between X
and X’ arise, the first order conditions for a maximum are:
6U/6X + 6U/6E°6E/6X - lPx s 0 and if dU/dX + aU/dE-dE/6X - APX < 0
then X=0,
dU/BX’ + aU/6E°dE/6X’ - APx's O and if dU/dX’ + aU/dE-BE/BX’ - lPx’ < 0
then X’=O,
M-PxX-PX’X’ = O.

10

apples.

The necessary condition for ecolabeled apples to be purchased is that
the difference in marginal costs of the new method of production (reflected in
the price) does not exceed the marginal value of the environmental
improvement to the consmner of the last unit sold. This condition is
illustrated in Figure l on the next page.

In Figure 1, the individual’s demand for X in a single shopping trip is
represented by curve D, and the demand for X’ in a single trip is represented
by D’. Demand for X’ is greater than X by the marginal value of the
environmental improvement to the consumer of the last unit sold. The
marginal costs of supplying X and X’ are represented by MC and MC’.
Marginal costs are constant and equal to price because the individual is
assumed to be a price taker in the apple market. For the individual in Figure
1, the difference between the marginal costs is less than the difference in
demand. Consequently, consumer surplus is greater with X’ and some
ecolabeled apples are bought.

The way the curves are drawn in figure 1, some of both apples are
demanded by the individual. We might expect that in a single shopping trip,

however, many consumers would buy only regular or ecolabeled apples or

11

at. «3.32.» 0.3% a 5 8:3: be 8%. 52— 9::— uefiameew— 0.5»:—

EE 295 a E 83% B: 33:35

   
 

.
_
_
.
.

X

——-—--_-——(

X

 

—‘--——-—

 

/ Owe/Tm

02" s m

 

8?.

12

at. «5.32% 29:» a a. be: 8.9... 3.2380 n»...— uefiameeUé 953%.

EB 29% a E 33% 3: 52525

X

 

s
_
_
_
_
_
_
_
_
_
_
_
_
_

USTm

 

 

  

UEH‘ AH

35V

8?—

 

l3

none at all. Ifan individual’s demand for regular apples (D) in a single
shopping trip is entirely below the price for regular apples (i.e., MC), a
situation which is represented in Figure 2 on the previous page, they would
not buy regular apples during that particular shopping trip. The way the
curves are drawn in Figure 2, the individual would only buy ecolabeled
apples. Similarly, if an individual’s demand for ecolabeled apples in a single
shopping trip (D’) was entirely below the price for ecolabeled apples (i.e.,
MC’), the individual would not buy ecolabeled apples during that particular
trip. Ifboth curves were entirely below their prices, the individual would not
buy either type of apple.

Consumers are not omniscient and it is costly and difiicult for them to
observe whether a producer has truly improved environmental quality.
Knowing this, some producers may develop a false environmental claim in
hopes of receiving the higher price Px’ without incuning the additional
marginal costs from the use of environmental technology. This results in an
asymmetric information problem between producers and consumers, a
problem which will decrease the effectiveness of ecolabel markets if not
addressed.

Asymmetric information implies that the consumer does not know the

14

probability associated with the ecolabel claim being truthful. Assume that
consumers perceive a rt probability that the environmental claim is true. The
closer rt is to one, the more trust there is that the claim is honest. This is
incorporated into the model by weighting the E production function by rt.
Thus, (2) becomes:
(4) Maximize U(Y,X,X’,rt-E(X,X’))
s.t. PYY + PXX + PX’X’ =M, X,X’,Y_>_O

where the producer’s claim is represented by the E function and the trust in
the claim is represented by the probability weighting function rt. The greater
the value of rt, the more likely consumers would buy the ecolabeled apples.

Producers may attempt to increase consumer trust in the claim by
developing and offering credible proof about their environmental claim.
There are various forms of proof that a producer can provide to increase
consumer trust. These include satisfying or complying with national
standards such as those being developed for organic products, using a third-
party labeler/certifier who is widely known for being accurate and truthful,
making a self-declaration or giving no proof at all. Providing credible proof
increases marginal costs of production and therefore increases Px’. It also,

however, increases the probability rt and thus benefits to consumers.

15

Actions taken to reduce environmental damage may also improve food
safety-primarily in terms of reduced chemical residues. Therefore, we must
account for both the value of the environmental improvement and the food
safety improvement to be able to say what the value of ecolabeling is.
Assume that consumers perceive ecolabeled apples to be safer because of
lowered levels of chemical residues per apple.7 We denote R as the level of
residue one ingests from consuming a unit of X or X’. Because chemical
residues will be reduced only if the environmental claim is true, R is also
weighted by rt. Assume that actions taken to lower environmental damage
are perceived to simultaneously lower residues. Because these two effects
are simultaneous in the consumer’s perception, rt is the same for both. Thus,
consumers will receive both improved environmental quality and less residues
with the same level of probability.8 We have:

(5) Maximize U(Y,X,X’,rt-E(X,X’),rt-R(X,X’))

s.t. PYY + PxX + PX'X’ =M, X,X’,Y20

 

7We do not include improved nutritional attributes within the safety
variable because no respondents mentioned these as a reason for buying
ecolabeled apples.

1'There are health and food safety attributes other than pesticide
residues, some of which might have a different rt fi'om environmental
attributes. However, those attributes are not included in this model.

16

R has a strictly negative effect upon utility (dU/dR<0), R(X)>R(X’), and X
and X’ have strictly positive effects upon residues ingested so that (GR/6X>O)
and (dR/aX’>O).

The demand functions for regular and ecolabeled apples in a market
where both are offered for sale are:
(6) X = X(Px,Px’M,EJt.R)

X’ = X’(Px,Px’,M,E,rt,R)
Own price is expected to negatively affect X and X’. Substitute price, M, E,
rt and improved health attributes in the form of lower levels of R are expected
to positively affect both X and X’.
Survey Design9

The potential demand for ecolabeled and regular (e. g. unlabeled)
apples is estimated using data from a phone survey of a random sample of
U.S. households. The questionnaire elicited apple purchasing intentions with
and without ecolabeling. (The questionnaire is in Appendix C.) The

respondents interviewed normally do the food shopping for the household.

 

9This section is based on Jeffrey R. Blend and Eileen van Ravenswaay,
1998, “ Demand for Ecolabeled Apples: Survey Methods and Descriptive
Results,” Department of Agricultural Economics Staff Paper No. 98-20,
Michigan State University, East Lansing, MI.

17

Because ecolabels are new, a simulated market was created within the
questionnaire. This technique was first used by van Ravenswaay and Hoehn
(1991a) to estimate consumer willingness-to-pay for reduced pesticide
residues on apples. In that study, respondents were asked the amount of
apples they would purchase at various prices, pesticide residue levels and
pest damage levels. Because these choices were not widely available to
apple shoppers at the time of their study, a simulated market was created to
measure respondent purchasing intentions as if the choices did exist in a
typical apple market. Other studies examining similar consumer choices with
respect to food purchases have also used this technique (Born, 1994, Buzby,
Skees and Ready, 1995, van Ravenswaay and Wohl, 1995, Wohl, van
Ravenswaay and Hoehn, 1995, Owens, Swinton and van Ravenswaay, 1997).

In the simulated market scenarios, respondents were presented with
descriptions and prices of regular and ecolabeled apples and asked which
they would buy and how much they would buy. Questions were asked in
terms of household purchases during a single shopping occasion in late
autumn and early winter.

To simulate markets with and without ecolabeling, two market

scenarios were presented to respondents. The first involved a market in

18

which the respondent’s preferred or regular apples were available at various
prices. The second market scenario was identical except that both regular
and ecolabeled versions were available. The quality and display of the apples
were the same for both market scenarios.

In the first market scenario, respondents were asked to imagine
themselves in their typical apple purchasing setting and were given a
randomly selected price per pound for regular apples (PX). They were told
that the prices of other fruits were what would normally be expected, that all
apples were the same price per pound no matter how displayed and that
apples were not selling for less at other stores. Thus, PY was held constant.
Respondents were also told that all apples were available in their favorite
varieties, qualities and sizes. In this way, the effects upon apple demand of
substitute fruits, packaging, competitor prices and non-environmental
attributes were held constant.

Given the market scenario and price (PX), respondents were asked
whether or not they would buy the regular apples. If the answer was yes,
they were asked how much they would buy at the given price. They were

allowed to give their answers in terms of individual apples, pounds, peeks and

19

bushels although all answers were converted into pounds.10 The respondent’s
answer to this second question is the quantity demanded of regular apples (X)
when these are the only brand available.

In the second market scenario, the ecolabeled apple was introduced. A
verbal description of its attributes was given. The price of regular apples (PX)
was the same as in the first market scenario. The price of ecolabeled apples
(Px’) was equal to the price of the regular apples plus a randomly assigned
price premium. Given prices for both apples and the ecolabel description,
respondents were then asked which and how much they would buy at the
given price(s). The respondent’s answer to this second question is the
quantity demanded of regular (X) and ecolabeled (X’) apples when both are
available.

To test the effects of different types of ecolabels on purchasing
intentions, four ecolabel descriptions were randomly varied across
respondents. (A detailed description of how the ecolabel descriptions were
developed is in Appendix A.) The ecolabel descriptions varied in terms of

two dimensions: comprehensiveness of the environmental claim (i.e.,

 

loOne medium sized apple is equal to 1/3 pounds, one peck is equal to
10 ‘/2 pounds and one bushel is equal to 42 pounds.

20

information about E(X’) relative to E(X)) and proof substantiating the
environmental claim (i.e., information about rt). Two variations of each of
these dimensions were incorporated into the descriptions of the ecolabels.
The four ecolabel names are shown in Table 1. (All tables are located at the
end of the paper.)

There were fewer environmental claims made in the IPM (Integrated
Pest Management) label because it specified production practices for pest
control only as opposed to the wider range of practices that could reduce
environmental impact. The ECO claim was more comprehensive because it
specified production practices which lower a number of farm-related
environmental impacts. These included efficient water and energy usage, the
use of naturally occurring fertilizers for better water and soil quality, natural
insect controls and reduced pesticide use. It was expected that consumers
would purchase greater amounts on average of ecolabeled apples with a mere
comprehensive claim. The reason was that the more comprehensive claim
specified a greater number of environmental attributes suggesting a greater
improvement in E.

Proof was described in terms of whether the ecolabel claims were

externally verified by a certifying organization. The ECO and IPM ecolabels

21

were either described as being certified by the USDA or not at all.
Certification by the USDA rather than a private agent or a non-profit
environmental group was chosen because USDA is familiar to most food
consumers. Other types of certifying organizations could be specified, but
this would require a larger sample size than our research funding permitted.
It was expected that consumers would buy a greater amount on average of
ecolabeled apples with a certified claim since this would increase their trust
(rt) in the honesty of the claim.

The four ecolabels were randomly varied across respondents with
equal probability so that approximately equal numbers would receive each.
Approximately half the sample were given the IPM ecolabel and half were
given the ECO ecolabel. This was done to determine whether the number of
environmental impacts specified by the ecolabel claim affects consumer
purchasing choices in a market with ecolabeling. Approximately half were
told that the ecolabel was certified. Certification status was not mentioned to
the other half of the sample. This was done to determine whether proof
ofl’ered that the claim is honest affects purchasing choices in a market with
ecolabeling. The number of respondents receiving each ecolabel is shown in

Table 2.

22

Three regular apple prices (PX) of $.59, $.89 and $1.19 per pound were
randomly varied across respondents with equal probability. Additional prices
could be specified, but this would require a larger sample size than our
research funding permitted. The prices were chosen to reflect variation
around the average U.S. apple price.“ Three price premiums for ecolabeled
apples of $.00, $.20, and $.40 (Px’=Px+premium) were randomly varied
across respondents with equal probability. Additional price premiums would
have required an increase in sample size that our research firnding did not
permit. These three values are based on price premiums for organic apples in
the U.S.12 (A detailed description of how the prices were developed is in

Appendix A.) Both prices were varied across respondents for the purpose of

 

"U.S. apple prices were obtained from United States Department of
Agriculture, EcOnomics Research Service, Fruit and Tree Nuts Situation god
mum, March 1997, FTS-279, p. 5 and United States Department of
Agriculture, Economics Research Service, Fr_uit und Troe Nuts Situation god

1 Y k1 su , October 1996, FTS-278, p. 6.

12Organic apple price premiums were obtained from Harnnrit, James,

0.4 343.11: . ' ns mr Willin e s 0 Pa o R q- - F . . Bo-m s. ,
1986, Santa Monica: The RAND Corp., R-3447-EPA and Jolly, D.,

n r Willin to P Price Premium f r r ni A l
Poachos, March 1989, Department of Agricultural Economics, University of
California, Davis and Jolly, D., “Differences Between Buyers and Non-
buyers of Organic Produce and Willingness to Pay Organic Price Premiums”,
Agribusiness, 9297-111.

23

determining their effect upon consumer demand of regular and ecolabeled
apples.

Randomly varying three regular apple prices and three price premiums
across the sample resulted in nine different price combinations that a
respondent could receive. The four ecolabel descriptions multiplied by the
nine price combinations resulted in 36 total versions of the survey market

scenario randomly varied across respondents. These are summarized in

Figure 3.
Two Two Three Three 36
ecolabels: X forms of X regular X ecolabel = survey
proof: apple price versions
prices: premiums:
IPM Certified
$.59/lb $.00/lb
ECO Not
Certified 5. 89/lb S. 20/lb
$1.19/lb $.40/lb

Figure 3-The Survey Versions
Respondents were asked several questions apart from the market
scenarios in order to identify factors affecting their purchases. These factors
included demographics such as income and education, motivations for

purchasing ecolabeled apples (i.e., perceptions of E(X’), R(X’), and rt), and

24

whether or not the householder purchased organic apples (i.e., potential
substitute for X’). For each factor, statistical tests were run to determine
whether those who gave different answers made different choices.

Respondents were asked, for example, where they normally buy apples
because purchasing location affects the shopping setting and thus might afl‘ect
one’s choice. It was expected that those who buy apples at locations other
than supermarkets and grocery stores might face a different purchasing
setting, difl’erent selection of apples and different substitutes and thus make
difi‘erent purchasing choices, ceteris paribus. Based upon their responses,
respondents were designated as either those who buy apples at a
supermarket/ grocery store or those who do not. This was referred to as the
‘grocery’ variable. It was then determined whether or not there was a
statistically significant difference in purchasing choices between the two
groups.

Respondents were asked their age, gender, household size, level of
education and their household’s annual income.13 This was done in order to

account for the effects of demographics upon purchasing decisions. It was

 

11"The demographic data was coded in such as way as to be consistent
with the figures reported in the U.S. Bureau of the Census Report from 1996.

25

expected that household size and income would have positive effects upon
the amount purchased of both apples and that education would positively .
afi‘ect the amount of ecolabeled apples purchased.

Organic apples are a potential substitute for ecolabeled apples and are
often sold at limited locations. To account for this, respondents were asked if
and how often they bought organic apples. Those who bought them and did
so ‘often’ or ‘always’ were identified as organic apple buyers.“

During apple season in October and November, consumers may buy
more apples. This was accounted for by noting the date upon which the
interview was given. Respondents interviewed in November were identified
as ‘in season’ and those after November were ‘out of season’.

The ‘grocery’, ‘organic’ and ‘seasonal’ variables provided a test of the
validity of the survey technique used for measuring the number of apples
purchased. In our case, the measurement technique was the survey questions
given within the market scenarios. The validity of a measure is the extent to
which it measures the theoretical model under investigation (Mitchell and

Carson, 1989). In the context of this study, the validity of the survey

 

1‘Because a small percentage of apple purchasers buy organic in the
U.S., a more costly and complicated approach of creating ‘organic’ demand
questions was avoided.

26

instrument is a fimction of whether all respondents are viewing the same
economic conditions within their market scenario (aside from those which are
intentionally varied including prices, claim comprehensiveness and proof). A
valid survey would be one in which respondents were perceiving the same
purchasing setting, accounting for the same substitutes, and understanding the
same questions being asked of them. If the amount purchased of apples was
found to be afi’ected by whether one bought apples at a grocery store or some
other location, the fact that one buys organic or whether the survey was
conducted in or out of apple season, this might suggest that some respondents
are perceiving difi‘erent economic conditions from those in the theoretical
model.

Respondent familiarity with the ecolabel claim was also taken into
account because it may indicate greater knowledge of or experience with the
environmental standards within that claim and thus afi'ect one’s choice. For
example, those familiar with the claim may perceive a higher probability that
it is truthful and thus buy more of the ecolabeled apples on average. To
account for familiarity, respondents who received the IPM ecolabel were
asked how much they had seen, heard, or read about IPM. Those who said

that they had heard a fair amount or a great deal about IPM were designated

27

as familiar with the claim. Respondents receiving the ECO ecolabel were not
asked about familiarity with ECO since ECO is not marketed anywhere at this
time.

The theoretical model claims that consumers value both environmental
improvement and health and food safety attributes in the ecolabeled apple.
To test this, we accounted for the different motivations behind particular
brand selections (i.e., perceptions of E, R, and rt). This was done by asking
respondents open ended questions about why they would or would not buy
the ecolabeled apples. Respondents could give up to three reasons for buying
or not buying ecolabeled. Their first responses were coded into categories
such as “concern for the environment”, “health and food safety concerns”,
“ecolabel price too high” and “lack of trust in the ecolabel?“ Respondents
whose first answer fell into the environmental concern category were
identified as those whose primary reason for buying ecolabeled was
environmental improvement. It was then tested whether respondents who
bought them primarily for environmental improvement bought more of the

ecolabeled apples on average. The same process was followed for those

 

lsMany did not give second or third responses so only the first
responses were used in the econometric equations. Table 21 gives statistics
concerning all three responses.

28

whose first answer was health and food safety.
Sample Selection“

The phone interviews were conducted by the Institute for Public Policy
and Social Research (IPPSR) at Michigan State University between
November 3, 1997 and February 11, 1998. IPPSR purchased the phone
numbers fiom Gensys Sampling Inc. who generated a proportional sample of
random numbers from the lower 48 states. ‘7 Of 1453 eligible phone numbers
that were contacted, 972 interviews were completed resulting in a
participation rate of 66.9 %. (The sample selection process is described in
more detail in Appendix B.)

Only those respondents who ever buy apples were asked the apple
purchasing questions. Ninety-two percent of respondents (i.e., n=893)
reported that they buy fresh apples for their household. We refer to these

respondents as the subsample. There was little difference in the

 

"This section is based on Jeffrey R. Blend and Eileen 0. van
Ravenswaay, “ Demand for Ecolabeled Apples: Survey Methods and
Descriptive Results,” 1998, Department of Agricultural Economics Stafl'
Paper No. 98-20, Michigan State University, East Lansing, MI.

1"Alaska and Hawaii were excluded for cost reasons. If these states
had been included either the sample size or the interview length would have
had to be reduced to stay within our research funding limit.

29

demographic composition of the full sample and subsarnple. All the results
reported in the remainder of this section and the next section of the paper are
for the subsample only.

Households which earned higher incomes were over-represented in our
subsample compared to the latest available census figures as were larger
households and householders with more education. ‘3

The household composition of the subsample was somewhat different
from the national average (Table 3). The average household size in our
subsample was 2.93 as compared to the national average of 2.65.19 The
percentage of households in our subsample with children under 18 was 43.2%
which is greater than the national average of 34.3%. The percentage of
sin e-person households in our sample was 15.3% compared to the national
average of 24.9%.

The ages of the householders were similar to census figures (Table 4).

Householders with some college or completed college degrees were over-

 

mU.S. Bureau of the Census, Statjstical Abstract of mo Unitod Stutos;
1296 (116‘h edition) Washington, D.C., 1996. Pages 60, 160, 61, and 465.

’9 These figures taken from U.S. Bureau of the Census, 1996.

° ° lAb tr c f h Uni e tates 1996 (116th edition) Washington,
D.C., 1996, pp. 60, 160, 61, and 465.

30

represented while those with less than a high school education were under-
represented (Table 5). The average age of subsample respondents was 46
and the average education level consisted of some college. Females made up
just over three-quarters of our subsample (Table 6). When asked about
household income, 18.7% of respondents either refused to report that
information or did not know their income (Table 7). Of the households that
did report income, both the ‘$10,000-$49,000' and ‘Over $50,000' categories
were over-represented while the “under $10,000" category was under-
represented. Average household income of the subsample was about
$53,000.

Table 8 gives the percentage of respondents who purchase apples at
various locations. Table 9 gives the percentage of respondents who purchase
organic with some frequency. Table 10 gives the percentage of respondents
who have heard something about IPM.

Model Specification

The amounts of regular (Que) and ecolabeled (Qm) apples purchased
in a market where both are available are measured by the responses to the

questions about willingness to buy and quantity purchased in the second

market scenario. The amount of regular apples purchased without

31

ecolabeling (OR) is measured by the responses to the same questions in the

first market scenario. All amounts are expressed in terms of the pounds of

apples purchased for the household in a single shopping trip in late fall and

early winter.20 (I able 11 provides definitions and summary statistics for all
variables used in the econometric models.)

The explanatory variables included in all three equations are regular
price (PX), household income (M), household size (Hh), respondent age
(Age), gender (Gen) and education (Educ), and the validity variables which
from here on are referred to as ‘grocery’ (Groc), ‘organic’ (Org) and
‘seasonal’ (D). The equations for regular and ecolabeled apples in the second
market scenario with ecolabeling both contain ecolabeled price (Px’), claim
comprehensiveness (L) and proof (C). The equation for ecolabeled apples
contains the two ‘motivation’ variables which indicate whether the primary
reason for purchasing ecolabeled apples was health and food safety or

whether it was environmental concern. These are referred to as ‘health

 

”Respondents purchased between 0 to 126 pounds of ecolabeled
apples and regular apples in a market without ecolabeling. They purchased
between 0 and 42 pounds of regular apples in a market with ecolabeling. The
highest amounts purchased were by those who bought bushels of apples (42
pounds per bushel) in a single shopping trip. Because respondents were
allowed to purchase individual apples (1/3 pound each) and peeks of apples
(10.5 pounds per peck), some purchased fi'actions of pounds.

32

motivation’ (<I>) and ‘environmental motivation’ (‘1’). It also includes a
variable which indicates whether respondents who received the IPM apple
are familiar with IPM. This is referred to as ‘familiarity’ (F). Because
familiarity and the two motivation variables were asked only to select
respondents, Ow, was estimated with and without them.21

The appropriate model for consumer decision-making difiers
depending on the time period involved (e. g., purchases in a single shopping
trip, monthly purchases, yearly purchases). The quantity in the single trip is
small and may often be zero for people who regularly eat apples. The
quantity purchased in a year is large and is unlikely to be zero if the person
regularly eats apples. In a single shopping trip, a two stage decision model
may be more appropriate than a single stage model. In effect, the individual

determines her quantity in advance of her shopping trip based on her

 

21The number of observations for QLAB with the motivation and
familiarity variables is much lower than without. The equation for Qm is
first estimated without motivation and familiarity. The values obtained fi'om
this first equation are reported in the main results in for ecolabeled apples
(Tables 16 and 19). The equation for Qua; is estimated a second time with
familiarity included and a third time with the health and environmental
motivation variables included. These two tables report only the coeficients
upon familiarity, health and environment from these last two equations. This
is signified in both tables by a double line separating the coefficients for these

three variables from the others.

33

household size, preferences and income. Once in the store, she observes
price and quality and decides to buy or not.

The Tobit model (Tobin, 195 9) is commonly used for modeling this
two-stage purchasing choice. The Tobit jointly models the likelihood of
purchasing ecolabeled apples and the quantity purchased. The Tobit imposes
the restriction that the same factors have the same effects upon the decision to
buy apples as they do on the quantity purchased. For a single shopping trip,
this may not hold.

One would expect, for example, that a variable such as price would
have a substantial effect upon the decision to buy apples in a single shopping
trip but not necessarily upon the amount purchased. The reason for this is
that consumers typically know how much of apples or other food items their
household consumes and would likely have made the quantity decision before
aniving at the store. However, with many substitutes for apples, price would
likely have a significant effect upon whether they are bought at all.

The Cragg (1971) Double-Hurdle model allows us to model the same
two-stage purchasing decision but relaxes the restriction imposed by the
Tobit. It allows for different factors to affect the decision to purchase and the

amount purchased. With the Cragg model, a Probit model is estimated for the

34

decision of whether or not to purchase the apples. A truncated regression is
then run upon the amount of apples purchased by those who bought some. In
this way, coefiicients are estimated separately for each of the two decisions
and difi'erent factors are allowed to influence each of these two decisions.

To determine whether the Tobit restriction held in our study, two tests
were run upon the equations for regular and ecolabeled apples in a market
with ecolabeling. The first test created by Pin and Schmidt (1994) determines
whether the likelihood value of the Tobit is significantly different from the
sum of likelihood values of the Probit and truncated regressions from the
Cragg model. It was found that for both regular and ecolabeled apples in a
market with ecolabeling, the difference in values was significant.22 Thus, the
results were sigrificantly difl‘erent without the restriction.

A second test consisted of running a Tobit and then running a Probit
upon the decision to buy apples. The Tobit coefficients were normalized by
dividing them by a from the re gession. The normalized Tobit values were

then compared to the values of the Probit coefficients. Ifthere are any

 

22The test is whether 2*(Likelihood Tobit-Likelihood Probit +
Likelihood Truncated) > Chi-Squared (13=nmnber of parameters) or x213.
1213 = 22.36. For regular apples, the test statistic equals 38.6 and for
ecolabeled apples it is 103.2. Both are above the critical value.

35

changes in the sigr of statistically sigrificant variables or differences in
magnitude often or more, then the two results are considered significantly
different. The sigr of the gender variable changed for the ecolabeled apples
and upon gocery for both apples. Thus, it was suspected that the results
without the restriction were different.

Based upon these tests, Cragg’s Double Hurdle model was chosen as
more appropriate to run. For the sake of comparison and providing
recommendations to other survey researchers, the results from the Tobit were
also reported.

A linear model often gives similar estimates to the Tobit of
coefficients and elasticities of the explanatory variables at their mean values.
Thus, it is worthwhile to compare the estimates from the Tobit to those
yielded from a linear model estimated using OLS. This comparison provides
a test of the sensitivity of the results to different modeling assumptions.

Because regular and ecolabeled apples are presented together in the
second market scenario, their error terms may be correlated. To account for
this, a Seemingly Unrelated Regessions (SUR) technique is also run to
determine whether the results change sigrificantly when correcting for

existing correlation.

36

Separate equations were run for regular and ecolabeled apples which
included four interaction terms between regular and ecolabeled prices and .
certification and comprehensiveness. This was done in order to determine
whether the efi‘ects of own and substitute prices are different across the
difi‘erent ecolabels.

Results

The amounts of regular and ecolabeled apples respondents would
purchase in a single shopping trip in markets with and without ecolabels form
the dependent variables for the three demand equations. The first equation
estimates the parameters for the demand function for regular apples (QR)
given the first market scenario without ecolabeling. The second and third
equations estimate the parameters in the demand functions for the regular
(Om) and ecolabeled apples (QLAB) given the second market scenario with
both apples available.

We found sigrificant consumer demand for ecolabeled apples. Over
halfof respondents (56%) would try the ecolabeled (i.e., either IPM or ECO)
apples presented in our second market scenario (Table 12). On average, they
bought 1.5 pounds of the ecolabeled apples and 1.3 pounds of the regular

when both were available.

37

Many respondents were willing to pay more for the ecolabeled apples.
The average marginal willingness-to-pay for a pound of ecolabeled apples
was calculated at $.32/lb.23 At a regular apple price of $.89 per pound, this
translates into a 36% price premium.

The geater the price premium on ecolabeled apples, the lower the
percentage of respondents who would buy some (Table 13).“ Seventy-two
percent of those who were offered regular and ecolabeled apples at the same
price said they would buy the ecolabeled.25 Fifty-two percent of those
receiving a $.20 price premium bought some of the ecolabeled. At a $.40
price premium, the percentage who said they would buy was 42%.

Results of the Cragg Double Hurdle model are found in Tables 14

through 16. Column one lists the variable names (Refer to Table 11 for their

 

23A1] calculated price premiums were calculated from the Tobit.
Premiums were calculated at the average quantity demanded of regular apples

given the second market scenario (about 1.3 pounds). A value of $.27/lb was
calculated from a linear model estimated by OLS.

2“The reported percentages of respondents buying ecolabeled apples
includes both those who bought ecolabeled only and those who bought some
of both types of apples.

2’The Food Marketing Institute (whose study is mentioned in the
introduction) found that there is a large goup for whom environmental
fiiendliness would be a “tie-breaker” in choosing among brands within a food
product category, although only a few would pay more.

38

definitions). The second column in each table lists the marginal effects of the
Probit upon the decision of whether or not to buy the apples. Specifically, it
gives the change in the probability of the average respondent purchasing
apples caused by a one unit change in the explanatory variable. Using the
amount purchased of ecolabeled apples and education (Educ) as an example
in Table 14, each additional year of education would raise the probability of
the average respondent buying ecolabeled apples by .026, on average. The
third column displays the coeflicients of the truncated regession. The forth
column gives the marginal effects of the truncated regession. In particular,
the forth column gives the change in the quantity of apples purchased by
those who bought positive amounts caused by a one unit change in the
explanatory variables. Using the same example, each additional year of
education results in .023 fewer pounds of ecolabeled apples purchased on
average by those who purchased some.

Results of the Tobit model are found in Tables 17 through 19 for each
of the three demand equations. Variable names are found in the first column
(Refer to Table 11 for their definitions). The Tobit coefficients and their
standard errors are listed in the second column. The McDonald and Mofiitt

(1980) transformed marginal effects are listed in columns three through five.

39

For each of the demand equations estimated by Tobit, the explanatory
variables were jointly significant in predicting the model. Leaving out similar
goups of explanatory variables (i.e., all demographics or the three validity
variables) did not improve any of the equations, so all were included.

The coefficients from the Tobit cannot be interpreted in the same way
as they are for an equation estimated using OLS. Their value does not
represent the expected change in the dependent variable given a one unit
change in the explanatory variable. The coefficients must undergo a
transformation developed by McDonald and Moffitt (1980) in order to
estimate marginal effects. This transfonnation decomposes the marginal
efi'ects and elasticities (calculated at the means) of the variables into three
parts.

The first part which is denoted as 6E(Y)/6x (where Y is the dependent
variable and x denotes one of the explanatory variables) gives the change in
the average amount of apples purchased given a one unit change in the
explanatory variable for all respondents. The second part, denoted as
6E(Y")/6x (where Y“ is all non-zero observations of the dependent variable),

gives the change in the average amount of apples purchased by those who

40

would buy the apples given a one unit change in the explanatory variable.26
The third part, 6F(z)/dx (where z is a normalized index and F (0) is the
cumulative standard normal distribution function), gives the change in the
probability of buying the apples given a one unit change in the explanatory
variable.

Using OM and Px’ in Table 18 as an example, aone dollar increase in
ecolabeled apple price would result in a 2.2 increase in the average pounds of
regular apples purchased by respondents in a single shopping trip
(6E(Y)/6Px’), a 1.7 pound increase in amount purchased for those
respondents who bought regular apples (dE(Y“)/6Px’), and an increase in the
probability of respondents buying regular apples of .41 (6F(z)/6Px’).

The dummy variables in these tables are intercept shifters. Using
organic. (O) as an example in Table 18, all respondents who normally buy
organic apples would buy 1.5 less pounds of regular apples on average in a
single shopping trip. Organic apple shoppers who bought regular apples
would buy 1.2 pounds less of them on average. All shoppers who normally

buy organic apples would have a .28 lower probability on average that they

 

2"One could also think of dE(Y*)/ax as the partial. derivative of Y with
respect to x given that Y is geater than zero or dE(Y|Y>0)/ax.

41

would buy regular apples in a market with ecolabeling.

The McDonald and Moffrtt technique also decomposes elasticities of
the explanatory variables calculated at their means into three parts. These are
the elasticity of the probability of buying the apple, nF(z), and the elasticity
of apple demand by those who purchase some of the apples, nE(Y’), which
sum up to the total elasticity of demand (the percentage change in the
dependent variable given a one percent change in the explanatory variable),
711300-27

Demand elasticities calculated fiom the Tobit model at the means of
own price, substitute price, income and other sigrificant variables are given in
Table 20 for all three equations. The way to interpret columns two through
four in Table 20 is as follows. Using the example of OM and Px’ again, a
one percent increase in ecolabeled apple price would result in a 1.57 percent
increase in the total amount purchased of regular apples by all respondents,
nE(Qm), a .50 percent increase in the amount of regular apples purchased

by those who bought regular, TIP-(Qnm'), and a 1.07 percent increase in the

 

2"The presentation of the McDonald and Moffit decomposition
technique is adapted from Nonis, Patricia E. and Sandra S. Batie, “Virginia
farmers’ Soil Conservation Decisions: An Application of Tobit Analysis,”
Southern Journal of Agricultural Economics, July, 1987: 79-90.

42

probability of respondents buying regular apples, nF(z).

The results from the Cragg model indicate that both prices have
significant effects (of the expected sigrs) upon the probability of buying
ecolabeled apples (Table 14). Education has a sigrificantly positive effect
upon the probability of buying. Household size, age and enviromnental
motivation have sigrificantly positive effects upon the amount of ecolabeled
apples purchased whereas ecolabel price and purchase location (i.e., the fact
that one buys apples at a gocery store or supermarket) have significantly
negative effects. Being male and being familiar with the IPM claim have
significantly negative effects upon the decision to buy ecolabeled and
significantly positive effects upon the amount purchased. This suggests that
both of these goups would buy ecolabeled apples less often, but would buy
more on average when they did decide to purchase them. Buying apples in
season (i.e., during November) has a sigrificantly positive effect upon the
decision to buy ecolabeled.

Regular apple price, respondent education, household size and

environmental motivation have sigrificantly positive effects upon the amormt

43

of ecolabeled apples purchased in the Tobit model (Table 17).28 Ecolabeled
price and purchase location botln have significantly negative effects upon the
amount of ecolabeled apples purchased. For the Tobit model, own price
elasticity for ecolabeled apples was calculated at -2.0 and cross-price
elasticity at 1.6 (T able 20).

The results from the Cragg model (Table 15) indicate that both prices
have a significant effect of the expected sign upon the probability of buying
regular apples in a market with ecolabeling. Being male has a significantly
positive efi‘ect and normally buying organic apples a significantly negative
effect upon the probability of buying regular apples. Household size and age
positively and significantly increased the amount purchased of regular apples
by those who bought some whereas gocery had a significantly negative
effect.

For the Tobit (Table 18), the following variables have a significantly
negative efiect upon the amount of regular apples purchased in the market
with ecolabeling: regular price, buying organic and purchase location (i.e.,

purchasing apples at a gocery or supermarket). Ecolabeled price and gender

 

2"Unless otherwise noted, ‘significant’ means statistically significant at
a 10% level.

44

(i.e., being male) both have significantly positive effects on the amount
purchased. For the Tobit model, own price elasticity for regular apples was
calculated at -1.7 and cross-price elasticity at 1.6 (Table 20).

Cragg’s model indicates that income and education significantly
increase the probability of buying regular apples in a market without
ecolabeling while own price and age negatively affect that probability (T able
16). Household size, age and gender positively affect the amount purchased
by those who bought regular apples whereas gocery negatively afi‘ects the
amount purchased.

In the Tobit model, gocery is the only significant variable for regular
apples in the market without ecolabeling (Table 19). Own price is of the
expected sign but not significant.

The gocery variable is significantly negative for all three equations
indicating that those who buy apples at a gocery store or supermarket bought
less on average. The Cragg model indicates that gocery primarily affects the
amount purchased rather than the decision to buy for all tlnree apples. It also
indicates that age and household size significantly increase the amount
purchased of all three types of apples.

Using the Cragg model, certification had a significantly positive effect

45

on the amount purchased of regular apples by those who bought some.
Respondents who bought some regular apples bought about .5 pounds more
when ecolabeled apples were certified. It did not affect demand for
ecolabeled apples. In the Tobit model, certification did not significantly
afl'ect regular or ecolabeled demand.

The comprehensiveness of the environmental claim did not
significantly affect the amount purchased of regular or ecolabeled apples in
either model.

The average marginal willingness to pay for ecolabeled apples did not
geatly differ across the different ecolabels. The marginal willingness-to-pay
for ecolabeled apples with USDA certification (calculated from the Tobit
model) was $.29/lb and for those without certification $.35/lb. There was an
even smaller difl‘erence between ecolabels with different levels of
comprehensiveness. Respondents were marginally willing to pay on average
$.30/lb more for IPM apples and about $.33/1b more for ECO apples.

The price of ecolabeled apples which are certified and more
comprehensive had a significantly geater effect upon the amount of regular
apples purchased than did the price of those which are non-certified and less

comprehensive. In other words, the cross-price elasticities of the certified

46

and ECO apples on regular apples were geater than those of the non-certified
and IPM. This was indicated by the observation that for regular apples in a
market with ecolabeling, the interaction terms of ecolabeled
price‘certification (e. g. Px’ * C) and ecolabeled price‘comprehensiveness
(e.g. PX” L) were both positive and significant.

Health and food safety was the reason most frequently given for
purchasing ecolabeled apples (Table 21). Thirty-seven percent mentioned
health or food safety, including the avoidance of chemicals in food and better
health for their family, as their first answer. Twenty-seven percent mentioned
concern for the environment as their first answer and 19% mentioned the
desire to try out the ecolabeled apples. Health and food safety did not affect
the amount purchased of ecolabeled apples by those who bought some.29

Because health and environmental motivations were only recorded for those

 

29Health and environmental motivation for purchasing ecolabeled
apples were measured by asking respondents who bought ecolabeled why
they did so. Both motivational variables were also measured in a second
way. They were both included in the ecolabeled apple demand equation for
those who purchased ecolabeled apples as well as for those who did not.
These variables now measured whether health or environmental motivation
were significant factors in respondents’ choices with respect to buying
ecolabeled apples, including the decision not to buy. Measured this new way,
both health and environmental motivation were positive and significant in
both models. Thus, both motivations had a significantly positive effect on
ecolabeled demand for all respondents including non-buyers.

47

who bought ecolabeled apples, we could not estimate their effect upon the
decision to buy ecolabeled apples.

Respondents’ top reasons for not purchasing ecolabeled apples were
that they were too expensive (46%), that they did not know enough about the
ecolabeled apples (22%), that they preferred their usual apples (9%) and that
the regular apples were just as safe to consume (8%) (Table 22). Nine
percent of responses indicated distrust of the ecolabel, IPM or the USDA.

On average, respondents bought about the same amount of apples with
and without ecolabeling. In the market without ecolabeling, respondents
bought 2.9 pounds of apples on average compared with 2.8 pounds of apples
in the market with ecolabeling (e. g. 1.5 pounds of ecolabeled + 1.3 pounds of
regular). However, the amount of regular apples purchased in the second
market scenario decreased by roughly the same amount of ecolabeled bought
indicating a substitution towards ecolabeled apples fi'om regular apples.

Regular apples became more price elastic with ecolabeling as a result
of the substitution effects from the ecolabeled apple. Their own price
elasticity without ecolabeling was about -. 14 and with ecolabeling was about
-1.7. This discrepancy is not surprising. Price would likely not affect apple

demand in a single trip, but rather the decision to buy only. Witln the

48

introduction of ecolabeled apples for the first time, one would expect
substantial substitution effects between the two apples and thus a geater ‘
sensitivity of demand irn a single shopping trip to price.

The results using a linear model estimated by OLS were qualitatively
similar to those of the Tobit. Quantitatively, there were some significant
differences between the two models. In Tables 23 and 24, the total marginal
effects of each variable upon all respondents from the Tobit (dE(Y)/6x or the
change in the average amount of apples purchased given a one unit change in
the explanatory variable for all respondents) are listed alongside those fiom
OLS and SUR for regular and ecolabeled apples in the market scenario with
ecolabeling. The SUR results were very similar to those from OLS indicating
the correlation between the residual terms of each equation did not
significantly affect the coefficient values or conclusions which could be
drawn. Because the OLS and SUR results were so sinnilar, a comparison is
made only between the Tobit and OLS results.

Aside from education, the same variables were statistically significant
for ecolabeled apples from the Tobit and OLS equations (Table 23).
Calculated price elasticities for ecolabeled apples were similar for both

equations. With Tobit, the marginal effects of both prices were slightly larger

49

and household size slightly smaller than OLS. There was a large difference
between the two estimation techniques in the marginal effects for education
and gocery.

For regular apples, own price and gocery were negative and
significant using both estimation techniques (Table 24). However, the
marginal efl‘ect of own price was larger and of gocery smaller for Tobit than
for OLS. Ecolabeled price was significant and its marginal efi'ect much larger
for Tobit whereas certification was significant and larger using OLS. Own
price elasticity for regular apples was similar using botln estimation techniques
as was income elasticity. Cross price elasticity was larger using the Tobit.
Conclusions

There is substantial consumer demand for ecolabeled apples. A
significant percentage of the sample (56%) would try ecolabeled apples if
they were available. In a market with both regular and ecolabeled apples,
respondents bought more of the ecolabeled on average.

A sizable percentage of consumers would pay more for ecolabeled
apples. Fifty-two percent of respondents receiving a $.20/lb price premium
bought some of the ecolabeled and at a $.40/lb prenniurn, 42% bought some.

The average marginal willingness-to-pay for ecolabeled apples was calculated

50

at $.32/lb.

In a market with ecolabeling, both regular and ecolabel prices would
affect consumer purchasing choices. In the Tobit model, regular and ecolabel
prices afl‘ected the quantities purchased of both apples. In the Cragg model,
both prices had significant effects upon the probability of buying regular and
ecolabeled apples. Prices have a strong effect upon the choice of which apple
to buy, but a less significant effect upon how many are bought. Thus, there
would be substantial substitution effects between apples if both were
available.

The geater the price premium on ecolabeled apples, the less
consumers would buy. Thus, retailers would likely face a tradeofi‘ between
the size of the price premium they set and the pounds of ecolabeled apples
sold. Those interested in promoting ecolabeled apples nnight offer them
initially at a lower price per pound relative to regular apples with the
understanding that purchases of regular will decline.

Advertising campaigns for ecolabeled apples should be targeted at
educated and older consumers and larger households. In the Cragg model,
older respondents who bought ecolabeled apples bought more on average.

Household size had a positive effect upon the amount of both regular and

51

ecolabeled apples purchased, and education had a positive effect upon the
probability of purchasing ecolabeled. For the Tobit model, respondents with
higher levels of education and those from households with more occupants
bought more ecolabeled apples on average. One marketing strategy would be
to target specific forms of media and shopping locations known to attract
more educated and older shoppers. Another would be to target locations
fi'equented by larger families such as those which have children.

Consumer demand for both regular and ecolabeled apples may not be
afl‘ected by claim comprehensiveness. Claim comprehensiveness had almost
no efl‘ect upon the amount purchased of either apple under either estimation
technique. Average marginal willingness to pay only differed by three cents
between ECO and IPM apples. However, only one way of wording the IPM
and ECO claims was tested. In addition, only one set of standards was tested
for each claim. Further research is needed to test different wordings and
different sets of standards.

Certification of the ecolabeled claim did not affect the demand for
ecolabeled apples in either model. Proof in the fornn of govemment
certification does not increase consumer trust in the label or the decision to

buy ecolabeled. However, only two forms of proof were tested, USDA

52

certification and no proof. Further research is needed to determine whether
other certifiers (e. g. non-profit environmental goups) and other forms of .
proof (e. g. national standards) would affect consumer demand for ecolabeled
apples.

Consumers who buy ecolabeled apples primarily for environmental
improvement bought more on average and may have a longer term
commitment toward purchasing such apples. Those who just want to try them
out would likely try fewer pounds irnitially. This suggests that ecolabelers
should target their products towards stores krnown for their concern over
environmental quality such as food co-ops and geen gocers.

Healtln and food safety attributes resulting fiom ecolabeled production
should be highlighted in a marketing campaign. Health and food safety was
mentioned most often as the primary reason for buying ecolabeled apples.
Although it did not significantly affect the amount of ecolabeled apples
purchased using either estimation technique, it was close to significant in the
Cragg model. This suggests that while many consumers would make the
decision to buy ecolabeled apples primarily for the health and food safety
attributes, they may not perceive added value from purchasing more of the

ecolabeled apples. Thus, health and food safety attributes may be most

53

effective in a campaign aimed at convincing consumers to give ecolabeled
apples an initial try.

Familiarity with the IPM claim may not lead to geater consumer
demand for the ecolabeled apples. In the Tobit model, respondents familiar
with the IPM claim actually bought less ecolabeled apples than those
unfamiliar with the claim, although this was not statistically significant. The
Cragg model showed that familiarity had a negative effect upon the
probability of purchasing the ecolabeled apples and a positive effect upon the
amount purchased (both effects significant). This suggests that those familiar
with IPM may be less apt to buy ecolabeled because either IPM is not
perceived to improve environmental quality or those familiar with IPM do not
value environmental improvement in the same manner. It also suggests that
familiarity may have a positive effect upon the amount purchased by those
who choose to buy some. We only measured familiarity with IPM. Furtlner
research might test the effects of familiarity with other claims. Only a small
percentage of consumers were identified as being familiar with IPM.

Ecolabelers may benefit from targeting organic apple buyers and those
who buy apples at locations other than supermarkets and gocery stores, as

both goups bought more ecolabeled apples on average. However, these

54

consumers only make up only a small percentage of the population.

Researchers who estimate consumer demand for food items in a single
shopping trip should think carefully about the econometric model they use.
This dissertation has shown that the Cragg Double Hurdle model is probably
more appropriate for estimating purchasing choices in a single trip. One
would expect the two stages of the purchasing choice to be made difi‘erently
in a single shOpping trip and thus to be affected differently by variables such
as price. The Tobit may be more appropriate for longer term demand where
the decision to buy and the amount purchased are more correlated with one
another over numerous trips.

Evidence to support this assertion includes the two statistical tests
which determirned that the Cragg model results were significantly difl‘erent
from the Tobit model results. It also includes the fact that price is
insignificant in the Tobit model while significantly increasing the probability
of buying the regular apples in a market without ecolabeling using the Cragg.
Thus, price has a geater effect upon the decision to buy regular apples than
on the amount purchased, a distinction that the Cragg model allows the
researcher to see. Finally, the Cragg gives a more detailed explanation of

how some of the demogaphic variables affect consumer decisions.

55

Researchers who conduct future apple surveys should account for
organic apples and for shoppers who buy apples at locations other than
gocery stores and supermarkets. The results suggest that these respondents
may not perceive exactly the same scenario as the others. Those who buy
apples at locations other than a gocery store or supermarket bought more on
average. This result is explained by the fact that a larger percentage of
respondents who purchased them at non-gocery locations bought apples in
larger quantities (i.e. peeks, bushels). Perhaps this is how they are typically
sold at other locations explaining why some shoppers go there. The negative
and significant effect of buying organic on regular apples may be explained
by the different apple selection, information and values that orgarnie shoppers
possess. Giving the survey interview after apple season significantly and
positively afieeted the decision to buy ecolabeled apples in the Cragg model,
but did not significantly affect apple demand in any other case. Thus,
researchers interested in constructing apple surveys may not have to aeconmt
for apples being in season if they are studying regular apples.

Care should be taken in determining the ecolabel price premium and in
educating consumers about ecolabeled apples. Expense and lack of

krnowledge were the top two reasons given by respondents for not purchasing

56

ecolabeled apples (Table 22). This reinforces the result that ecolabel price
and respondent education significantly affect the amount purchased of
ecolabeled apples.

There are many other items that could be used to test consumer
demand for ecolabeled foods including meat products, other types of produce,
processed foods and dairy products. For any specific product, further
research miglnt involve the use of visual ecolabels. Visual dimensions of such
ecolabels could be tested as to their effects upon demand. One might also
test the types of environmental information (i.e., improved water quality vs.
lower levels of chennieal runoff) that shoppers most strongly respond to.

Further research might examine additional aspects of ecolabels such as
standard strietrness and precision of standards (process vs. performance
standards). It might also look more closely at the production side and the

potential costs involved with meeting ecolabel standards.

57

TABLES

58

Table 1

Names of Ecolabels in the Questionnaire

PROOF OFFERED
TO SUBSTANTIATE

DIMENSIONS OF CLAIM?
E C OLABEL W ORDIN G

ENVIRONMENTAL
CLAIMS MADE?

 

m' "
ECO Certified ‘
. ECO

AMOUNT OF

 

 

“Tl USDA
Less l IPM Certified
f r J IPM .

Table 2

g Number of Res .ondents for Each Ecolabel
, .

- TYPE OF
i ECOLABEL

 

 

Not . Certified
Certified ' I

248 242

243 239

 

 

59

Table 3
Household Composition

 

      
 

  
   
 

 

 

 

 

 

 

 

 

 

 

Household Percent of Percent of Percent of
Composition Census'I Sample“ Subsam le"
Average household
size (number of 2.65 2.87 2.93
People)
Percentage of l
households with 34.3% 42.2% 43.2%
children under 18
Percentage of Single- 249% 17.0% 15.3%
. __person households
'Souree: U.S. Bureau of the Census, StaujsuoaLAosttuotofthsfluitodjtates;

1296 (116“ edition) Washington, D.C., 1996. Pages 60, 160, 61, and 465.
"‘ Sample’ refers to the firll sample of 97 2 respondents that were contacted.
‘Subsample’ refers to the 893 respondents in the full sample that buy apples.

60

Table 4

 

 

 

 

 

 

 

 

Age of the Respondents
Age of Percent of Percent of Percent of '
Householder‘ Censusb Sam 1ec Subsamplec I
15-24 years 5.3% 6.5% 6.6%
25-34 years 19.3% 18.8% 18.1%
35-44 years 23.3% 22.1% 24.5%
45-54 years 18.0% 19.1% 19.6%
55-64 years 12.4% 11.6% 11.8%
65-74 years 11.9% 8.2% 8.3%
75 ”a” and 9.6% 6.5% 6.5%
older
Refused --- 5.0% 4.5%

 

 

 

 

 

 

'Because respondents had to be at least 18 years old to complete this survey,
the first age category is 18-24 years for sample and sub-sample respondents.
l'Souree: U S Bureau of the Census WW5:

129 (116* edition) Washington, D. C. 1996. Pages 60, 160, 61, and 465.
“ Sample’ refers to the fnnll sample of 97 2 respondents that were contacted.

‘Subsample’ refers to the 893 respondents in the full sample that buy apples.

61

Table 5
Educational Attainment of Respondents

 

       
    

 

 

 

 

 

 

 

 

 

 

Education level Percent of Percent of Percent of
Census' Sam leb Subsam le"
Did not graduate o 0
high school 6'7 /° 6'0 A
High “hm“ 33.6% 28.1% 27.3%
diploma
5"” “mg” 24.5% 30.1% 30.6%
allege ”r . 23.6% 32.7% 33.9%
graduate educatnon
Rem” _ 2.4% 2.0%
'Source: U.S. Bureau of the Census; WW

1996 (116“ edition) Washington, D.C., 1996. Pages 60, 160, 61, and 465.
I"Sample’ refers to the fnnll sample of 972 respondents that were contacted.
‘Subsample’ refers to the 893 respondents in the finll sample that buy apples.

62

Table 6
Gender of Respondents

 

   
 

     
 

    
       
     

Percent of
Sam le' Subsam le‘I

Female 75.0% 76.8%

Male II 25.0% 23.3% I

" Sample’ refers to the full sample of 972 respondents that were contacted.
‘Subsample’ refers to the 893 respondents in the full sample that buy apples.

Gender Percent of

      

 

 

 

63

Table 7

Household Income

 

 

 

 

 

 

 

 

 

 

 

Total Percent of Percent of Percent of Percent of
household Census Sample“ Subsample" subsample
income in Households‘ who reported

1996 income‘
Less than 0 o o 0

$10,000 12.3 /o 3.0/0 2.1 /o 2.6/0
510,000. I , , , ,
$49,000 55.7/0 43.8/0 44.7/0 55.0/0
5504”" °' 31.9% 33.4% 34.5% . 42.4%
more
No answer I --- 19.8% 18.7% ---
Mean
household $34,076 --- --- $53,003
income

 

'Souree: U.S. Bureau of the Census,
122d (116‘ll edition) Washington, D.C., 1996. Pages 60, 160, 61, and 465.
"‘ Sample’ refers to the full sample of 972 respondents that were contacted.
‘Subsample’ refers to the 893 respondents in the hill sample that buy apples.
‘This column was included to compare tlnose respondents in the subsample
who reported income with Census figures.

 

Table 8
Q: Where do you usually purchase fresh apples? (Open ended)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

' Location Percent
ResLondents
Supermarket or grocery store 89.7 % I
Farm, roadsnde stand or 5.3 %
orchard
Farmer’s Market 3.4 %
Green grocer or frunt specralty 0.7 %
store
Bulk/Discount Store 0.2 %
Don’t Know 0.1 %
Refused 0.7%
Table 9
Q: When you buy apples, how often do you buy organic apples?
Response Percent
Respondents _
Often or always 2.9 %
Rarely or 15.3 %
occasnonally
Do not buy them 73.9 %
Do not know 7.5 %
Refused 0.3%

 

 

 

 

65

Table 10
Q: How much have you seen, heard, or read about Integrated Pest
Management or IPM?
Wore: This question given only to those receiving a scenario with IPM
apples; N =404)

 

 

 

 

 

 

 

Response I Percent
Respondents
A great deal T 2.7 %
A fair amount 4.2 %
A little 26.5 %
Nothing 66.3 %
Don’t Know 0.2 %
Refused 0.0 %

 

 

 

 

66

Table 11

Variable names, definitions, means and standard deviations‘

 

 

 

 

 

 

 

 

 

 

 

 

 

Definition Mean Std.
__W—_ _____ _________l-___ ___Dev
OR 1 Pounds of regular apples purchased for the 8.6988 18.4988

1 household in a single shopping trip when
l ecolabeled apples are not available
Que Pounds of regular apples purchased for the 3.8949 9.1442
household in a single shopping trip when
’ both types of apples are available
Qm Pounds of ecolabeled apples purchased for 4.4943 14.8275
l the household in a single shopping trip when
both types of apples are available
Px ! Price of regular apples per pound (in .8823 .2453
. dollars)
Px’ Price of ecolabeled apples per pound (in 1.0794 .2940
dollars)
C Form of proof offered for the environmental .5050 .5003
claim:
1=USDA certification, O=No Proof
L Comprehensiveness of ecolabel claim: .5006 .5003
1=More-ECO ecolabel, 0=Less-IPM
ecolabel
M Household Income (in thousands of dollars) 53.0028 36.09
Kb ‘ Household Size (Number ofoccupants in 2.9287 1.6094
household)
Age Age of householder (in years) 46.1596 16.1943
Gen Respondent’s gender: 1=Male, 0=Female .2318 .4222

 

 

 

 

.‘These numbers are reported for the 893 respondents in the full sample that

buy apples.

67

 

Table 11 (Cont’d)

 

Definition

Educ Number of years of education householder
has completed

Mean

Std.
Dev. 7

14.0331

2.7065

 

Groc Householder’s typical apple purchasing
location:
1=Grocery store or supermarket, 0=Other
locations

.8959

.3056

 

Frequency that householder normally buys
organic apples:
1=Often or always, 0=Never, rarely or
occasionally

Org

.0291

.1682

 

season (in November):
1=Given in season during November,
0=Given out of season from December
through February

.3359

.4726

 

D I Whether survey was given during apple

1=Yes, 0=No

Whether householder is familiar with IPM:

.0659

.2546

 

<I> Respondent’s first reason for buying
‘ ecolabeled apples is:
1=Health or food safety, 0=A different
reason

.3689

.4830

 

‘1’ Respondent’s first reason for buying
ecolabeled apples is:
1=Concern for the environment, 0=A
| different reason

 

 

.2722

 

.4455

 

68

 

Table 12
Percent respondents purchasing regular and ecolabeled apples when
both are available

 

   

Ecolabele Some of No apples Don’t Refused

 

 

(1 only" both apples know_
41.7% 13.9% 6.3% 4.4% 0.7%

 

 

 

 

 

 

'Ecolabeled apples refer to botln IPM and ECO apples.

Table 13
Percent respondents purchasing regular and ecolabeled apples when the
price premium on ecolabeled apples is $ llb.

 

Price Regular Eco- Some No Don’t Refused
Premium on Apples labeled of both apples Know
ecolabeled Apples apples

 

5.00 16.3% 57.2% 15.4% 7.2% 2.6% 1.3%
5.20 I 39.8% 38.1% 14.3% 5.4% 1.4% 0.0%

5.40 H 44.0% 29.0% 13.3% 9.9% 3.1% 0.7%
'Ecolabeled apples refer to both IPM and ECO apples.

 

 

 

 

 

 

 

 

 

 

69

Table 14
Cragg Double Hurdle for ecolabeled apples in a market with ecolabeling

 

 

 

 

 

 

 

 

 

 

 

 

 

Dependent Variable (estimation
technique) Marginal effects
Variable Whether or not Amount of regular of the Truncated
name regular apples apples purchased regressron upon
were purchased' by those who amount
(Probit) bought some purchased
V__. ________._._.-NW___.___-__ _22 _ _ ,
PX 7513*" 6.6724 .8784
(.1353) (5.3694) (.7069)
Px’ -.8894*" -8.l623* -l .0746“
(.1140) (4.8833) (.6429)
M .0002 .0030 .0004
(.0006) (.0199) (.0026)
Hh .0142 .9403" .1238"
(.0123) (.4613) (.0607)
Age -.0014 .1085" .0143”
(.0012) (.0534) (.0070)
Gen -.1112*** 3.7615" .4952"
(.0444) (1.8524) (.2439)
Educ 0255‘" . -.1738 -.0229
(.0072) (.2708) (.0357)
L .0506 .6130 .0807
(.0363) (1.3378) (.1761)

 

'The values listed in this column are the marginal values of the Probit and
thus give a change in the probability of buying the apples given a one unit
change in the explanatory variable.

70

 

Table 14 (cont’d)

 

 

 

 

 

 

 

   

 

 

 

 

Dependent Variable (estimation
technique) Marginal effects
V . bl of the Truncated
ana e Whether or not Amount Of regular regression upon
name regular apples apples purchased by amount
were purchased' those who bought purchased
_- _’-__t___ som_ru_-_ted ___- -_-__V
C -.0133 -1.0754 -.1416
(.0364) (1.3484) (.1775)
Groc -.0536 -6.3647** -.8379"
(.0599) (2.5201) (.3318)
Org .1653 .9537 .1255
(.0914) (3.5129) (.4625)
D l .0624“ -2.1041 -.2770
_ _-__-_ ' .0380 _ _557--- -__2-_35-__
F' -.1973* 1.4127“ .7132“
(.1023) (.8188) (.4134)
(IF ---" 2.7356 .3841
‘1 (1.6835) (.2364)
‘P‘ ---" 3.8977" .5473”
(1.8950) (.2610)
a --- 2.7452 .....

 

 

 

‘= Significant at the 10% level,**= Significant at the 5% level,*"=

Significant at the 1% level

‘The values listed in tlnis column are the marginal values of the Probit and
thus give a change in the probability of buying the apples given a one unit
change in the explanatory variable.
I'Because (D and ‘P were measured only for those who bought ecolabeled,
they cannot be measured for their effect upon the decision to buy.

71

 

Table 15
Cragg Double Hurdle for regular apples in a market with ecolabeling

 

 

 

 

 

 

 

 

 

 

 

 

Dependent Variable (estimation
technique) Marginal effects
V . bl of the Truncated
arra e Whether or not Amount of regular regression upon
name regular apples apples purchased by amount
were purchased' those who bought purchased
-- __ 'robit _______ ___e -.-_ru__-d_- --_ _-_ - _ _ - V
PX -.8452*** -3 .5062 -1 .0056
(.1377) (2.5510) (.7317)
Px’ 7552*“ 1.0373 .2975
(.1136) (2.1911) (.6285)
M .0003 -.0032 -.0009
(.0006) (.0115) (.0033)
Hh -.0080 7275"" 2086*“
(.0123) (.2444) (.0701)
Age .0000 0714*“ 0205*“
(.0012) (.0248) (.0070)
Gen .0918“ .6809 .1953
(.0436) (.7896) (.2265)
Educ -.0042 -.0598 -.0171
(.0072) (.1254) (.0360)
L -.0098 -.5383 -.1544
(.0365) (.6612) (.1897)

 

 

'The values listed in this column are the marginal values of the Probit and

thus give a change in the probability of buying the apples given a one unit
change in the explanatory variable.

72

 

Table 15 (cont’d)

 

 

 

 

 

 

 

Dependent Variable (estimation
technique) Marginal effects
Variable Whether or not Amount of regular of the Truncated
name regular apples apples purchased regressnon upon
were purchased' by those who amfrunzd
(Probit) bought some Pm ‘3
_ -___- _- ____-_______-___-_- _-_-__---_..
C .0305 1.7641“ .5060“
(.0366) (.7234) (.2075)
Groc -.0446 -2.4535** -.7037**
(.0365) (1.1198) (.3212)
Org -.3142*** .5051 .1449
(.0791) (2.7083) (.7768)
D l -.0528 .4246 .1218
(.0384) (.7072) (.2028)
c --- 3.4594 .....

 

 

 

 

 

‘= Significant at the 10% level,**= Significant at the 5% level,”*=
Significant at the 1% level

'The values listed in this column are the marginal values of the Probit and
thus give a change in the probability of buying the apples given a oneunit
change in the explanatory variable.

73

Table 16

Cragg Double Hurdle for regular apples in a market without ecolabeling

 

 

 

 

 

 

 

 

 

 

 

 

 

Dependent Variable (estimation Marginal?
technique) effects of the
Variable Whether or not Amount of regular Truncated
name regular apples apples purchased by r egressron
were purchased‘ those who bought some upon amount
_ robit -_-__-_-_____--_- __-"_"°"_s°"_ -
PX -.1505*" -l .5345 -.5216
(.0452) (.9899) (.7739)
M .0006* -.0031 .0011
(.0003) (.0071) (.0024)
Hh .0063 .6421 *" 2183‘"
(.0075) (.1650) (.0561)
Age -.0015** 0592“" 0201*”
(.0008) (.0171) (.0058)
Geo -.0145 9842* .3346"
(.0261) (.5422) (.1843)
Educ .0141 *** -.0362 .0123
(.0043) (.0928) (.0315)
Groc .0278 -3.3003*“ -1.1218""
(.0380) (.7402) (.2516)
Org .0180 -1.5593 -.5300
(.0644) (1.4755) (.5016)
D .0158 .2806 .0954
(.0230) (.1806) (.1634)
a --- 3.4594 .....

 

 

‘= Significant at the 10% level,**= Significant at the 5% level,"*=
Significant at the 1% level

74

 

Table 16 (Cont’d)

‘The values listed in this column are the marginal values of the Probit and
thus give a change in the probability of buying the apples given a one unit
change in the explanatory variable

75

Table 17
Tobit coefficients and calculated marginal effects for ecolabeled apples

 

Variable

 

 

 

 

 

 

 

 

 

 

 

 

Tobit coeff. (aE(Y)/ax) (6E(Y*)/dx) (aF(z)/ax)
name __ ______________ ______
9.04305" 3.9981"* 3.0100 5252
(2.0001) (.8833)
-9.4586*" 4.1818" -3.1483 -5494
(1.6709) (.7360)
.0017 .0007 0006 .0001
(.0081) (.0036)
3451* .1526“ 1149 0200
(.1778) (.0787)
-0124 -0055 -0041 -0007
(.0182) (.0081)
-.9111 -.4028 -3033 -0529
(.6475) (.2861)
1915* .0847“ .0637 .0111
(.1079) (.0476)
-0495 -0219 -.0165 -.0029
(.5335) (.2359)
.2760 .1220 .0919 .0160
(.5346) (.2364)
-3.9809*** -1.7600"* -1.3251 -2312
(.8782) (.3913)
1.4056 .6214 .4679 .0816
(1.5141) (.6693)
-.0518 -0229 -0173 -.0030
(.5613) (.2482)

 

 

 

 

 

 

76

 

Table 17(cont’d)

 

          
 
 
 

  

name

Variable

  

Tobit coeff.

-2.7705
(2.4059)

 

(6E(Y)/dx)

 

(6E(Y*)/ax) *

    

 

(aF(z)/ax)

 

 

.6038
(6890)

.2958

.2074

.0265

 

1.5394"
(.7352)

.8042

.5640

.0722

 

 

 

6.7963

 

 

 

 

 

‘= Significant at the 10% level,“‘= Significant at the 5% level,""“=
Significant at the 1% level

‘The familiarity, health and environmental concern variables are separated

Likelihood Value‘: -1755.8419
Chi-Squared‘: 64.94, P-value‘: .0000
‘ These values for the equation without F, <I> and ‘1’.

77

from the other variables by a double line to indicate that they were estimated
in separate equations. The results above the double line were estimated in a
regession without these three variables included. Two separate equations
were run with familiarity (F) and the two motivational variables (II) and ‘1').
We only report the coefficients of these three variables from those last two
equations.

 

Tobit coefficients and calculated marginal effects for regular apples in a

Table 18

market with both types of apples

 

Variable

 

’ (aE(Y)/ax) (aE(Y*)/ax) ’ (aF(z)/ax)

 

 

 

 

 

 

 

 

 

 

 

 

Tobit
name coefficient __ _-___-___-_-___-_ ____ _ __ __
-7.0380"“* -2.9439**‘ -2.2581 -.5531
(1.5198) (.6317)
5.2287"‘”‘ 2.1871**“ 1.6776 .4109
(1.2693) (.5265)
.0030 .0013 .0010 .0002
(.0064) (.0027)
.0753 .0315 .0241 .0059
(.1352) (.0566)
.0068 .0028 .0021 .0005
(.0136) (.0057)
.9108“ .3810“ .2922 .0716
(.4789) (.2002)
-.0967 -.0405 -.0310 -.0076
(.0785) (.0329)
-.6010 -.2514 -.1928 -.0472
(.4023) (.1685)
.1789 .0748 .0574 .0141
(.4031) (.1686)
-2.0192"** -.8446**“ -.6478 -.1587
(.6687) (.2814)
-3.5886"" -1.5010“‘ -1.1514 -.2820
(1.4063) (.5861)
.1280 .0536 .0411 .0101
(.4274) (.1788)

 

 

 

 

 

 

78

 

Table 18 (Cont’d)

 

I Variable l Tobit
name coefficient

,, (aEm/ax)

(6E(Y*)/3X) (3F(l)/3X).

 

 

 

‘=i Significant at the 10% level,**= Significant at the 5% level,“"‘=

Significant at the 1% level

Likelihood Value: -l392.7057
Chi-Squared: 42.37, P-value: .0000

79

Tobit coefficients and calculated marginal effects for regular apples in a

Table 19

market without ecolabeled apples.

 

 

(6E(Y*)lax) (6F(z)/6x) ’

 

 

 

 

 

 

 

 

 

I Variable l Tobit (6E(Y)ldx)
name coefficient ___-__ __ __ _______ f
-.9419 -.6100 -.4277 -.0531
(1.0037) (.6500)
.0045 .0029 .0020 .0003
(.0076) (.0049)
.1334 .0864 .0606 .0075
(.1658) (.1074)
-.0233 -.0151 -.0106 -.0013
(.0166) (.0108)
-.2520 -.1632 -.1144 -.0142
(.5771) (.3738)
.0230 .0149 .0104 .0013
(.0960) (.0622)
-5.0969‘""‘ -3.3013""“ -2.3 145 -.2871
(.8179) (.5346)
-.8844 -.5728 -.4016 -.0498
(1.4169) (.9179)
.3799 .2461 .1725 .0214
(.5091) (.3298)
6.591 1 --- ..... .....

 

 

 

 

 

 

 

*= Significant at the 10% level,‘*= Significant at the 5% level,“'*=
Significant at the 1% level

Likelihood Value: -2354.083
Chi-Squared: 42.76, P-value: .0000

80

 

Table 20
Calculated Tobit elasticities of prices, income and significant variables
for all three types of apples

 

Variable

    

Re ; larA : [es in zrst mr_ke scenario

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Px - 1324 -.0601 -0723
M .0383 0174 .0209
Groc -.7332 -.3329 -.4003
Realm aaples in ire-3W
Px -1.7260 -.5538 -1.1722
Px’ 1.5690 .5034 1.0656
M .0443 .0142 .0301
Gen .0610 .0196 .0414
Groc -.5033 -.1615 -.3418
Org -.0302 -.0097 -0205
Px 1.5767 5248 1.0519
1’," -2.0207 -.6726 -l .3481
M 0174 .0058 .0116
Kb .2010 .0669 .1341 -
Educ 5294 1762 .3532
Groc -.7078 -.2356 -.4722
‘I’ 0559 .0261 .0299

 

81

 

 

 

Table 21

Q: Why did you choose to buy the ecolabeled‘ apples? (Open ended)
Wore: This question asked only to those who bought ecolabeled apples)

 

 

 

 

 

 

 

 

 

 

 

 

Stated Reason First Reason Second Third Reason
Percent Reason Percent
Respondents Percent Respondents
(N=496) Respondents (N=103)
Ecolabeled apples are
better for the 27.4 % 23.5% 23.3%
environment
To avoid chemicals in 20.9 % 18.4% 214%
my food
T” “7 them/1?“? the“ 18.9% 6.3% 2.9%
out of currosrty
Ecolabeled apples are
safer/healthier for me 16.2 % 32.5% 23.3%
and my family
The “Safe“ apple " 10.2 % 10.2% 9.7%
quality 15 better
Feels like] m donng 5.5 % 3.1% 4.8%
something good
I we “wit mm" 1.6 % 2.4% 4.9%
generations
Don’t know 1.4 % 3.1% 6.8%
Refused .6% .4% 2.9%

 

 

 

 

‘Ecolabeled apples refer to both IPM and ECO apples

82

 

Table 22

Q: Why did you choose not to buy the ecolabeled apples‘? (Open-ended)
(Note: This question asked only to those respondents who did not buy

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ecolabeled apples)
Stated Reason First Reason Second Third
Percent Reason Reason
Respondents Percent Percent
Respondents Respondents
|| (N=3 52) (N=87) fl=362
— a
They are too expensive 46.3 % 14.9% 22.2%
Do not know enough
about ecolabels to buy 21.6 % 42.5% 44.4%
them
Prefer own brand of 8.5 % 6.9% 2.8%
apples
Regular apples are just
as safe and healthy for 8.2 % 14.9% 2.8%
my family
Do not trust/belleve the 4.5 % 6.9% 0.0%
ecolabel
They are. an advertnsrng 2.3 % 4.6% 8.3%
gimmick
Do not trust the o o o
governmen t/USD A 2.0 /o 0.0 /o 0.0 /o
Do not like to try new 1.1 % 1.1% 2.8%
things
Don’t know 4.3 % 5.7% 13.9%
Refused 1.1% 2.3% 2.8%

 

‘Ecolabeled apples refer to both IPM and ECO apples.

83

 

Table 23
Comparison of results between Tobit, OLS and SUR for ecolabeled
apples in a market with ecolabeling

 

 

 

 

 

 

 

 

 

 

 

Variable Name Tobit model- OLS SUR
Marginal (Linear (Linear
effects functional functional
-Ec-01Wbe- we nuke” menri
PX 3.9981‘” 3.3109“ 3.3532"
(.8833) (1.3171) (1.3352)
[1.0200]‘
Px’ 4.1818” -2.9552**" -3.0055"""
(.7360) (1.0946) (1.1073)
(5581]
M .0007 -.0003 .0007
(.0036) (.0055) (.0056)
L0025]
Hh .1526“ .2191‘ .2183‘
(.0787) (.1188) (.1205)
L1346]
Age -.0055 -.0007 -.0016
(.0081) (.0120) (.0122)
L0083]
Gen -.4028 -. 1209 -.1184
(.2861) (.4252) (.4313)
L2784]
Educ .0847“ .0189 .0010
(.0476) (.0698) (.0702)
L0460]

 

 

 

 

‘The bracketed expression is White’s robust standard error given to correct

for heteroscedasticity.

84

 

Table 24 (cont’d)

 

 

effects

Variable Name Tobit model-
Marginal

 

OLS
(Linear
functional
0-

 

' SUR
(Linear
functional

form

 

E labele a 7 (es in the second market scenarzo

 

 

 

 

 

 

 

 

 

 

L -.0219 -.2645 -.2910
(.2359) (.3538) (.3588)
L3985]
C .1220 .3181 .3341
(.2364) (.3545) (.3597)
L3572]
Groc -1.7600*** -3.0334*** -3.0194***
(.3913) (.6007) (.6050)
[1.7205]
Org .6214 .2587 .2430
(.6693) (1.0457) (1.0528)
L3793]
D -.0229 -.3894 -.4089
(.2482) (.3749) (.3795)
_ gf .2909
Cross price elas -2.0207 -2.0846 -2.0914
Own price elas 1.5767 1.9061 1.9048
income elas .0174 -.0104 .0238
MSE --- 14.897 5.0001

 

 

 

‘= Significant at the 10%leve1,"= Significant at the 5% level,""=

Significant at the 1%level

85

Table 24

Comparison of results between Tobit, OLS and SUR for regular apples

in a market with ecolabeling

 

Variable Name Tobit model-
Marginal
effects

 

 

OLS
(Linear

functional

_orfm _

 

SUR
(Linear

functional

form

RNHLIa '6 WJWW" scenaw

 

 

 

 

 

 

 

 

Px -2.9439*** -1 .9829" -2.0070”
(.6317) (.8079) (.8164)
[.5929]'I
Px’ 2.1871“* .8512 .8736
(.5265) (.6703) (.6745)
L8512]
M .0013 .0036 ---—"
(.0027) (.0034)
L0036]
Hh .0315 .0827 .1082
(.0566) (.0731) (.0725)
L0827]
Age .0028 .0052 .0072
(.0057) (.0074) (.0074)
L0052]
Gen .3810" .3263 .3475
I (.2002) (.2609) (.2636)
J L3263]

 

‘ 'The bracketed expression is White’s robust standard error given to correct

for heteroscedasticity.

I’In order for SUR to run as SUR, the explanatory variables have to be

different between the two equations. Thus, income and education were left

out of the regular equation so that the equations would be different.

86

 

 

Table 24 (cont...)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Name Tobit model- OLS SUR
Marginal (Linear (Linear
effects functional functional
# form)___ form
f f R r _g_ 7 (es in the second market scenario
Educ -.0405 -.0573 --- <
(.0329) (.0426)
L0629]
L -.2514 -.4476** -.4495"
(.1685) (.2175) (.2196)
[22338]
C .0748 -.0020 -.0073
(.1686) (.2181) (.2198)
L2161]
Groc -.8446*** -1.7885*“ -1.8082*“
(.2814) (.3663) (.3700)
L9545]
Org -1.5010*"' -.9469 -.9457
(.5861) (.6415) (.6378)
L3853]
D .0536 .2925 .3139
(.17 88) (.2300) (.2322)
T' _ [.2646]
Own price elas. -1.7260 -1.3406 -1.3379
Cross price elas. 1.5690 .7043 .7122
Income elas. .0443 .1468 ---
_ _’___ ---______ 3.0452 3.0571

 

 

 

 

 

‘= Sigmcfiant at 1° level,**= Significant at the 5% level,*"=

Significant at the 1% level

87

APPENDICES

88

APPENDIX A

DEVELOPMENT OF ECOLABEL DESCRIPTIONS AND PRICES”

The ecolabels presented in the questionnaire were developed over
several months. Descriptions of ecolabels were pretested in personal
interviews with consumers and refined based on the pretest results. Many
rounds of pretests were conducted before the final wording was selected.

The initial set of ecolabels was created by examining existing
agricultural ecolabeling programs in the U.S. These programs included the
Massachusetts-IPM Partners with Nature Program, the Wegmans IPM label 2
for canned com and standards developed by California Clean Growers.
Initially, we looked at three dimensions of the ecolabels: (1) whether process
standards (i.e., production practices) or performance standards (i.e.,
environmental outcomes) were used to define the environmental claim, (2)
how comprehensive the claim was (i.e., how many types of process or

performance standards were specified), and (3) whether certification was

 

30This section taken from J eifrey R. Blend, and Eileen 0. van
Ravenswaay, 1998, “ Demand for Ecolabeled Apples: Survey Methods and
Descriptive Results,” Department of Agricultural Economics Staff Paper No.
98-20 , Michigan State University, East Lansing, MI.

89

performed and, if so, by whom. Initial label descriptions consisted of
difi‘erent combinations of standards, comprehensiveness, and certifiers.
Examples of some of the initial ecolabel descriptions follow:
Example 1: Process standards which are non-comprehensive and producer
certified
Grown using Integrated Pest Management
Grown with 33% less of Pesticide X
Certified by the Smith Farms
Example 2: Performance standards which are non-comprehensive and
certified by the federal govemment
Production resulted in 75% less nitrates in groundwater
Superb rating on the Soil Health Index
Certified by the USDA
Example 3: Process standards which are comprehensive and certified by a
well-known consumer group
Used the safest fertilizers and pesticides on the market at the lowest possible
levels
Met 80% of the total points possible to qualrfi» as IPM certified

Produced in a manner which encourages healthy soil

90

Produced with 60% less of chemicals X, Y, and Z
Produced with 40% less irrigation water
Grown using 30% less total energy (in Kilowatts)

Certified by the Society for a Better America Consumer Group
Example 4: Performance standards which are comprehensive and not
certified

Improved 50% on the Soil Nutrient Yardstick over previous year
Doubled the number of Sandhill Crane and endangered Gray Wolves
observed on the farm
Water runofir from farm contained a 40% lower concentration of toxic
residue than average levels in the county
No detectable residues found upon this product
Introduced four new species of rare wildflowers to farm
Production resulted in 20% less on-farm solid waste than the year before
Energy-efficient transport of this product resulted in 5 0% less air emissions

Example 5: No process or performance standards and no certification

This product grown with a commitment to preserving our environment

Initial pretest interviews revealed that these descriptions were too

91

complicated and confusing. Participants did not like being read a ‘laundry
list’ of environmental attributes and thought the standards were very vague.
We also learned from agricultural scientists that the development of
performance standards was hampered by the current inability to measure
environmental performance.31 Consequently, we decided to omit the

distinction between the process and performance standards. The pretests did

 

indicate, however, that certification status and comprehensiveness of the
claim were important in the purchasing decision. As a result, they were
included as dimensions in the further rounds of ecolabel pretesting.

IPM ecolabel was chosen because several U.S. agricultural ecolabels
are based upon IPM standards including Partners with Nature in
Massachusetts and the Wegmans label on canned vegetables in New York

state. These existing labels provided a framework within which to create the

 

' 31Discussions of these difficulties found in S. Riha, L. Levitan, and J.
Hutson, “Environmental Impact Assessment: The Quest for a Holistic
Picture,” Proceedings of the Third National IPM Symposium, USDA, ERS,
Washington, D. C., 1997 and Wayne S. Roberts and Scott M. Swinton,
"Economic Methods for Comparing Alternative Crop Production Systems: A
Review of the Literature," American Journal of Alternative Agriculture,
1996, ll(1):10-l7.

92

IPM description.”

The ECO ecolabel was created to see if consumers distinguish other
farm-related environmental impacts. Unlike the IPM ecolabel, the ECO
ecolabel was not based upon an existing ecolabel.

Food labels at local food stores provided examples of simple and clear
language that respondents would recognize.33 Labels were examined on flesh
and frozen produce and juices, processed foods, health foods, and organic
foods.

The final versions of the ecolabel began with a definition of IPM or
ECO, depending upon which version the respondent received. This was
followed by the environmental claim itself. The IPM claim was:

0n the label it is stated that IPM means that the apples were

 

32The IPM definition and practices mentioned in the claim were
obtained from two web-sites and from the We grnan’s IPM label for canned
corn. Valuable assistance was also provided by Mark Whalen, Entomology
Dept. at MSU, Lois Levitan, Cornell University, Kurt Petzoldt with New
York IPM, Cornell Extension Station and Laura Tourte, UC Davis-extension.
Bill Coli at Umass at Amhurst and Bryan Hubbel at Univ. of Georgia at
Athens provided assistance with the specific wording of the IPM claim.

33Stores visited consisted of several supermarkets, a farmer’s markets
and an organic-oriented food co-op, the names of which are Meijers, Kroger,
Save-A-Lot, and East Lansing Food Co-Op or ELFCO, and the Haslett

Farmer’s Market.

93

grown using Integrated Pest Management. Signs and
brochures in the store explain that IPM uses a number of
different methods to prevent pest damage to fresh apples. These
methods lower the need for environmentally harmfitl practices
on the farm. For example, insects that damage apples are
controlled by their natural enemies. Fields are monitored for
pests to reduce unnecessary chemical usage. Apple varieties
are planted which resist disease. When pesticides are needed,
the least harmful are used.

The ECO claim was:
Signs and brochures in the store explain that “ECO-apples ”
are grown using techniques based on environmental principles.
These techniques result in more efficient use of water and
energy. They include the use of naturally occurring fertilizers
to protect water and soil quality. When possible, natural insect
control methods are used to prevent pest damage on fresh
apples. Man-made pesticides are used only as a last resort.
To help respondents visualize the ecolabel, a physical description was

provided. The size, shape, color and lettering upon the label were described.

94

Respondents were told that the letters on the label were IPM [or ECO] in
capital letters. We included three capital letters on all versions of the
ecolabel to ensure that the letter size would not affect demand.

The regular apple prices used in the market scenarios were chosen
based upon U.S. apple data. Retail apple prices fi‘om the months in which the
survey was given, November through February, were averaged over the last
five years of available data (1992-1996) resulting in a figure of $.88 per
pound.“ A central price of $.89 per pound was identified because most
apples in the U.S. are sold at a price which ends in ‘9' on a per pormd basis.”
Two additional prices were then chosen on either side of this figure at $.30
increments to represent variation about the average U.S. apple price.

Prices for the ecolabeled apple were selected by looking at the market
for organic apples. In van Ravenswaay and Blend (1997), it was suggested
that ecolabeled apples would best capture a significant portion of the apple
market by providing a cheaper environmental alternative to organic. Thus,

the ecolabeled apple prices used in our market scenario were kept at or below

 

34All prices given are in December, 1997 dollars.

3’It was decided from earlier work (van Ravenswaay and Hoehn, 1991)
that the prices given would be on a per pound basis since this was the most
common way respondents understood them.

95

 

 

the lowest organic prices.

In order to determine the lowest prices for organic apples, a price range
was identified. This was done by looking at prices in the Lansing area and
referring to studies by Jolly (1989 and 1991) and Hammit (1986) who
calculated price premiums for organic apples in California.36 The lowest
prices for organic apples were about $1 .39-51 .49 per pound.” This was set
as an approximate upper bound price for the ecolabeled apples.

Because the highest price for regular apples was $1.19/lb, the price
premium on ecolabeled apples could be no greater than 30 or 40 cents per
pormd if it were to stay roughly within its bound. Thus, the highest premium
was set at $.40/lb. The lowest premium was set at zero cents (i.e., same price
for regular and ecolabeled apples) so that an estimate could be made of the
difi‘erence in demand between ecolabeled and regular brands, ceteris paribus.

A third premium was set halfway between these two at $.20/lb.

 

”Stores visited included East Lansing Food Co-op, Kroger grocery
store, and the Haslett Farmer’s Market.

3“’It was found that organic prices ranged from $1 .SO/lb up to $2.50 per
pound from the sources consulted. The lower prices were fi'om price data
reported in Hammitt and Jolly. The highest price was seen at the local food
co-op.

96

APPENDIX B

SAMPLE SELECTION38

The Institute of Public Policy and Social Research (IPPSR) at
Michigan State University conducted the survey interviews. IPPSR
purchased the phone numbers from Gensys Sampling Inc. who randomly
generated 3945 numbers from a straight 48 state proportional sample.

IPPSR assigned a five digit case ID number to each phone number.
Each ID number was then randomly assigned to one of the 36 survey
versions. In this way, approximately equal numbers of respondents received
each survey version.

IPPSR dialed the phone numbers until they obtained the desired sample
size. The calls were placed between November 3, 1997 and February 11,
1998 and the survey interviews lasted an average of 7 minutes. Of the 3203

numbers dialed, 145 3 reached households with persons who were eligible for

 

38This section is based on Jeffrey R. Blend, and Eileen 0. van

Ravenswaay, 1998, “ Demand for Ecolabeled Apples: Survey Methods and
Descriptive Results,” Department of Agricultural Economics Stafl‘ Paper No.
98-20 , Michigan State University, East Lansing, MI.

97

participation in the survey.39 Of these eligible cases, 972 interviews were
completed resulting in a participation rate of 66.9 % (972/1453). Of the '
eligible households, there were 393 refusals. One thousand seven hundred
and fifty numbers or 54.6% of those dialed were ineligible for participation.
One thousand three hundred and twelve of the ineligible numbers consisted of
business lines, FAX or computer numbers and verified non-working numbers.
Three hundred and twenty-nine could not be contacted because either no one
answered or the line was constantly busy after 20 attempts. One hundred
thirty seven of the ineligible numbers reached respondents who were unable
to participate due to physical handicaps, respondents with language barriers,
those who were too young (teenager lines) and those who were absent during

the study period.

 

35'Eligibility required the number to be that of a residential household
and that the person who purchases food for the household be 18 years or
older.

98

APPENDIX C

SURVEY INSTRUMENT”0

Hello, my name is and I am calling from the Institute of
Public Policy and Social Research at Michigan State University. Is this
(phone number)?

We are calling on behalf of the Agricultural Economics Department at
Michigan State University. We are conducting a national survey of food
consumers. May I speak to the person who usually does the food shopping
for your household? Thank you.

The survey takes about 7 minutes. Before we begin, let me tell you
that any information you give me will be kept strictly confidential. Let me
also tell you that this interview is completely voluntary. Should we come to
any question that you don't want to answer, just let me know and we'll go on
to the next question.

I’d like to start by asking you some general questions about the fresh
apples you may buy for your household. By fresh apples, I am referring only
to raw apples. When answering the following questions, do not include
canned or processed apples, apple juice, apple sauce, or cider. Fresh apples
are in season in late summer and early fall.

Q1. Do you buy fi'esh apples for your household? [IF THEY ASK WHAT
IS MEANT BY HOUSEHOLD: Your household includes yourself,
your dependents, and persons with whom you share income and
household living expenses]

Yes—-[GO TO Q3]
No—-[GO TO Q2]
Don’t know
Refused-«[GO TO Q3]

99°91"

 

“This section is based on Jeffrey R. Blend and Eileen 0. van
Ravenswaay, 1998, “Demand for Ecolabeled Apples: Survey Methods and
Descriptive Results,” Department of Agricultural Economics Staff Paper No.
98-20, Michigan State University, East Lansing, MI.

99

Q2. Could you tell me why you do not buy fresh apples for your
household? (ALLOW 2 CHOICES)

 

1. My household does not like/eat apples
2. Apples are too expensive
3. Apples are not available
4. Apples are unsafe or unhealthy
5. Other (Specify )
90. Do buy apples, but only in season or only occasionally—[GO TO
Q3]
98. Don’t Know
99. Refused
GO TO Q17

Q3. Where do you usually purchase apples?

Supermarket, grocery, or convenience store
Farmer’s market

Farm, roadside stand, orchard, or cider mill
Green grocer or hit specialty store
Organic food store or cooperative

Other (Specify )

Don’t know

Refused

99999939!“

Q4. Have you ever bought organic apples?

1. Yes—-[GO TO Q4a]

5. No--[GOTO Q5]

8. Don’t Know-«[GOTO Q5]
9. Refused-[GOTO Q5]

Q4a. When you buy apples, how often do you buy organic apples? Would
you say rarely, occasionally, often, or always?

Rarely
Occasionally
Often
Always

Do not know
Refused

.‘°.°°:“P°l"f"

190

Q5. How much have you seen, heard, or read about Integrated Pest
Management or IPM?

1. Nothing

2. A little

3. A fair amount

4. A great deal

8. Don’t know

9. Refused

Q6

Q6a.

Q6b.

Q7.

Imagine you are at the place where you normally buy fresh apples, and
you are planning to buy some. Suppose that all apple varieties and
sizes are the same price per pound whether displayed individually or
packaged in bags, boxes, or bushels. All your favorite varieties are on
display and are of the size and quality you prefer. There are no special
Sales on any other fresh fruits. If the price of apples was per
pouird 3nd no other place was selling them for less, would you buy any
app es.

1 Yes-[GOTO Q7]

5. No--[GOTO Q6a]

8. Don’t krrow--[GO TO Q10]
9. Refirsed--[GO TO Q10]

Why would you not buy any apples?

1. Too expensive-«[GO TO Q8]

2. Other (Specify_)-—[GO TO Q8]

9. Respondent contradicts what was read in scenario, Specify
L_)- [GO TO 06b]

98. Do not known-[GO TO Q8]

99. Refirsed--[GO TO Q8]

If [INSERT RESPONDENT’S CONTRADICTION FROM
Q6a] was not a factor, and if the price of apples was [INSERT PRICE
FROM Q6] _ per pound, would you buy any apples?

1. Yes-«[GO TO Q7]
5. No—-[GO TO Q8]
8. Don’t know

9. Refused

How much would you buy?

101

Q8.

Q9.

Q10.

RECORD NUMBER and UNITS

Ifyou were in exactly the same setting, and the price of apples was
per pound, would you buy any apples?

1. Yes--[GO TO Q9]

5. Non-[GO TO Q10]

8. Don’t krrow--[GO TO Q10]
9. Refused—[GO TO Q10]

How much would you buy?
RECORD NUMBER and UNITS

[There are four versions to Q10. Respondents are assigned randomly
to receive one of these version. Versions 1 and 3 present the
respondent with an IPM ecolabel and versions 2 and 4 present the
respondent with an “ECG ” ecolabel. Versions 1 and 2 include the
statement that the ecolabel is certified by the USDA, whereas versions
3 and 4 do not.]

[VERSIONS 1 AND 3 OF Q10, WITH IPM ECOLABEL]

Imagine you are in the same setting and you notice something new at
your store. Your store sells the same apples it always has and apples
grown a new way. The apples grown the new way have small, round
half-inch wide labels that have a white background and green lettering.

The letters on the label are IPM in capital letters. [The label states that
IPM apples are certified by the USDA to have been grown and handled
according to government standards] On the label it is stated that IPM
means that the apples were grown using Integrated Pest Management.
Signs and brochures in the store explain that IPM uses a number of
different methods to prevent pest damage to fresh apples.

 

These methods lower the need for environmentally harmful practices
on the farm. For example, insects that damage apples are controlled by
their natural enemies. Fields are monitored for pests to reduce
unnecessary chemical usage. Apple varieties are planted which resist
disease. When pesticides are needed, the least harmfirl are used.

You are planning to buy fresh apples of the size and quality you prefer.
The price of the regular apples is the same as before, (INSERT PRICE
FROMOQ6) per pound and the price of the [Certified] IPM

apples rs per pound. Would you buy the regular apples, the

102

IPM apples, some of both, or none at all?

Regular only-«[GO TO QlOa]
IPM only—[GO TO Q10b]
Some of both-«[GO TO Qloa]
None at all-«[GO TO Q15]
Don’t lcnow--[GO TO Q16]
Refused-[GO TO Q16]

PFPP’N?‘

[The bracketed sentence involving USDA certification of IPM and the
bracketed word “certified ” are included only in version 1 of the apple
scenario. They are not included in version 3]

Q10

[VERSIONS 2 AND 4 WITH THE “ECO” ECOLABEL]

Imagine you are in the same setting and you notice something new at

your store.

Your store sells the same apples it always has and apples grown a new
way. The apples grown the new way have small, round half-inch wide
labels that have a white background and green lettering.

The letters on the label are E C O or “Eco” in capital letters. [The
label states that ECO-apples are certified by the USDA to have been
grown and handled according to government standards] Signs and
brochures 1n the store explain that “ECO-apples” are grown using
techniques based on environmental principles. These techniques result
in more efficient use of water and energy. They include the use of
naturally occurring fertilizers to protect water and soil quality. When
possible, natural insect control methods are used to prevent pest
damage on fresh apples. Man-made pesticides are used only as a last
resort.

You are planning to buy fresh apples of the size and quality you would

prefer. The price

Q6)—

of the regular apples is the same as before, (INSERT PRICE FROM
per
pound and the price of the [Certified] ECO-apples rs _per pound.

Would you buy

the regular apples, the ECO-apples, some of both, or none at all?

1. Regular only-«[GO TO Q10a]
2. IPM only-«[GO TO Q10b]

103

 

99°91”

Some of both-[GO TO Q10a]
None at all--[GO TO Q13]
Don’t krrow--[GO TO Q17]
Refused-«(GO TO Q17]

[The bracketed sentence involving government certification of EC 0 apples
and the bracketed word “certified” will be included only in version 2 of the
apple scenario. It will not be included in version 4]

Q10a. How much of the regular apples would you buy?
RECORD NUMBER AND UNITS (if blank record zero)

Q10b. How much of the IPM apples would you buy?

on

Q12

RECORD NUMBER AND UNITS (if blank record zero)

Why would you buy the [IPM/ECO] apples? (Open-ended) (Allow up
to 3 choices)

PP°P‘.V'PP’.N.'"

IPM apples are better for the environment

I care about future generations/firture of our planet

IPM apples are safer and healthier for my family and myself
To avoid chemicals in my food

Feels good/feels like I’m doing something good

Other (Specify )

Don’t know

Refused

Why would you not buy the [IPM/Eco] apples? (Open-ended) (Allow
up to 3 choices)

wave? we

They are too expensive/Not worth it-[GO TO Q13]

The regular apples are just as safe/healthy for my family-{GO
TO Q13]

I do not trust/believe the IPM label—[GO TO Q13]

I do not know enough about IPM to buy them—~[GO TO Q13]
IPM is an advertising gimmick-«[GO TO Q13]

Other (Specify )--[GO TO Q13]

Respondent contradicts what was read in scenario, Specify

(_)—-[GO TO Q12a]
Do not known-[GO TO Q13]

104

 

99. Refused-[GO TO Q13]

Q12a Imagine that this (INSERT RESPONDENT’S CONTRADICTION
FROM Q12) is not a factor. The price of the regular apples is the same
as before, (INSERT PRICES FROM Q10) per pound, and the
price of the [Certified] ECO-apples is per pound. Would you
buy the regular apples, the ECO-apples, some of both, or none at all?

1. Regular only-[GO TO QlOa]
2. IPM only—[GO TO Q10b]

3. Some of both-«[GO TO Q10a]
4. None at all—-[GO TO Q13]

8. Don’t lcnow--[GO TO Q17]

9. Refused—{GO TO Q17]

Q13 Suppose that the price of the regular apples is the same' as before,
[INSERT THE PRICE GIVEN FROM Q6] per pound and no
other place is selling them for less. The price of the [IPM/Eco] apples
is now per pound. Would you buy the regular apples, the

IPM/Eco apples, some of both, or none at all?

Regular only-«[GO TO Q13a]
IPM only—[GO TO Q13b]
Some of both--[GO TO Q13a]
None at all—-[ GO TO Q17]
Don’t know--[GO TO Q17]
Refirsedu-[GO TO Q17]

99°?pr

Q13a. How much of the regular apples would you buy?
RECORD NUMBER AND UNITS (if blank record zero)

Q13b. How much of the IPM apples would you buy?
RECORD NUMBER AND UNITS (if blank record zero)

DEMOGRAPHIC QUESTIONS

[If they answered yes to Q], then use the following statement only. If they
answered no to Q], then use the following statement and add on the
alternate statement]

105

These last few questions for statistical purposes only and cannot be linked to
you in any way.

[If they answered no to Q], then use the following statement]

Although you do not buy apples for your household, this information is still
valuable to the survey.

Q17. Are you male or female? [ASK ONLY IF IN DOUBT] (Circle one)

1. Female

2. Male
Q18a. Including yourself, how many people in your household are in the
following age categories? (Write in the numbers)

Under 5 years?

<0-10> Under 5 years

<98> Don’t Know
<99> Refused
Q18b. 5 to 17 years?
<0-10> 5 to 17 years
<98> Don’t Know
<99> Refused
Q18c. 18 to 64 years?
<0-10> 18 to 64 years
<98> Don’t Know
<99> Refused
Q18d. 65 years or older?
<0-10> 65 or older
<98> Don’t Know
<99> Refused

Q19. In what year were you born?

19 <00-79> YEAR BORN
18 <90-99> YEAR BORN

106

<998> Don’t know
<999> Refused

Q20. Please indicate the highest level of education you have completed.

(Circle one).

0.

1-11.

12.

13-15.

16.
17.
18.
20.
98.
99.

Did not go to school

Grade school

High school graduate or GED holder

Some college (One to three years)
College Graduate (Four years)
Some graduate school or professional (law, medical) school
Graduate degree
Technical School or Junior College Graduate
Do not know
Refused

To find out if people with different financial situations make difl‘erent food
choices, we’d like to know the general range of incomes of all households we
interview. Your answers will be kept strictly confidential.

Q21. Now, thinking about your household’s total annual income fi'om all
sources (including your job), did you household receive $30,000 or more in
1996? (Circle one)

1.
5
98.
99.

Yes ($30,000 OR MORE) [GO TO INC4]

No (Less than $30,000) [GO TO INC2]
DON'T KNOW-NO OPINION
REFUSED-NO ANSWER

>INC2< Was it $20,000 or more?

0.
1.
5
98.
99.

NOT APPLICABLE
Yes ($20,000 - 29,999)

No (LESS THAN $20,000)
DON'T KNOW-NO OPINION
REFUSED-NO ANSWER

>INC3< Was it $10,000 or more?

0.
1.
5

98.

NOT APPLICABLE
Yes ($10,000 - 19,999)

No (LESS THAN $10,000)
DON'T KNOW-NO OPINION

107

99.

REFUSED-NO ANSWER

>INC4< Was it $60,000 or more?

0.
l.
5
98.
99.

NOT APPLICABLE

Yes (more than $60,000)[goto INC7]

No (more than $30,000 and less than $60,000)[goto INCS]
DON‘T KNOW-NO OPINION

REFUSED-NO ANSWER

>INC5< Was it $40,000 or more?

0.
1.
5
98.
99.

NOT APPLICABLE

Yes ($40,000 or more)[goto INC6]
No ($30,000 - $39,999)

DON'T KNOW-NO OPINION
REFUSED-NO ANSWER

>INC6< Was it $50,000 or more?

0.
1.
5
98.
99.

NOT APPLICABLE

Yes ($50,000 - 59,999)

No ($40,000 LESS THAN $49,999)
DON'T KNOW-NO OPINION
REFUSED-NO ANSWER

>INC7< Was it $80,000 or more?

0.
1.
5
98.
99.

NOT APPLICABLE

Yes ($80,000 or more)[goto INC9]

No (More than $60,000 LESS THAN $79,999) [goto INC8]
DON'T KNOW-NO OPINION

REFUSED-NO ANSWER

>INC8< Was it $70,000 or more?

0.
l.
5
98.
99.

NOT APPLICABLE

Yes ($70,000 - 79,999)

No ($60,000-369.9999)
DON'T KNOW-NO OPINION
REFUSED-NO ANSWER

>INC9< Was it $100,000 or more?

108

0.
1.
5

98.
99.

NOT APPLICABLE

Yes ($100,000 or more)[goto INCl 1]
No ($80,000-$99,999)[gotoINC10]

DON'T KNOW-NO OPINION
REFUSED-NO ANSWER

INCIO Was it $90,000 or more?

0.
1.
5
98.
99.

NOT APPLICABLE

Yes ($90,000 - 99,999)

No ($80,000-$89,9999)
DON'T KN OW-NO OPINION
REFUSED-NO ANSWER

INC11< Was it $110,000 or more?

0.
l.
5
98.
99.

NOT APPLICABLE

Yes ($110,000 - or more)

NO ($100,000-5109,9999)
DON'T KNOW-NO OPINION
REFUSED-NO ANSWER

>NC12< Was it $150,000 or more?

0.
1.
5
98.
99.

NOT APPLICABLE

Yes ($150,000 to $200,000)
No ($1 10,000-$149,9999)
DON'T KN OW—NO OPINION
REFUSED-NO ANSWER

>NC13< Was it $200,000 or more?

0.
1.
5
98.
99.

NOT APPLICABLE

Yes ($200,000 or more)

No ($150,000-$199,9999)
DON'T KN OW-NO OPINION
REFUSED-NO ANSWER

109

 

BIBLIOGRAPHY

110

BIBLIOGRAPHY

Ahmad, Mubariq, Eoolabeling and Tropical Timber. Linking Consumer

Aw wmnoss goo Sustainable Forest Utiligatin 1n,o Ph. D. Dissertation, 1997, East
Lansing, MI: Michigan State University.

Blend, Jeffrey R. and Eileen 0. van Ravenswaay, “Demand for Ecolabeled
Apples: Survey Methods and Descriptive Results,” 1998, Department of
Agricultural Economics Stafi‘ Paper No. 98-20, Michigan State University,
East Lansing, MI.

Buzby, Jean C., Jerry R. Skees, and Richard C. Ready, “Using Contingent
Valuation to Value Food Safety: A Case Study of Grapefruit and Pesticide

Residues,” in Julie Caswell (ed), Valoing Food Safog and Notritioo, 1995,

Boulder, San Francisco, Oxford: Westview Press.

Cragg, J ., “Some Statistical Models for Limited Dependent variables with
Application to the Demand for Durable Goods,” Econometrica, 1971,
39:829-844.

Eom, Young Suk, “Pesticide Residues and Risk and Food Safety Valuation:
A Random Utility Approach,” American Journal of Agricultural Economics,
1994, 76(4):760-77 1.

Food Marketing Institute (FMI) W
W, 1997, Washington, D. C.: Food Marketing Institute.

Fin, T. and P. Schmidt, “A Test of the Tobit Specification Against an
Alternative Suggested by Cragg,” Review of Economics and Statistics, 1984,
66:174-177.

Grodsky, Jamie A., “Certified Green: The Law and Future of Environmental
Labeling,” Y 1 oum R 1 ti n, 1993,10:147-227.

Harnmit, James, Organic Carrots; Congomor Willingoogs to Pox to Rooooo
Eooo Born; Risks, 1986, Santa Monica: The RAND Corp, R-3447-EPA.

The Hartman Group, F an he Envir nm 11 ' A ns m r’ P r 'v
1996, Bellevue, WA: The Hartman Group. (Summer).

111

Hollingsworth, Craig S. ,William M. Coli and Ruth V. Hazzard, Editors,

tea t dPestMan men Ma sa h set uid line _0mm0!1\ .- 1
Dofinitions, University. of Mass. Extension IPM Program. Project number
94-EPMP-l-0049.

Jolly, D., Consmor Willingooss to Pay Prioo Premiums for thaoio Aoolos
Moog, March 1989, Department of Agricultural Economics, University

of California, Davis.

Jolly, D., “Differences Between Buyers and Non-buyers of Organic Produce
and Willingness to Pay Organic Price Premiums,” Agribusiness, 1991, 9:97-
1 l 1.

McDonald, J. F. and R. A. Moffitt, “The Uses of Tobit Analysis,” Review of
Economics and Statistics, 1980, 62(2):318-321.

Mitchell, Robert C. and Richard T. Carson, V
V 1 nM h ,1989, WashingtonD..:C

Resources for the Future.

Nonis, Patricia E. and Sandra S. Batie, “Virginia farmers’ Soil Conservation
Decisions: An Application of Tobit Analysis,” Southern Journal of
Agricultural Economics, July 1987: 79-90.

Owens, Nicole N., Scott M. Swinton, and Eileen O. van Ravenswaay,
“Farmer Demand for Safer Corn Herbicides: Survey Methods and Descriptive
Results,” 1997, East Lansing, MI: Michigan State University. Research
Report 547 for the Michigan Agricultural Experiment Station.

Riha, S., L. Levitan, and J. Hutson, “Environmental Impact Assessment: The
Quest for a Holistic Picture,” Proceedings of the Third National IPM
Symposium, USDA, ERS, Washington, D. C., 1997.

Roberts, Wayne 8., and Scott M. Swinton, “Economic Methods for
Comparing Alternative Crop Production Systems: A Review of the
Literature,” American Journal of Alternative Agriculture, 1996, 11(1): 10-17.

Sparling, Ed, Karen Wilken, and John McKenzie, “Marketing Fresh Organic
Produce in Colorado Supermarkets,” 1992, Colorado State University, Dept.
of Agricultural and Resource Economics Report.

Tobin, J., “Estimation of Relationships for Limited Dependent Variables,”
Econometrica, 1958, 26(1):26-36.

112

United States Bureau of the Census, Statistical Abstract of the Unitoo Statos;
1996 (116th edition), Washington, D.C., 1996.

United States Department of Agriculture, Economics Research Service, Emit
and flea Nuts Situation and Qotlook Report, March 1997, ITS-279, p. 5.

United States Department of Agriculture, Economics Research Service, Eat

goo Treo ths Simation and Outlook Repog. Yearbook Issoo, October 1996,
PIS-278, p. 6.

United States Environmental Protection Agency, Office of Pollution
PreventionandToxics, VI in fEn r nm 11 1M k in T
Mates, February 1993a, Washington, D. C. EPA741-R-94-003.

United States Environmental Protection Agency, Office of Pollution

Prevention and Toxics, Status Report on the Use of Environmental Lafils
mamas, September 1993b, Washington, DC. EPA742-R-9-93-001.

United States Environmental Protection Agency, Office of Pollution

Prevention and Toxics, :1 ho Us; of Life Cyolo Assossmont in Bum ootal
Laboljog, September 1993c, Washington, DC. EPA742-R-99-003.

United States Environmental Protection Agency, Office of Pollution
Prevention and Toxics, Detorminants of Effaoti ryon ass for Environmoptal

Wm April 1994 Washington D C EPA742-
R-94-001.

van Ravenswaay, Eileen 0. and John P. Hoehn, “Contingent Valuation and
Food Safety: The Case of Pesticide Residues in Food,” 1991a, East Lansing
MI: Michigan State University. Department of Agricultural Economics Stafi‘
Paper No. 91-13.

van Ravenswaay, Eileen 0. and John P. Hoehn, "Consumer Willingness to
Pay for Reducing Pesticide Residues in Food: Results of a Nationwide
Survey,” 1991b, East Lansing, MI: Michigan State University. Department of
Agricultural Economics Staff Paper No. 91-18.

van Ravenswaay, Eileen 0., flblic Poroeptions of Ago'ohomioals, 1995,
Ames, IA: Council on Agricultural Science and Technology, Task Force

Report No. 123.

van Ravenswaay, Eileen 0., “Emerging Demands on Our Food and
Agricultural System: Developments in Ecolabeling,” 1996, East Lansing, MI:
Department of Agricultural Economics Staff Paper No. 96-88. -

113

van Ravenswaay, Eileen O. and Jeffrey R. Blend, “Using Ecolabeling to
Encourage Adoption of Innovative Environmental Technologies in
Agriculture,” 1997, East Lansing, MI: Michigan State University.
Department of Agricultural Economics Staff Paper No. 97-18. Prepared for
the symposium on “Flexible Incentives to Promote the Adoption of
Environmental Technologies in Agriculture, ” held on June 8-10, 1997,
Gainesville, FL.

van Ravenswaay, Eileen 0. and Jennifer B. Wohl, “Using Contingent
Valuation Methods to Value the Health Risks from Pesticide Residues When
Risks are Ambiguous,” in Julie Caswell (ed.), Valuing Food Safoty and
Nutrition, 1995, Boulder, San Francisco, Oxford: Westview Press.

Wohl, Jennifer B., Eileen O. van Ravenswaay and John P. Hoehn, “The
Effect of Ambiguity on Willingness to Pay for Reduced Pesticide Residues,”
1995, East Lansing, MI: Michigan State University. Department of
Agricultural Economics Staff Paper No. 91-18.

1.”-.--

114