Efficacy of environmental labeling


Material Information

Efficacy of environmental labeling an economic analysis with two examples from Florida agriculture
Physical Description:
xi, 200 leaves : ill. ; 29 cm.
Athearn, Kevin R., 1968-
Publication Date:


Subjects / Keywords:
Food and Resource Economics thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Food and Resource Economics -- UF   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 2004.
Includes bibliographical references.
Statement of Responsibility:
by Kevin R. Athearn.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 003100327
sobekcm - AA00004687_00001
System ID:

Full Text







Copyright 2004


Kevin R. Athearn


I would like to thank the members of my supervisory committee. Tom Spreen

helped me formulate the initial research topic and provided much encouragement and

guidance throughout the Ph.D. program and research process. James Stems helped keep

me focused and contributed valuable input, especially on the grant proposal, organic

citrus research, and later stages of the dissertation writing process. Clyde Kiker has been

a source of inspiration for creative thinking throughout my graduate studies. Sherry

Larkin provided much feedback and thoughtful critique, especially on the grant proposal

and early stages of my research. Steven Slutsky provided valuable comments.

I would like to acknowledge the contributions of other individuals. Chris Andrew

has been a mentor and source of encouragement during my graduate studies. Donna Lee,

Jim Seale, and Bob Degner contributed to the early stages of my research. Also, I would

like to thank the organic citrus growers and handlers who participated in the research.

I am grateful for the funding provided by the Food and Resource Economics

Department to support my graduate studies, travel, and research opportunities. Also, the

University of Florida Alumni Fellowship contributed substantially. Finally, I am

thankful for the grant provided by the Organic Farming Research Foundation.

Last, but not least, I would like to thank my friends at the University of Florida and

my family for their support. In particular, I would like to thank my wife, Lisa, for her

love and friendship, and for her advice, encouragement, and patience.



ACKNOWLEDGMENTS ....................................................... .................................. iii

LIST O F TA B LE S ............................................................................................................ vii

LIST O F FIGU RES ..................................................................................................... ix

ABSTRACT............................................................................................................................ x


1 IN TRO D U CTION .................................................................................................. 1

Background on Eco-Labels.............................................. ..................................... 1
General Description of Eco-Labeling................................ ............................ 1
Examples of Eco-Labels................................................................................ 3
Special Characteristics of an Eco-Label.......................................... ..............8
Intentions behind Eco-Labeling .......................................................................10
Research Questions and Objectives.................................................................. 12

2 CONSUMER DEMAND RESPONSES TO ECO-LABELS.............................15

Literature Review .................................................................................................15
Product Characteristics and Quality in Consumer Theory ..................................15
Information and Uncertainty in Consumer Decisions.......................................25
Theory of Impure Public Goods.......................................................................28
Prior Model of Eco-Labeling.......................................................................29
Empirical Research......................................................................................30
Model of Consumer Response to Eco-Labels ....................................................36

3 PRODUCER SUPPLY RESPONSES TO ECO-LABELS ......................................46

Literature R eview ..................................................................................................46
Theoretical Models of Producer Decisions .................................................46
Empirical Research......................................................................................51
Model of Producer Response to Eco-Labels ......................................................53

EN V IRON M EN T ...................................................................................................60

Literature Review .................................................................................................60
Conceptual Background and Assumptions............................................................. 64
Two-Product, Partial-Equilibrium M odel..................................................... ... ..66
Price-Endogenous Programming M odel..................................................................78
Summary of Environmental Effects ................................... ....................................97

ALTERNATIVE................................................................................................ 103

Literature Review ...............................................................................................103
Environmental Effectiveness............................................................................... 104
Pareto Efficiency ................................................................................................. 108
Cost-Effectiveness ....................................................................................................117
Equity............................................................................................................ ........... 125
Conclusions on Efficacy........................................................................................ 128


Introduction....................................................................... ..................................132
Research Objectives......................................................................................... 133
Research M ethod ................................................................................................135
Acreage, Production, and M arkets.................................................. ..................136
Acreage............................................................................................................ 136
Yields and Production Volumes................................................................... 139
M market Channels.......................................................................................... 140
Grove and Grower Characteristics.......................... .............. ............................ 144
Farm Sizes ...................................................................................................144
Other Grower Characteristics............................................................................ 145
Organic Citrus Grower Typology................................................................. 147
Grove Care Practices, Costs, and Profitability ....................................................... 150
Organic Grove Care Practices ........................................................................150
Organic Grove Care Costs............................................................. ......... 152
Single Enterprise Profitability Analysis ........................................................ 155
Partial Budgets.............................................................................................159
Investment Analysis ........................................ ............... ...........................163
Incentives, Disincentives, and Alternatives ............................................................64
Reasons for Growing Organically ................................................................... 164
Problems and Difficulties............................................................................. 166
Alternative Land Uses .................................................................................. 168
Dissemination of Organic Knowledge................................................................. 169
Current Information Sources ................................... ...................................... 169
Research Needs ...........................................................................................170
Summary of Findings ...............................................................................................171
Conclusions and Avenues for Further Research...................................................175

7 CONCLUSION............................................... ................................................... 178


EQU ILIBRIU M M O DEL....................................................................................... 184

B GAMS PROGRAM FOR PRICE-ENDOGENOUS MODEL................................. 189

LIST O F REFEREN CES ............................................................................................ 191

BIOGRAPHICAL SKETCH ......................................................................................200


Table page

2-1. First-order conditions for four different consumer types ........................................41

4-1. Two-product, partial-equilibrium model results...................................................78

4-2. Price-endogenous programming model results under 1st market scenario................91

4-3. Price-endogenous programming model results under 2nd market scenario ..............93

4-4. Summary of price-endogenous programming model results...................................95

6-1. Estimated acreage for certified organic citrus in Florida by season......................137

6-2. Regional distribution of organic citrus acreage..................................................... 138

6-3. 2003-04 organic production estimates by variety..............................................140

6-4. Organic grove care practices. ................................... ...........................................151

6-5. Distribution of growers among grove care cost categories ...................................153

6-6. Representative production budgets............................ ..........................................154

6-7. Gross margin and average variable cost estimates for organic grapefruit .............157

6-8. Gross margin and average variable cost estimates for organic round oranges........157

6-9. Gross margin and average variable cost estimates for tangerines .........................157

6-10. Short-run break-even yields (boxes per acre).......................................................158

6-11. Short-run break-even prices ($/box).......................................................................158

6-12. Partial budget comparing conventional and organic grapefruit in the Indian River
region under high-cost, fresh, cultural programs ........................................ ...160

6-13. Partial budget comparing conventional and organic Valencia oranges in the central
region under medium-cost cultural programs for processed marketa...................161

6-14. Importance of factors in growers' decision to adopt organic methods..................165

6-15. Difficulties associated with growing certified organic citrus..............................167

6-16. Sources of organic citrus production information...............................................170

6-17. Sources of organic citrus marketing information...............................................170

6-18. Highest priority research needs ...........................................................................171

6-19. Additional concerns among organic citrus growers ............................................171


Figure page

2-1. Constraint set in 3-dimensional Y-X-B space...................................................42

2-2. Constraint set in 2-dimensional X-B space ..........................................................43

4-1. First market equilibrium scenario...................................................................... 70

4-2. Second market equilibrium scenario. .....................................................................75

4-3. Conceptual diagram of the price-endogenous model ..............................................79

4-4. Distribution function representing willingness to pay a premium for eco-label.......82

6-1. Distribution of groves by size category................................................................145

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



Kevin R. Atheam

August 2004

Chair: James A. Stems
Cochair: Thomas H. Spreen
Major Department: Food and Resource Economics

We evaluated the efficacy of voluntary environmental labeling programs and

compiled data on the organic citrus sector in Florida. Using ornamental plants as an

example, the theoretical analysis produces conclusions that are relevant for organizations

that develop eco-label programs, environmentalists, and policy-makers. An empirical

study of organic citrus provides information valuable to citrus growers and handlers,

agricultural researchers and extension agents, as well as policy-makers.

First, we developed models of consumer and producer response to voluntary

environmental labeling programs. These models draw on economic theory relating to

product quality and characteristics, information, and public goods.

Second, we analyzed the potential effectiveness of eco-labeling for protecting the

environment under various market conditions. We used a two-product,

partial-equilibrium model and a price-endogenous programming model to identify and

simulate the effects of eco-labeling on production quantities, land-use, and the

environment. Our analysis provides insight regarding the conditions under which

environmental labeling is most effective at protecting the environment and regarding

possible unintended consequences of eco-labeling.

Third, we investigated the potential for voluntary eco-label programs to serve as an

effective alternative to government-mandated environmental policies. We evaluated

efficacy, in terms of environmental effectiveness, Pareto efficiency, cost-effectiveness,

and equity. Our analysis identifies several factors that reduce the efficacy of voluntary

labeling programs relative to government-mandated environmental policies.

Fourth, we collected primary data on the organic citrus sector, using in-depth and

telephone interviews with growers and handlers in Florida. Data include acreage,

production volumes, market channels, production practices and costs, grower

characteristics, incentives and disincentives for organic production, alternatives,

information sources, and research needs. In addition to providing general qualitative and

quantitative information on the sector, our study of organic citrus connects to and

supports the preceding theoretical analysis, regarding alternative land uses and incentives

for adoption of organic certification.


Voluntary certification and labeling programs have proliferated as part of a

market-oriented approach to reducing adverse environmental impacts of agriculture and

industry. The term "eco-labeling" is commonly used to describe these programs. With a

focus on agriculture, we used economic theory and methods to analyze the potential

effectiveness of eco-labels at improving environmental outcomes and their efficacy as an

environmental policy alternative. An environmental management practice (EMP) label

for ornamental plants serves as an example, and a case study of organic citrus in Florida

complements the theoretical analysis.

Background on Eco-Labels

What is an eco-label? What are some examples of eco-labels? What

characteristics make an eco-label different from other types of labels? What is the

general intent or purpose of an eco-label? These questions are addressed in the following


General Description of Eco-Labeling

Environmental labeling programs take various forms. Labeling programs may rely

on first-party or third-party verification. Labels may contain positive, negative, or neutral

messages. Programs may be mandatory or voluntary. A label may be a seal-of-approval,

may verify that a single environmental criterion has been met, may provide a hazard

warning, or may disclose information in a report-card format (U.S. EPA 1998). Criteria

used for environmental certification and labeling may include a full life-cycle assessment,

or apply only to one stage in the product life-cycle (e.g., production or consumption

only). Also, labels may be relevant at different stages of the marketing channel. For

example, labels may be intended for retail consumers or for business-to-business


Our study considers voluntary, third-party verified labeling programs that convey

positive messages to the consumer in the form of a seal-of-approval pertaining to the

environmental impact of the production process. The label or seal-of-approval is

awarded by a third party (governmental or nongovernmental organization) certifying that

production of a good meets certain environmental standards. The seal-of-approval

implies that the labeled good's production caused less environmental harm than the

production of its unlabeled counterpart. This type of eco-label is common for products in

agriculture, forestry, and fisheries.

A wide variety of eco-labels can be found in the marketplace for food and fiber,

and for all sorts of consumer products. Consumers Union (2003b) describes and

evaluates more than 100 different eco-labels found in the United States. The term,

eco-label, has evolved to include certification criteria relating to social responsibility,

animal welfare, and other consumer concerns besides just environmental impact.

Consumers Union (2003a, p. El) defines eco-labels as "seals or logos indicating that an

independent organization has verified that a product meets a set of meaningful and

consistent standards for environmental protection and/or social justice." A brief

description of selected eco-labels found in agriculture, fisheries, and forestry follows.

These include Dolphin Safe, Marine Stewardship Council, Forest Stewardship Council,

Bird Friendly, Fair Trade, Free Farmed, Protected Harvest, The Food Alliance, USDA

Organic, and a new eco-label being developed for ornamental plants.

Examples of Eco-Labels

The dolphin safe label found on canned tuna is one of the most widely recognized

eco-labels in the United States. According to federal law, tuna labeled as dolphin safe

must not be caught by setting nets on dolphin (Teisl, Roe, and Hicks; Consumers Union

2003b). Vessels fishing with purse seine nets in the Eastern Tropical Pacific must have

independent observers aboard to verify that nets were not set on dolphin (Teisl, Roe, and

Hicks; Consumers Union 2003b). U.S. consumers do not have the choice to buy tuna that

is not considered dolphin safe, because the sale of such tuna is restricted by federal laws

and embargoes (Teisl, Roe, and Hicks). Essentially a mandatory program, the dolphin

safe label does not fit the category of voluntary eco-labeling that is the focus of this


The Marine Stewardship Council is a nonprofit organization that certifies fisheries

that are not being over fished, have low bycatch or other adverse ecosystem impacts, and

have effective management institutions (MSC 2002). Fish products from certified

fisheries may bear the MSC logo. Fisheries currently certified by the Marine

Stewardship Council include Alaska salmon, Burry Inlet Cockles, Loch Torridon

Nephrops, New Zealand Hoki, South West Mackerel handline fishery, Thames Herring,

and Western Australian Rock Lobster (MSC 2003). The Marine Stewardship Council

label is similar to the type of voluntary label discussed in our study, except that for MSC

certification the entire fishery is certified, not individual fishers.

The Forest Stewardship Council is an international organization that certifies wood

and wood products from sustainably managed forests (Consumers Union 2003b).

Certification criteria include social and economic impacts as well as environmental

impacts of the forest industry (Consumers Union 2003b; FSCUS). Specific principles

include protection of biodiversity, water resources, and forest ecosystem functions;

promotion of community economic well-being; and respect for workers' rights and

indigenous peoples' rights (FSCUS). Products that meet its certification requirements

may use the FSC label.

A variety of shade-grown or bird-friendly coffee labels exist. One such eco-label is

the Smithsonian Migratory Bird Center's Bird Friendly coffee label. Many species of

song birds found in North America migrate to Central and South American coffee-

growing regions during the winter months. Whereas coffee often is grown on unshaded,

monoculture plantations that reduce bird habitat, bird-friendly certification recognizes

traditional coffee farms that maintain a tree canopy or other aspects of bird habitat.

(Consumers Union 2003b; Gobbi; Smithsonian Migratory Bird Center).

The Fair Trade label applies primarily to social responsibility and equitable trading

practices, but considers environmental impact as well. The nonprofit organization

TransFair USA, a member of Fair Trade Labeling Organizations International, certifies

and promotes Fair Trade products, including coffee, tea, and cocoa. It works to achieve a

"fair deal for farmers and workers, environmental sustainability, and profitability for all

parties in the chain of production" (TransFair USA: El). The Fair Trade certification for

coffee requires that importers purchase directly from certified cooperatives of small-scale

producers at a minimum floor price and extend credit to the producers if requested. It

encourages democratic, cooperative production, environmental protection, and long-term,

stable, trading relationships (TransFair USA). Different from most other eco-labels, it is

a "social" label, and it pertains to buyer and seller relationships in the marketing channel.

The Free Farmed label represents an animal welfare standard for meat and dairy

production. Administered by Farm Animal Services (founded by the American Humane

Association), the program certifies farms that meet guidelines pertaining to the humane

treatment of animals. Criteria include food and water access and quality, comfortable

living conditions, and animal health (Consumers Union 2003b). This label pertains to an

ethical issue that concerns a segment of consumers.

Protected Harvest is an integrated pest management (IPM) certification program

first developed for Wisconsin potatoes. Intended to reduce the amount and toxicity of

pesticides used in potato production and provide recognition of this in the marketplace,

Protected Harvest guidelines were designed by a collaborative effort of The World

Wildlife Fund, the University of Wisconsin, the Wisconsin Potato and Vegetable

Growers Association, and other organizations. Protected Harvest is now an independent

certification organization. Certification criteria are based on a point system pertaining to

pest management practices and pesticide use. Certified potatoes are sold at retail

displaying the Protected Harvest seal-of-approval (Consumers Union 2003b; Protected

Harvest; WWF).

The Food Alliance program certifies fruit, vegetable, and livestock farms with

"environmentally and socially responsible management practices" (The Food Alliance:

El). The Food Alliance is a nonprofit organization formed by "farmers, consumers,

scientists, grocers, processors, distributors, farm worker representatives and

environmentalists" (Consumers Union 2003b: El). The Food Alliance standard prohibits

the use of genetically engineered seed or livestock, and prohibits the routine use of

antibiotics and hormones. Its certification criteria measure performance in the areas of

pest management, soil and water conservation, working conditions, and wildlife habitat

conservation (The Food Alliance). Food products from certified farms may display The

Food Alliance-Approved logo.

The U.S. Department of Agriculture (USDA) National Organic Program was

developed under a mandate from the Organic Foods Production Act of 1990. A national

organic standard was published in the Federal Register on December 21, 2000. Full

implementation of organic production, handling, and marketing regulations went into

effect on October 21, 2002. After this date, use of the USDA organic seal was allowed

on products containing at least 95% organic ingredients (USDA-AMS 2003a and 2003b;

USDA-OC). According to the National Organic Standard, an organic production system

is one that integrates "cultural, biological, and mechanical practices that foster cycling of

resources, promote ecological balance, and conserve biodiversity" (USDA-AMS 2002).

The National Organic Standard contains a list of allowed and prohibited substances, land-

use and record-keeping requirements, and guidelines for soil fertility and pest

management. The use of most synthetic chemicals and conventional pesticides,

antibiotics and growth hormones, genetic engineering, sewage sludge, and irradiation is

prohibited in organic agriculture. Product handling and processing guidelines are

contained in the national standards as well (USDA-AMS 2002 and 2003a; USDA-OC).

Organic farms and handling operations that wish to label their products as organic, and

use the USDA organic seal, must be certified by USDA-accredited certifying agents

(USDA-AMS 2003c). Although use of the USDA organic seal (as well as the word

organic) is regulated by federal law, organic labeling is still voluntary, in the sense that

producers have the choice of whether to adhere to organic standards, or not; and

consumers have the choice of purchasing certified organic products, or those that are not.

A new eco-label proposed for Florida's ornamental plant nursery industry is based

on certification of environmental management practices. "The state of Florida is the

second largest producer of ornamental plants in the United States" (Hodges and Haydu,

p.1), and the ornamental plant industry is "the second largest agricultural sector in

Florida" (Larson Vasquez and Nesheim, p.1). Large amounts of pesticides and fertilizers

are used in the industry, posing contamination risks for Florida's water resources and

associated negative impacts on wildlife and human health (Leppla; Hodges, Aerts, and

Neal; Larson Vasquez and Nesheim; Stamps). Integrated pest management (IPM) and

best management practice (BMP) guidelines have been developed for the sector (Mizell

and Short; Stamps). These guidelines promote scouting and monitoring, cultural and

biological practices, and more efficient use of water and chemical inputs, to reduce

unnecessary applications and minimize the risk of negative environmental impacts.

Although a first stage in the development of this certification and labeling program

focuses on IPM for woody ornamentals, the ultimate goal is to expand the scope of

certification to include other nursery and ornamental crops and "additional criteria of

public concern, including wildlife and habitat preservation, farm worker safety, and

BMPs for nutrient and water resource management" (Leppla, p.3). In our study, these

practices are referred to collectively as environmental management practices (EMPs).

The leatherleaf fern, an ornamental plant produced in Florida, is used as an example in

chapters that follow.

Our study focused on voluntary, environmental labels on retail products that

present positive information in the form of a seal-of-approval indicating that an

independent (third-party) organization has certified a product's production as adhering to

established EMPs. The Forest Stewardship Council, bird-friendly, Protected Harvest,

Food Alliance, and USDA Organic labels, as well as the new certification and labeling

program being developed for ornamental plants, are good examples of the type of label

emphasized in our study. In particular, we refer to the EMP program (for ornamental

plants) and to the USDA organic program.

Special Characteristics of an Eco-Label

Eco-labels that are used voluntarily (to present positive information in the form of a

third-party verified seal-of-approval pertaining to the environmental impact of the

production process) are different than most other labels found on food and consumer

products. Two key characteristics make eco-labels unique.

First, an eco-label conveys information about the production of a good, but not

necessarily about the product itself. For example, Alaska salmon certified by the Marine

Stewardship Council may be indistinguishable from salmon caught in an uncertified

fishery, except that the certified salmon displays the MSC logo on the package or at the

retail counter. The physical attributes of the salmon product (such as appearance, texture,

freshness, nutritional content, and eating quality) may be essentially the same. Likewise,

there may be no physical difference between uncertified wood and wood that has been

certified by the Forest Stewardship Council. The certification and label apply only to the

environmental impact of the harvesting process, and the sustainable management of the


Second, an eco-label informs consumers about a public attribute (environmental

impact) tied to a private good. A public good is often defined as a good that is

nonexcludable and nonrival. That is, once a public good is provided to (or by) one

individual, others can enjoy the benefits of the good (because it is difficult or costly to

exclude others). Also, if one individual enjoys or appreciates the benefits or services

provided by a public good, it does not reduce the benefits or services available to others.

For example, benefits from the purchase of dolphin-safe tuna (avoiding contributing to

dolphin mortality) accrue in a marginal sense not only to the individual consumer but to

all those who are concerned about the killing of dolphin. Similarly, everyone who enjoys

birds that migrate to coffee-growing regions is affected marginally by one individual's

choice of bird-friendly vs. uncertified coffee. Whereas most product labels inform the

consumer about private attributes-those that only directly and immediately affect the

consumer of the good (nutrition, safety, horsepower, etc.)-an eco-label informs the

consumer about public attributes (environmental impacts that can affect numerous

people). Although the product itself (e.g., a bag of coffee, a piece of fish, or a grapefruit)

is a private good, the eco-label signifies a public attribute associated with the production

of the private good.

Some eco-labels signal information about both public and private attributes. For

example, organic and pesticide-free labels inform consumers about the environmental

impact of production (a public attribute), and also about the likelihood of pesticide

residues (a private attribute) on the product. The theoretical analysis presented in

subsequent chapters focuses on eco-labels whose significance to the consumer primarily

pertains to a public attribute.

Intentions behind Eco-Labeling

Various stakeholders bring assorted perspectives on the purpose behind any

eco-labeling program. For example, consumers, producers, and environmental groups,

may support eco-labeling in pursuit of different, though interrelated, objectives.

For consumers, the introduction of an eco-label provides additional information

that was not previously available or readily accessible. In some cases, an eco-label may

introduce a new claim about a product that had not previously been made. The

introduction of a meaningful eco-label allows consumers who are aware of and concerned

about the environmental impact of production to adjust their purchasing behavior in a

way that more accurately reflects their values and preferences, without incurring high

search costs. By providing more information and additional choices, an eco-label enables

consumers to obtain greater satisfaction from their purchases in the marketplace.

In other cases, an eco-label serves to verify claims that have already been made

about certain products. In this sense, an eco-label increases the credibility of an

environmental claim, especially when the certification and labeling program is supported

by a trusted third-party organization. For example, Consumers Union (2003b) has

documented numerous claims that are misleading or relatively meaningless, such as

"environmentally friendly" or "free range." A clear and meaningful eco-label, however,

provides the consumer with better information than generic claims provided by the

product's manufacturer. Consumers Union expresses the opinion that before

implementation of the national organic standards, "unregulated use of the term 'organic'

was leading to at best confusion, and at worst fraud" (Consumers Union 2001: El).

Uniform guidelines for what constitutes organic agriculture, and a standardized label

under the National Organic Program, thus served to clarify and improve trust in the

"organic" claim. For consumers, eco-labels reduce the costs of searching for information

about impacts of production, improve the credibility of environmental claims, reduce

uncertainty about product attributes, and provide additional product choices. A major

eco-labeling objective for consumers is the reduction in search costs and provision of

better, more credible information about product attributes.

Producers of goods also can benefit from eco-labeling programs. Profit-

maximization is an important objective for producers, although other objectives may also

enter their decision-making process. Eco-labels enable producers who generate

environmental benefits to differentiate their products in the market and to potentially

obtain higher prices and higher returns. In addition to the incentives created by price

premiums for eco-labeled products, producers may be motivated to adopt environmental

certification because of a reduced risk of liability for environmental damage, improved

public relations, and possibly lower production costs (Henriques and Sadorsky; Khanna

and Damon; Khanna and Anton; Leppla; WWF). Producer groups pursue eco-labeling

schemes as a means to improve profits or satisfy other objectives.

Environmental groups support eco-labels as a means to reduce the environmental

impacts associated with production, and thus improve environmental quality. When

consumers choose eco-labeled products, they reward producers who practice

environmental stewardship. Consumers who purchase eco-labeled products (often at a

premium) help keep environmentally certified producers in business and create incentives

for other producers to adopt certified practices. Thus, eco-labels have the potential to

improve environmental outcomes associated with market interactions. From the

perspective of environmental groups, and nonprofit organizations that develop

eco-labeling programs, environmental protection is the primary objective.

Research Questions and Objectives

A policy analyst would ask whether a particular policy or program effectively

meets policy objectives as defined by the stakeholders or decision-makers. Economists

often contribute to policy analysis by evaluating the economic efficiency of a policy. An

evaluation of economic efficiency might involve an analysis of costs and benefits, based

on the notion of Pareto efficiency, or an analysis of the cost-effectiveness at meeting a

particular policy objective. Economists also can contribute to policy analysis by

assessing the distributional effects or equity of a policy. In any case, an evaluation of the

likely consequences of a policy or program is an essential first step, on which to base

further evaluation in terms of efficiency or equity.

Our study focused on the objective of environmental protection and assessed the

likely consequences of eco-labeling in terms of environmental impacts under different

market conditions. An understanding of consumer and producer behavior is necessary

for analyzing market responses and ultimately the environmental impact of eco-labeling.

We presented models of consumer and producer decisions regarding environmental

certification and labeling. A two-product, partial equilibrium model and a price-

endogenous programming model were used to assess the potential effects on production,

land-use, and the environment. We extended our analysis of the efficacy of

environmental labeling to include environmental effectiveness, Pareto efficiency,

cost-effectiveness, and equity. The theoretical analysis relies on the example of an EMP

label for ornamental plants (leatherleaf ferns) to add concreteness, and a case study of the

Florida organic citrus sector complements the theoretical analysis. The rest of this

dissertation is organized as follows.

In Chapter 2, we review literature and present a model of consumer behavior in

response to eco-labels. Branches of consumer theory pertaining to product characteristics

and quality, information, and impure public goods are examined. Using a product

characteristics approach, our consumer model provides a basis for predicting consumer

demand response to eco-labeling.

In Chapter 3, we review literature and present a model of producer response to

certification and labeling programs. The literature review focused on theories of

producer choice of product quality, advertising, and technology adoption, and

summarizes empirical studies of producer decisions regarding environmental

management, certification, and labeling. Our producer model considers three separate

decisions pertaining to adoption of environmental management practices, certification,

and marketing products with an eco-label.

In Chapter 4, we introduce two theoretical models used to identify market

interaction effects and the potential impact of eco-labeling on the environment. We used

a two-product, partial-equilibrium model and a price-endogenous programming model to

address the following research questions: Under what conditions are eco-labels most

effective at achieving the objective of environmental protection? Under what conditions

are eco-labels least effective at protecting the environment? Is it possible that an

eco-labeling program could actually increase environmentally harmful production? Our

analysis provides a framework for anticipating the effects of eco-labeling on production,

land-use, and the environment under different market conditions.

In Chapter 5, we evaluate the efficacy of voluntary environmental labeling as a

potential alternative to government-mandated environmental policies. Specifically we

addressed the research question: What are the implications, in terms of environmental

effectiveness, Pareto efficiency, cost-effectiveness, and equity, of relying on voluntary

environmental labeling programs as an alternative to government-mandated

environmental policies?

In Chapter 6, we present a case study on organic citrus, based on in-person and

telephone interviews with growers and handlers in Florida. The case study provides

information on Florida organic-citrus acreage and market outlets, production volumes,

grove-care practices and costs, grower characteristics, incentives and disincentives for

organic citrus production, and research needs.

In Chapter 7, we synthesize the case study results with the preceding theoretical

analyses. We draw final conclusions and suggest avenues for further research.


An understanding of the effects of introducing an eco-label on consumer demand is

essential for analyzing the efficacy of eco-labeling. In this chapter, we review literature

and introduce a model that provides a framework for anticipating consumer demand

responses to eco-labels.

Literature Review

Although there is no well-developed theory of consumer response to labeling,

several branches of consumer theory relate to this concept. An eco-label provides

additional information to consumers, changing their perception of a product's

characteristics. Furthermore, an eco-labeled product is an impure public good. Theories

of product characteristics, information, and impure public goods are reviewed and

synthesized next as a basis for a model of consumer response to eco-labeling. We also

describe an existing consumer eco-label model found in the literature.

Many empirical studies evaluate actual consumer responses to specific

environmental labels. We reviewed several empirical studies that used a variety of

techniques to assess consumer demand for eco-labeled products.

Product Characteristics and Quality in Consumer Theory

Lancaster, Becker, Muth, and Gorman are credited with developing a theory of the

consumer based on product characteristics, although the origins of the theory can be

traced back farther. Also known as household production theory, market goods are

purchased for the characteristics that they produce (often in combination with other

market goods or time and labor inputs). Literature on the characteristics approach to

consumer theory is summarized next. Notation is altered to provide consistency.

In a classic paper, Gorman introduces a consumer model based on product quality

characteristics, specifically relating to the egg market. In his model, a consumer's utility

function is defined over the various characteristics of eggs and quantities of all other

goods. Gorman's consumer problem is

(2-1) MaxU =U ZJ; Yk

Z, = a,q,
s. t.
Ep,q, + Ek Y = I
i k

where Zj is the amount of characteristic (e.g., vitamin A) consumed; Yk are the quantities

of other goods consumed; qi is the quantity of egg type i purchased; ay is the amount of

characteristic contained in egg i; pi and Pk represent the market prices of eggs and all

other goods; and I is the consumer's money income. Important assumptions in his model

are that characteristics are measurable quantities, and that the total amount of a

characteristic consumed is the sum of the amounts contained in the eggs. Gorman's

model represents inputs (goods) that have constant marginal rates of technical

substitution in the production of outputs (characteristics), and considers joint products

(multiple characteristics) from a single input.

The necessary first-order condition for utility maximization in the Gorman model is

aU au
p. =\i -a,, +a -+
(2-2) a a Z2
p, =a,,irr +a,2)r2 +...

where X is the marginal utility of income, and tj is the imputed price for characteristic j.

The market price for egg type i is the sum of the imputed prices for characteristics times

the input-output coefficients.

Muth introduces a similar model, in which "[goods] purchased on the market by

consumers are inputs into the production of [characteristics] within the household"

(p.699). The consumer problem according to Muth's model is

(2-3) MaxU= U(Z,,...,Z,)

s.t. Z, = Zj(q ,,...,q,,)
p1q, +...+ p,,q,, =I

He demonstrates that the marginal rate of substitution (MRS) between two market goods

used to produce the same characteristic reduces to

(2-4) Zja
aZ,/ /q

The MRS between the two market goods equals the marginal rate of technical

substitution (MRTS) in production of characteristic (Muth). If the production function

were defined as in the Gorman model, the MRS would be the constant ratio of input-

output coefficients, ai/a2j. Muth notes that the MRS between two market goods used to

produce the same characteristic "is functionally independent of all [goods] not used in

the production of characteristicsj] by the household" (p.701). He does not consider the

case of joint products, in which one market good produces multiple characteristics.

Lancaster (1966a, 1966b, 1991) developed a detailed linear characteristics model.

His basic consumer model is similar to Gorman's. In Lancaster's model, a consumer

SIn Gorman's paper, X is incorrectly placed in the numerator on the right-hand side of the equation.

maximizes utility, which is defined over product characteristics, subject to a budget

constraint defined in goods-space and a consumption technology constraint that

transforms the market goods into characteristics-space. Lancaster specifies the consumer

problem as

(2-5) MaxU(z)

pq<_ I
z = Bq

where z and q are vectors of characteristics and market goods respectively; and B is a

matrix of input-output coefficients transforming goods space to characteristics space.

According to Lancaster (1966a, p.139)

A consumer's complete choice subject to a budget constraint...can be considered as
consisting of two parts:
a) An efficiency choice, determining the characteristics frontier and the associated
efficient goods collections.
b) A private choice, determining which point on the characteristics frontier is
preferred by him.

An important assumption in Lancaster's approach is that the consumption technology

(defined by the input-output coefficients in the linear model) is objectively determined,

independent of the consumer's perception or preferences. That is, every consumer faces

the same characteristics frontier. A change in product quality, or introducing a new good

with a different mix of characteristics, affects the shape of the characteristics frontier

(Lancaster 1966a). This induces a change in the efficient bundle of goods without

changing consumer preferences.

The change in the slope of the frontier is analogous to the change in the budget line
slope in the traditional case and, with a convex preference function, will result in a
substitution of one characteristics bundle for another and, hence, of one goods
bundle for another (Lancaster 1966a, p.143).

Lancaster asserts that his model allows one to predict the result of introducing a new

product or a differentiated product better than the traditional consumer theory does.

Hendler provides a critique of Lancaster's approach. He faults Lancaster for

ignoring the possibility of characteristics with negative utility. Also, Hendler (p.197)

points out two restrictive assumptions in Lancaster's model: a) "utility independence of

characteristics per consumption unit", and b) "the mixability of goods per consumption

unit." Hendler argues that it is unrealistic to compare the utility of characteristics

obtained from a linear combination of two goods with the characteristics contained in a

single good.

Michael and Becker pose a household production function approach similar to the

one described by Muth. Their production function is Zj = z/(qj, tj; E), where q4 represents

market goods, t is a household's time, and "E is a vector of variables which represents the

environment in which the production takes place" (p. 382). According to Michael and

Becker, the necessary first-order conditions for maximization of utility defined over the

characteristics Zj imply that marginal rates of substitution between characteristics are

(2-6) U, w(t/iz,)+ p,(aq,/8z,) ,
(2-6) = =
Uz2 w(atl z2)+ P,( q,/az2) ;2

where Uzj is the marginal utility with respect to characteristic Zj, w is the wage rate (value

of household's time), and nl and r2 are the implicit "shadow" prices of characteristics 1

and 2. We note that Michael and Becker's first-order condition (Eq. 2-6) is correct only

for the case of fixed-proportions production, but is incorrect for a generic variable-

proportions production function. They define the MRS between market goods as

U UIME, p,
(2-7) Uq UIMPI P
Uq2 UZ2MP22 P2

where MP is marginal product. We note that Eq. 2-7 is correct for a variable-proportions

production function, but does not apply to the fixed-proportions case. Michael and

Becker point out that if both inputs are used to produce the same final characteristic, then

the condition (Eq. 2-7) is simply that the ratio of marginal products equals the (input)

price ratio. This result is similar to the one obtained by Muth. Michael and Becker note

that the case of joint products (one market good producing multiple characteristics)

implies that the marginal value of a good is

(2-8) U:,MPI, =p,

"In general, the price of any [characteristic] is then affected by the level of output of the

other [characteristics] which use [the same input]" (Michael and Becker, p. 383).

Hedonic price theory distinguishes a product and its market value by the product's

characteristics. Rosen (p. 34) defines hedonic prices as "implicit prices of

attributes...revealed to economic agents from observed prices of differentiated products

and the specific amounts of characteristics associated with them." He describes a

consumer's value or bid function for product attributes as "the expenditure a consumer is

willing to pay for alternative [levels of characteristics] at a given utility index and

income" (Rosen, p. 38).

In Rosen's expanded model, the consumer chooses the quantity of a good to

purchase and chooses the level of characteristics associated with the good. The

consumer's problem is to maximize a utility function, U(y, zi,..., z,, q), subject to the

budget constraint, I = y + qp(z), where q is the quantity of the good, zl,..., z, is the bundle

of characteristics associated with the good, y is all other goods consumed, I is income,

andp(z) is the good's price as a function of the characteristics vector. In Rosen's model,

necessary first-order conditions imply that

(2-9) = p(z)



Equation 2-9 is the familiar condition that marginal rate of substitution should equal the

price ratio. Equation 2-10 demonstrates that the level of characteristic (or product

quality) should be chosen so that the marginal rate of substitution for quality with respect

to income (a consumer's value function for the characteristic) equals the rate at which

price increases with quality, pi, times the optimal quantity of the good. Rosen

acknowledges that this model restricts the consumer to purchasing only one type of good.

Ladd and Suvannunt present a consumer model in which utility is defined over

characteristics (Zj) embodied in multiple goods and characteristics (Xi) unique to one

good only. Similar to the results from Gorman and Michael and Becker, the necessary

first-order conditions of the Ladd and Suvannunt model are

(2-11) = azaU/a ax, au/aX, 1
SZaq, U/aI j I q, [ aU/9I

where X, is the characteristic unique to product i. Noting that income equals expenditure,

the authors convert the first-order conditions to

(2-12) p,= a E a- ,
aq, aZ, ax,

where E is total expenditure, and 8E/8Z, is "the (marginal) implicit or imputed price

paid for the jth product characteristic" (Ladd and Suvannunt, p. 505). They conclude that

for each product consumed, the price paid by the consumer equals the sum of the
marginal monetary values of the product's characteristics; the marginal monetary
value of each characteristic equals the quantity of the characteristic obtained from
the marginal unit of the product consumed multiplied by the marginal implicit price
of the characteristic (Ladd and Suvannunt, p. 504).

Equation 2-12 is very similar to Eq. 2-2 from Gorman and Eq. 2-8 from Michael and

Becker. Ladd and Suvannunt test the hypotheses of their model empirically using

nutritional elements of food items and obtain significant results.

Stigler and Becker (p. 76) argue "that tastes neither change capriciously nor differ

importantly between people." The authors advocate a characteristics approach to

consumer theory and explain the effect of advertising or social influences as changing

household production constraints not as changing consumers' preferences. They present

a consumer model in which utility is defined over characteristics Zj and the household

production function is Z =f(q,A,E, V), where q is the quantity of a market good, A is the

advertising for the market good, E is a human capital variable affecting production, and V

is a vector of other relevant variables. If the characteristic output, Z, is assumed

proportional to the amount of market good consumed, q, for any given level of A, E, and

V, Stigler and Becker's production function becomes

(2-13) Z =g(A,E,V)q.

This relationship is similar to the second constraint in Eq. 2-5 and the linear

characteristics models of Lancaster and Gorman, except that the coefficient g() is not the

same for every household. In Stigler and Becker's model, the coefficient go varies

according to household characteristics and advertising exposure. In particular,

8g/aA > 0. Thus, the relationship between shadow prices of characteristics and market

prices of goods, as in Eqs. 2-2, 2-6, 2-8, and 2-12, for a single market good and a single

characteristic is

(2-14) ,r, =

For a given market price p, an increase in g will lower the characteristics shadow price rz.

According to Stigler and Becker, an increase in g allows a household to obtain more of

the characteristic from a unit of market good, implying that a unit of characteristic

becomes less expensive. Stigler and Becker (p. 84) describe the relationship between

advertising and the demand for characteristics and market goods as follows:

An increase in advertising may lower the [characteristic] price to the household (by
raising g), and thereby increase its demand for the [characteristic] and change its
demand for the firm's output, because the household is made to believe-correctly
or incorrectly-that it gets a greater output of the [characteristic] from a given input
of the advertised product. Consequently, advertising affects consumption in this
formulation not by changing tastes, but by changing prices. That is, a movement
along a stable demand curve for commodities is seen as generating the apparently
unstable demand curves of market goods and other inputs.

Whereas Lancaster emphasizes that the consumption technology is objectively

measurable and the same for all consumers, Stigler and Becker relax this assumption.

They allow the coefficient, g, to vary with household attributes and perceptions

influenced by advertising.

Deaton and Muellbauer acknowledge some benefits associated with the Stigler and

Becker approach. They question the usefulness, however, when characteristics are

qualitative in nature (i.e., not easily observable or measurable). Deaton and Muellbauer

(p. 244) state

Our only qualm is that, when the intervening variables are not observable, there
may be little cutting edge to the distinction between preferences and constraints,

and the 'explanations' offered by the approach can sometimes be complicated ways
of making rather simple points.

Deaton and Muellbauer point out other shortcomings of the linear characteristics models,

such as the problem of corer solutions, in which a consumer's optimal bundle is

obtained from fewer goods than there are characteristics. In this case, the marginal rate

of substitution between characteristics does not equal the shadow price ratio(Deaton and

Muellbauer). The authors suggest a discrete choice formulation of the consumer problem

to deal with corer solutions. "When households choose between two discrete

alternatives, the relevant measure of demand to be empirically explained is the

probability of choosing one or the other" (Deaton and Muellbauer, p. 267).

Tirole explains the consumer's problem in a vertically differentiated market as a

discrete choice of whether to purchase or not at a given quality level. A taste parameter 0

describes a consumer's willingness to pay for a quality level z, such that the consumer

will purchase a good only if Oz > 0. A distribution function can be estimated for

consumers in a market, such that "F(O) is the fraction of consumers with a taste parameter

of less than 0" (Tirole, p.97). The taste parameter, 6, can be interpreted as the

individual's marginal rate of substitution between the quality characteristic and income,

UZ /aU, The individual consumer will purchase a market good if 0 > p/g, where p is

the market price and g represents the quality index. Tirole's model is similar to Eq. 2-14

and Stigler and Becker's description ofg as the amount of a characteristic contained in

the market good. The distribution function, F(O) can be converted to a demand function

of the form

(2-15) D(p)= N[1- F(p/g),

where D(p) is quantity demanded at a given quality level, and N is the number of

consumers in the market (Tirole). This formulation assumes unit demands, but could be

adjusted to account for differences in quantity demanded among consumers. It is clear

from Tirole's formulation that an increase in the quality index g will increase the quantity

demanded at any given price. While this approach is appropriate for modeling the

discrete choice of buying one market good, it does not provide insight as to the

relationship between demands for various market goods.

Information and Uncertainty in Consumer Decisions

Lancaster (1966b) suggests that consumers may not act efficiently, if they are

unaware that a particular good possesses certain characteristics or if they are uncertain

about these characteristics. In Lancaster's model, the efficient characteristics frontier is

the same for all consumers. This idea provides the basis for his "argument in favor of

public information on these matters and in favor of legal requirements, such as

composition and contents labeling, designed to increase knowledge of the available

consumption technology" (Lancaster 1966b, p. 18-19).

Akerlof demonstrates that quality uncertainty or asymmetrical information can

undermine markets for high quality products. Asymmetrical information occurs when

one party to a transaction is less well informed than the other party. In Akerlof's

example of the market for used cars, sellers have much more information about the

quality of the automobile than do potential buyers. Akerlof's hypothesis is summarized

as follows. If buyers cannot tell the difference between a good car and a "lemon" prior to

purchase, they would only be willing to pay a value reflecting their estimate of the

average quality of cars on the market. It is not likely, however, that owners of high

quality cars would be willing to sell at the average quality price. Thus, only low quality

cars would be sold at the going price. If buyers come to realize this, they may adjust their

estimate of the average quality of cars on the market and their willingness to pay

downward once again, further reducing the incentive for high quality cars to be sold. In

short, if quality cannot be credibly conveyed, consumers will not pay extra for uncertain

quality, and there will be no incentive for producers to supply quality products.

Product attributes may be classified according to how easily the attributes can be

verified by consumers. According to Nelson, search attributes are characteristics of a

product that consumers can easily identify before purchase, whereas experience attributes

can only be verified by consumers after they use the product. Darby and Karni describe a

third category, called credence attributes, which are unobservable to consumers even

after purchase and consumption. Credence attributes are especially vulnerable to

problems of asymmetric information and adverse selection, which erode incentives to

supply these attributes.

The environmental impact of a good's production is information that cannot be

verified easily by most consumers even after consumption. The environmental

characteristics embodied in a product are therefore considered credence attributes.

According to McCluskey, third-party monitoring is necessary for most producers to have

incentive to provide high-quality credence goods, such as environmentally friendly food

products. Using laboratory markets for products making environmental claims, Cason

and Gangadharan (p. 129) find that "unverified claims are not sufficient to improve

market outcomes." The results of their experiment suggest that certification by an

independent organization is essential for the market to improve the environmental quality

of goods. Teisl, Roe, and Levy test the response of survey participants when presented

with different types of eco-labels. Respondents gave significantly different credibility

scores depending on the label format, detail, and certifying organization (Teisl, Roe, and


Consumer choice under uncertainty also can be modeled according to the concept

of expected utility (Nicholson). If subjective probabilities are assigned to uncertain

outcomes 1 and 2, then expected utility is the probability of outcome 1 times its utility

plus the probability of outcome 2 times its utility. The expected utility approach can be

applied to consumer choice when product quality is uncertain. Consider a consumer who

obtains known utility, Uo, from a product without an environmental attribute, and utility,

U1, from a product with an environmental attribute. Let the consumer assign probability,

p, that an environmental claim is true, and probability 1-p that it is false. The consumer's

expected utility ispUI + (1-p)Uo. Consumers will make choices so as to maximize their

expected utility if they "obey the von Neumann-Morgenstem axioms of behavior in

uncertain situations" (Nicholson, p. 218).

By lowering search costs, eco-labels serve to reduce information asymmetry and

consumer uncertainty about credence attributes associated with the environmental impact

of the production process. Some eco-labels are more effective than others at boosting

consumer confidence in the credibility of environmental claims. The introduction of an

eco-label on an existing product can be modeled as an increase in the probability

consumers place on the environmental quality claim being true. Assuming linear utility,

the effect is the same as increasing g in the Stigler and Becker, and Tirole models,

essentially increasing the expected amount of the environmental characteristic provided

by the market good.

Theory of Impure Public Goods

An eco-labeled product is similar to an impure public good (i.e., a market good that

contains both private and public attributes). For example, the consumer of bird-friendly

coffee receives a private benefit, the cup of coffee, and also the satisfaction of

contributing to a public benefit, preserving bird habitat. Alternatively the unlabeled

product could be considered an impure public bad (i.e., a market good that contains a

private benefit, but reduces the welfare of others). For example, if conventional

agricultural production contributes to the contamination of local water supplies (through

fertilizer and pesticide runoff), then its products contain both a private benefit for the

consumer (food) and a public bad associated with harmful impact of production. An

EMP label for Florida leatherleaf ferns, however, informs consumers that production

followed management practices that protect water quality, avoiding the public bad

associated with uncertified production.

Comes and Sandler (1984, 1994, 1996) model consumer behavior in relation to

impure public goods and bads. In their basic model, a consumer chooses quantities of a

numeraire good and a market good that generates a private and public characteristic. The

total quantity of the public characteristic enjoyed by the consumer depends also on the

actions of others. The Comes and Sandler consumer problem is

(2-16) MaxU =U(Y,,YY3)

Y + pq = I
s.t. Y2= q

Y3 = Y3 + +

where YI is the numeraire characteristic representing all other goods; Y2 is the private

characteristic generated from consumption of the market good q; and Y3 is the total level

of the public good enjoyed by the consumer (Comes and Sandier 1994, p.406; 1996,

p.256). The characteristic Y2 is a function of the quantity of market good q, and Y3 is a

function of the level of public good provided by others, Y30, and the individual's

contribution through purchases of market good q.

Necessary first-order conditions for utility maximization in the Comes and Sandier

consumer model imply that

U2 U3
(2-17) U, U,
p = pT + y3

where Ui is marginal utility, and rE; is the shadow value of the characteristic to the

individual. This condition is similar to Eqs. 2-2, 2-8, and 2-12. Taking the contribution

of others as given and acting only in self-interest, the individual consumer does not

consider the public good value to others resulting from private consumption choices.

The Comes and Sandler model of impure public goods is not entirely suitable for

the case of voluntary eco-labeling, because only one good provides the joint public and

private characteristics. With voluntary eco-labeling, consumers typically have the choice

of purchasing two or more similar products, each providing the same private and public

attributes to different degrees.

Prior Model of Eco-Labeling

Van Ravenswaay and Blend formulate a consumer model of the choice between an

unlabeled and eco-labeled product. Their basic consumer model is

(2-18) MaxU(X,X',Q(X,X',E)

s.t. PX + P'X'= M

where "X is the quantity of goods purchased, Q is environmental quality, E is an

exogenous amount of environmental damage, P is the price ofX, and Mis income....Let

X' be the quantity purchased of the ecolabeled version of X' (Van Ravenswaay and

Blend, p. 124). The authors identify the condition for a consumer to choose the unlabeled

product X over the unlabeled product X' as

(2-19) P'-P >U/aQ[aQ/X-Q/aX]

Equation 2-19 demonstrates that a consumer will not purchase an eco-labeled product if

the price premium exceeds the marginal rate of substitution between environmental

quality and income times the difference in environmental impact between the two


In another version of the model, van Ravenswaay and Blend introduce a probability

variable representing consumer trust in an eco-label claim. The model is formulated as

(2-20) MaxU(X,X',(Q(X,X',E)Prob(Q)))

s.t. PX + P'X'= M

The authors do not explore consumer demand response to eco-labeling or impacts on

demand for X or X' in either model.

Empirical Research

Several studies evaluate potential consumer demand and willingness to pay

premiums for specific eco-labeled items. Empirical research on consumer response to

eco-labels utilizes a variety of techniques. These include contingent valuation, conjoint

analysis, hedonic price analysis, and demand system analysis.

The contingent valuation method (CVM), often used to value non-market

environmental amenities, can be tailored to assess willingness to pay for a new product or

for a labeled product. In an open-ended format, CVM asks the survey respondent "what

is the maximum amount that you would be willing to pay...." (Van Kooten and Bulte, p.

123). The dichotomous choice format presents respondents with a yes or no alternative,

such as "would you be willing to pay $A...." (Van Kooten and Bulte, p.124). Either

format can be used to elicit consumer valuation of label attributes and their willingness to

pay a price premium for a good with a certain label. In the spike model described by

Kristrim, consumers first are asked whether they would be willing to pay any amount for

an environmental improvement (or any premium for a label attribute). Then the valuation

function is elicited from those consumers willing to pay a positive amount.

Ozanne and Vlosky, Gr6nroos and Bowyer, and Jensen and Jakus conducted

contingent valuation studies for environmentally certified wood products. Their results

indicate that anywhere from 24% to 71% of consumers are willing to pay a non-zero

price premium, depending on the study and specific product. Ozanne and Vlosky found

that the average consumer would pay an 18.7% price premium for an environmentally

certified 2"x4"x8' wood stud with a base price of $1, but only a 4.4% price premium for

a new home ($100,000 base price) built with environmentally certified wood. In their

study, Gr6nroos and Bowyer concluded that consumers willing to pay extra for a home

built with environmentally certified material would, on average, pay a 1 to 2% premium.

Jensen and Jakus find a mean willingness to pay of 12.9% extra for a certified oak

shelving board, 8.5% extra for a certified oak chair, and 2.8% extra for a certified oak


Contingent valuation is used to evaluate consumer preferences for eco-labeled fish

in at least two studies. In a survey of U.S. consumers, Wessels, Johnston, and Donath

found that price, species, perception of fish stocks, and various consumer characteristics

had statistically significant effects on respondents' choice of certified vs. uncertified

seafood. Consumer willingness to pay a premium varied by species, with consumers

more sensitive to increased premiums for cod than for salmon (Wessels, Johnston, and

Donath). Johnston, et al. conducted a similar study of consumers in the U.S. and

Norway. The authors found that price, species (cod or shrimp), certifying agency,

membership in environmental organizations (U.S. only), fresh vs. frozen products (U.S.

only), gender (Norway only), income (Norway only), and country of residence all had

significant effects on the likelihood of purchasing certified vs. uncertified seafood.

Norwegian respondents were more price sensitive.

Loureiro, McCluskey, and Mittelhammer used a contingent choice method to

compare consumer preferences for organic, eco-labeled, and regular apples. The authors

estimated probabilities of choosing one type of apple over the others depending on

various consumer characteristics. They conclude that organic apples are perceived to

have higher levels of positive attributes that affect consumer choice, with eco-labeled

apples an intermediate choice. Eco-labeled apples were also the subject of a study by

Blend and van Ravenswaay. The authors surveyed households with a contingent choice

approach. The percentage of respondents choosing eco-labeled apples was 72.6% with

no premium, 52.4% with a $0.20 premium, and 42.3% with a $0.40 premium over the

base price of unlabeled apples (Blend and van Ravenswaay).

Hays used a contingent valuation approach with a discrete utility based framework

to determine "willingness to pay for goods labeled with safe drinking water guarantees"

(p.47). Among the 53% of consumers willing to pay extra for goods produced so as to

protect groundwater quality, 8.9% was the average maximum acceptable premium


Conjoint analysis is another method that can be used to elicit consumer preferences

and anticipate consumer responses to eco-labeled products. Conjoint analysis asks

respondents to rate or rank products with different combinations or levels of attributes.

The relative importance respondents place on price vs. levels of other attributes provides

an estimate of the implicit prices of attributes (Roe, Boyle, and Teisl). In conjoint

studies, different types of labels can be used to represent attributes and assess consumer

valuation of those attributes.

Several studies use conjoint analysis to assess consumer valuation of product

attributes similar to those represented by eco-labels. Sylvia and Larkin surveyed seafood

wholesalers to assess the relative value they placed on various characteristics of Pacific

whiting. The authors estimated the change in "conditional short-run, firm-level

demands" (p. 503) for fillets with improved characteristics. Holland and Wessells asked

consumers to rank salmon products according to the label, which presented information

on type of production (farmed or wild-caught), safety inspection, and price. They found

that consumers valued the safety inspection attribute most strongly, and that, on average,

respondents favored farmed salmon over wild-caught. Baker and Burnham used conjoint

analysis to assess consumer preferences and valuation of a genetically modified (GM)

attribute in relation to other attributes of corn flakes. About 30% of respondents based

their rankings on the GM content of the corn flakes (Baker and Burnham).

Whereas contingent valuation and conjoint analysis are used primarily for new

products, for which insufficient market data is available, hedonic price analysis and

demand system analysis rely on historical market data to estimate the effects of various

characteristics. Hedonic price estimates depend on both consumer valuations of

attributes and producer costs of supplying the attributes, but "estimated hedonic price-

characteristics functions typically identify neither demand nor supply" (Rosen, p.54).

Nimon and Beghin estimated hedonic prices for organic and "no-dye" attributes of

cotton apparel items. The authors found an average price premium of 33.8% for the

organic cotton attribute. A 14.8% discount was associated with the no-dye attribute,

reflecting in part the lower costs of production (Nimon and Beghin).

Teisl, Roe, and Hicks used demand system analysis to assess whether dolphin-safe

labels altered consumer behavior and increased the market share of canned tuna between

April 1988 and December 1995. The authors estimated a system of share equations with

an AIDS model and iterated seemingly unrelated regression using scanner data for

canned tuna and substitute meat products (other canned seafood, canned red meat, and

luncheon meats). They used monthly time series data and incorporated a media index.

Although the market share for canned tuna declined during this time period, they found

that the implementation of dolphin-safe labeling had a significantly positive effect,

causing the market share to decline by less than if the labeling program had not been

implemented. Their analysis considers the effect of the entire U.S. market converting to

dolphin-safe labeling at the same time, so consumers did not have the choice between

labeled and unlabeled tuna. The authors used compensating variation (CV) to measure a

partial welfare effect of the dolphin-safe labeling. They estimated national annual CV to

be between $6 million and $15 million.

Thompson and Glaser estimated own- and cross-price elasticities of demand for

organic and conventional baby food using a quadratic almost ideal demand system

(QUAIDS) model. The authors obtained monthly scanner data for the period between

April 1988 and December 1999 from A.C. Neilsen and Information Resources, Inc. They

note that organic baby food had a market share in 1999 of between 2.5% and 13%,

depending on the specific type of product. In their analysis, organic and conventional

baby food products were considered for five product categories. According to Thompson

and Glaser's results, own-price demand for organic baby food is highly elastic, although

these elasticities have declined over time as premiums have fallen and market shares have

risen. Their analysis shows that own-price demand for conventional baby food is

inelastic and that cross-price elasticity of demand for organic baby food is higher than

that for conventional (i.e., changes in the price of conventional baby food have a stronger

effect on demand for organic than changes in price for organic baby food have on

conventional). As expected, most of these cross-price elasticities are positive, indicating

that organic and conventional baby foods are substitutes.

Bjorner, Hansen, & Russell used a mixed logit model with panel data on Danish

households to estimate the impact of the Nordic Swan environmental label on choice of

toilet paper, paper towels, and detergents between 1997 and 2001. Controlling for brand

influences, advertising, and other variables, they found that the Nordic Swan label

increased consumer willingness to pay for toilet paper by 13 to 18 percent. The effect of

the label on consumer choice of paper towels and detergent was less significant.

Various general conclusions can be drawn from these studies. A considerable

segment of the population is willing to pay premiums for eco-labeled attributes, although

hypothetical estimates of this willingness may be inflated. Willingness to purchase an

eco-labeled product increases as the number of (positive) attributes embodied in the label

increases (e.g. organic vs. reduced pesticide or eco-labeled), but falls as the price

premium rises. When expressed in absolute terms, the acceptable premium level rises as

the price of the conventional product increases. However when expressed as a

percentage of the conventional price, the acceptable premium for an eco-label is lower for

more expensive products.

Consumer choice between an eco-labeled and unlabeled product is quite sensitive

to variations in perceived product "quality." It seems evident that most consumers are

more willing to pay for private attributes, such as product quality and food safety, than

for public attributes embodied in a product, such as diffuse environmental impacts. The

food safety attribute (e.g. pesticide exposure) seems to be of greater importance in

purchasing baby food than most other products, as evident by the significant market share

for organic baby food.

Model of Consumer Response to Eco-Labels

We formulate a consumer model for the choice of an eco-labeled good, its

unlabeled counterpart, and all other goods. Our model draws on elements of the product

characteristics approach, imperfect information, and the impure public good model. In

particular, it is similar to the model used by Comes and Sandler for an impure public

good. By adding the option to choose between an eco-labeled and unlabeled version of

the same commodity, however, this model differs in an important sense. The model also

has similarities to the consumer models presented by Stigler and Becker, and van

Ravenswaay and Blend.

Although our model is intended for generic application to a variety of eco-labels,

we use the example of an eco-label intended to distinguish leatherleaf ferns (an important

crop in Florida's ornamental plant industry) that have been produced using environmental

management practices (EMPs) that conserve water, reduce chemical use, and protect

regional water quality in Florida. Stamps details irrigation and nutrient management

practices for commercial leatherleaf fern production and asserts that such practices help

protect ground water quality. An eco-label representing certified adherence to these

EMPs would signal a public attribute (water quality protection) that affects residents and

visitors to the producing region. For simplicity, assume that ferns may be produced

either conventionally (without adherence to EMPs) or using "green" practices (EMPs that

protect regional water resources). Ferns produced by EMP-certified nurseries may be

sold with or without an eco-label. Ferns produced by uncertified nurseries must be sold

without an eco-label.

The consumer is assumed to possess a utility function that depends on the levels of

three characteristics: a private characteristic associated with fern consumption, a private

characteristic associated with the numeraire commodity representing all other goods, and

a public (bad) characteristic measured by a water quality index representing the level of

contamination of water bodies in Florida. It is assumed that one unit of the numeraire

good generates one unit of the numeraire characteristic. Also, the private characteristic

from fern consumption is satisfied equally well by either unlabeled ferns or eco-labeled

ferns, in a one-to-one relationship. Finally, we assume that one unit of conventional fern

production (and consumption) reduces the water quality index by one unit, whereas

"green" fern production has a negligible impact on water quality.

The consumer chooses quantities of the three market goods, so as to maximize

utility subject to an income constraint and technical constraints that relate the market

goods to characteristics valued by the consumer. The consumer's optimization problem


(2-21) MaxU(Y,X,B)
y,q. ,q1

y + p,q, + pe =I
S. t.
X =q, + q,
B BO + q +(1- p)q

where y, qu, and qe are quantities of the numeraire good, unlabeled ferns, and ferns

labeled with an environmental claim respectively. The variables Y, X, and B are

quantities of the private numeraire characteristic, the private characteristic associated

with ferns, and the public bad respectively. A higher level of B represents greater water

contamination (measured by a water quality index). Bo is the level of water

contamination generated by others (through the production and consumption of ferns not

adhering to EMPs) and is considered exogenous to the individual consumer's decision.

The exogenous parameters I, pu, and pe represent the consumer's disposable income, the

price of unlabeled ferns and the price of environmentally-labeled ferns respectively.

Finally, the parameter, p, represents the consumer's trust in the environmental claim.

Thus, the coefficient (1-p) has a subjective element based on consumer perceptions,

similar to g() in the Stigler and Becker model.

Whereas a small segment of consumers may have been taking the producer's word

or relying on a first-party (producer) environmental claim prior to eco-labeling, a third-

party verified seal-of-approval represents an improvement in information. In particular,

the credibility of the environmental claim increases. Consumers would attribute a higher

probability that an environmental claim is true. Thus, the introduction of an eco-label is

represented by an increase in the value of p on a 0-1 scale. Ifp = 0, an environmental

claim has no credibility. Ifp = 1, an environmental claim has perfect credibility. The

eco-label design, trustworthiness of the certifying organization, and strength of

monitoring and verification activities may affect the credibility of an eco-label.

Assuming a positive quantity of Y (all other goods) is always purchased, necessary

first-order conditions for utility maximization are

(2-22) Ux U p


(2-23) Ux +(1 -p)L Ur UY

where Uj is marginal utility with respect to characteristic. Since Y and Xare positive

characteristics, Ur > 0 and Ux > 0. Because B is a negative characteristic, UB < 0.

Equation 2-22 holds with equality if q, > 0, and Eq. 2-23 holds with equality if qe > 0.

The two first-order conditions expressed in Eqs. 2-22 and 2-23 are similar to Eqs. 2-2, 2-

8, and 2-11 from Gorman, Michael and Becker, and Ladd and Suvannunt. If positive

quantities of a market good are purchased, the market price equals the sum across all

jointly produced characteristics of the marginal implicit characteristic values times the

coefficients measuring the amount of a characteristic embodied in the market good.

Equation 2-22 implies that the marginal rate of substitution between X and income

(MRSxr) plus the marginal rate of substitution between B and income (MRSBr), which is

negative, equals the market price for consumers that purchase unlabeled ferns. If eco-

labeled ferns are purchased, Equation 2-23 implies that MRSxy + (1-p)MRSBy equals

market price, where (1-p) is the expected contribution ofqe to the public bad, B.

For consumers who purchase both unlabeled and eco-labeled ferns, or who are

indifferent between the two, Eqs. 2-22 and 2-23 together imply

(2-24) PU = Pr

where p, is the price premium (i.e., pe -Pu). Equation 2-24 suggests that the price

premium equals the marginal consumer's valuation of water quality (-UB/UY) times the

perceived difference in impact on water quality between one unit of the unlabeled and

one unit of the eco-labeled ferns (p). If p = 1, the price premium is the marginal

consumer's valuation of avoiding the water contamination resulting from conventional

production of one unit of ferns. This interpretation applies to the consumer who chooses

to purchase a positive quantity of ferns so that Ux/Uy = pe and is marginal in the sense

that -pUs/Uy = p,. Other (intramarginal) consumers may maximize utility by purchasing

a positive quantity of eco-labeled ferns so that Ux/Uy = pe, but have a water quality

valuation that exceeds the price premium (i.e., -pUs/Uy > p,).

Corer solutions imply that the conditions represented by Eqs. 2-22, 2-23, and 2-24

do not all hold with equality. Considering the possibility of corer solutions, four types

of consumers are distinguished with respect to our consumer model. The first type of

consumer (Type 1) does not buy any ferns. The second type of consumer (Type 2)

purchases unlabeled ferns, but no eco-labeled ferns. The third type of consumer (Type 3)

purchases eco-labeled ferns, but no unlabeled ferns. The last type of consumer (Type 4)

buys both unlabeled and eco-labeled ferns. The necessary relationship between marginal

characteristic values and market price for each consumer type is shown in Table 2-1.

Table 2-1. First-order conditions for four different consumer types
Type 1 Type 2 Type 3 Type 4
qu = 0, qe = 0 qu > 0, qe = 0 q, = 0, qe > 0 q, > 0, qe > 0
Ux +U< Ux +UB Ux +UP Ux +U,

U <" Ur< P= P= PUU,
Ux +(-p AU Ux +(1-p A Ux +(-pU Ux +(-p)U P,
U, U, U, U,
_U, UY Y

Demand for each type of fern is a function of the prices of both types of ferns, a

price index representing the numeraire, consumer disposable income, the parameter p,

and consumer preferences (which typically are assumed to be stable). Demand for the

unlabeled ferns can be specified as D,(p, pe, py, I, p). Demand for the ferns labeled with

an environmental claim can be formulated as De(pu, p, I, p). Additional parameters

could be added representing characteristics of the consumer population, such as

environmental awareness or other factors affecting the perceived relationship between

fern production, q, and water quality, B.

Joint production, the presence of multiple goods producing a single characteristic,

and the existence of corner solutions cause difficulties for assessing the comparative

static properties of demand from this model. Without restrictive assumptions,

comparative static results for the effect of a change in the parameter p are ambiguous for

X, Y, q,, and qe. If improved water quality is not an inferior good, and given well-

behaved utility functions, the effect on B' of increasing p is negative, where B' is the

individual consumer's contribution to the public bad, B.

A graph of the constraint set in characteristics-space, shown in Figure 2-1, provides

insight regarding the effects of eco-labeling, represented by an increase in the parameter

p. Taking Bo as given and setting it at the origin, the difference, B Bo = Bi, is the

individual's contribution to the public bad. The constraint set in characteristics space is

defined by the equations:

(2-25) Y + P + p, iX ( B' = I


(2-26) B' = aX + (1 -a)X(1- p),

where Eq. 2-26 restricts the relationship between B' and Xto be defined by a convex

combination of q, and qe.



B x

Figure 2-1. Constraint set in 3-dimensional Y-X-B space

In Figure 2-1, the consumer's constraint set is bounded by the pyramid ceBOYmx.

The relevant surface for utility maximization is the surface ceYmx. An increase in the

parameter, p, rotates the qe vector toward the X-axis, generating less B' per unit ofX. The

corer point e moves toward the point and the base of the pyramid bounded by ce shifts

out toward cf A change in p is equivalent to a price change in characteristics space.


q,' qe c

f e

Bo B

Figure 2-2. Constraint set in 2-dimensional X-B space

Figure 2-2 shows the constraint and indifference curves in two dimensional X-B

space. Because B is a negative characteristic, the indifference curves, represented by the

dashed lines, are positively sloped. The slope of the constraint line ce is

(2-27) Pr/P
(Pr/P)+ P.

As p approaches 1, the vector qe rotates toward qe', and the slope of ce falls (as it rotates

toward cj). A substitution effect implies lower levels of B' (which is desirable) and lower

levels of X. An income effect allows the consumer to obtain higher levels ofX for a

given level of B'. Assuming that B is a normal "bad" (i.e., water quality is not an inferior

good) the income effect further reduces the optimal level of B for the individual

consumer. For fixed market prices, pu and pe, an increase in the parameter p, representing

the credibility an eco-label lends to an environmental claim, has a non-positive effect on

B'. The effect on ordinary (uncompensated) demand for X and Yis ambiguous.

Intuitively, one would expect an eco-label to increase demand for the good making

the environmental claim and decrease demand for the unlabeled good. Indeed, for

consumer Types 1 and 2 (representing corner solutions, in which qe = 0 prior to labeling)

the effect of eco-labeling on demand for the unlabeled good is nonpositive, and the effect

on demand for the "green" good is nonnegative. For example, consumer Type 1 faces

first-order conditions described in Table 2-1. An increase in the value ofp raises the

perceived value of the eco-product relative to its price, and increases the likelihood that

consumer Type 1 will purchase the environmentally differentiated ferns. For consumer

Type 2, an increase in the value of p increases the likelihood that he will switch from

consumption of unlabeled ferns to eco-labeled ferns.

For consumer Types 3 and 4, however, the effect of eco-labeling (increasing p), on

the quantities qe and q, consumed, is theoretically ambiguous without additional

assumptions. The possible (although unlikely) result the qu could rise, depends on the

idea that the eco-label leads the (pre-labeling) consumer of qe to believe that his

contribution to the public bad, B, is lower than previously thought. After labeling, the

consumer could choose more qu and less qe to obtain more X, but with a lower perceived

contribution to the negative characteristic B. As long as the perceived reduction in B

owing to the increase in p outweighs the increase in B owing to the higher quantity of q,,

the consumer perceives a net reduction in B and an increase in X.

This theoretical possibility is ruled out, however, if we assume that eco-labeling

does not change the consumer's marginal valuation of the public bad, B. Specifically,

under the assumptions of separable utility and negative, constant marginal utility with

respect to B (within the relevant range), an increase in the value ofp has an

unambiguously positive effect on eco-demand and an unambiguously negative effect on

unlabeled demand by all consumer types. Under this assumption (and the condition that

both qe and q, are positive prior to labeling) the comparative static effects on qe and qu are

8q,- p. UpUrr -UUx,
(2-28) P2uU= -UBU
ap xU rrU(-p,2 p + 2p, e)


(2-29) q, = p, peUBU, + UUx
ap UxU, (-p, -p, + 2pp,)

where Ujj is the second partial derivative of utility with respect to characteristic. Under

standard assumptions, Uyy, Uxx, and UB are negative. Additionally, as long as pu does not

equal pe, Eq. 2-28 is clearly positive, and Eq. 2-29 is clearly negative. Under the

assumption of separable utility and negative, constant marginal utility with respect to B,

eco-labeling causes an increase in consumption of environmentally differentiated ferns

and a decrease in consumption of unlabeled ferns.

Much like advertising, eco-labels increase the perceived quality of the

environmentally differentiated product, increasing its demand relative to similar

competing products. In aggregate, eco-labels supported by a credible third-party

certification process can be expected to increase demand for the good making the

environmental claim and decrease demand for the unlabeled version of the same

commodity, ceteris paribus.


Whereas an understanding of consumer response to eco-labels provides insight

regarding potential effects on demand, the effects on production and ultimately the

environment depend on producer responses. In this chapter, we review literature and

introduce a producer model that provides a framework for anticipating supply responses

to eco-labels.

Literature Review

Theories of producer choice of product quality, advertising level, and technology

adoption provide a framework for analyzing producer response to certification and

labeling programs. We review literature on producer behavior regarding product quality,

where quality refers to a product characteristic that can be changed through adjustments

to the production process. Also, models of producer decisions concerning advertising,

technology adoption, and environmental certification are reviewed. In addition, we

summarize existing literature on empirical research pertaining to environmental

management, certification and labeling decisions of producers.

Theoretical Models of Producer Decisions

Dorfman and Steiner (p. 832) derive the marginal conditions for a firm to choose

the optimal level of product quality, defined on a single dimension. Given an average

cost function, c = c(q,z), and demand curve, q =f(p, z), where q is quantity, z is quality, c

is average production cost, and p is product price, the first-order (necessary) condition for

profit maximization with respect to the quality choice in their model is

(3- af af/az
(3-1) _z .
ap ac/az

The first-order condition expressed in Eq. 3-1 can be written as

(3-2) =
az 8z

Equation 3-2 equates the marginal change in price with the marginal change in average

cost, both with respect to product quality. In other words, firms will increase quality as

long as the resulting increase in unit price is greater than the increase in average

production cost, for any fixed quantity of output.

Rosen considers the case of individual production establishments (plants), each

producing only one type of product. An individual plant has a total cost function, C(q, z;

P), where q is production quantity, z is the vector of product characteristics, and fl

represents cost function parameters that vary among plants. Each plant chooses q and z

to maximize profit, H = qp(z) C(q, zi,..., z,), where p(z) is the product price as a

function of characteristics. The first-order (necessary) conditions for profit maximization

in Rosen's model are

(3-3) p(z)=C,(q,z,,...,z,)


(3-4) p,(z) =C, (q,z,,...z,)/q.

Equation 3-3 is the condition that marginal cost (relative to quantity) equals unit

price. Equation 3-4 is the condition that "marginal revenue from additional attributes

equals their marginal cost of production per unit sold" (Rosen, p.42). This latter

condition is essentially the same as the optimal quality condition identified by Dorfman

and Steiner. A firm will improve the quality of its product as long as the resulting

increase in price is greater than the increase in average cost, for a given quantity of


Stigler and Becker describe a firm's profit function with respect to quantity of

output and level of advertising as

(3-5) H = pq C(q)- Ap,.

Noting Eq. 2-14 and the relationship between implicit demand for a characteristic and

market demand for a good, Stigler and Becker substitute the shadow value, 7r, of the

product's characteristic times the coefficient, g(A), measuring the amount of quality

characteristic contained in the market good, for the market price. Then the profit function

can be written as

(3-6) H = )rg(A)q- C(q)- Ap,.

The necessary first-order condition for profit maximization with respect to advertising in

Stigler and Becker's model is

(3-7) =pq z qag=p,.
aA aA

Dividing through by q, this condition is similar to Eqs. 3-2 and 3-4. The only technical

difference is that the effect of advertising increases market demand (price), not by

shifting implicit demand for characteristics, but by increasing the input-output coefficient

describing the (perceived) quality of a market good.

Tirole analyzes the case of a monopolist choosing optimal quantity and quality of

output. Given an inverse demand function, P(q,z), and a total cost function, C(q,z),

where q is quantity and z is quality, the monopolist would maximize

(3-8) = qP(q,z)- C(q,z).

The necessary first-order condition with respect to the choice of quality for Tirole's

monopolist model is

(3-9) qPg(q,z)= C (q,z),

where Pz and Cz are partial derivatives with respect to quality.

Ideally a social planner would try to maximize social welfare, defined as "the

difference between gross consumer surplus and production cost" (Tirole, p. 100). The

social planners objective function would be

(3-10) W(q,z)= fP(x,z)dx-C(q,z).

Tirole's formulation in Eq. 3-10 assumes unit demands among the consumer population.

"[P(x,z)] is then the price that makes the xth consumer indifferent between buying one

unit of the good of quality [z] and not buying" (Tirole, p.100). The necessary first-order

condition for the social planner in Tirole's model is

(3-11) 1P,(x,z)dx=C,(q,z).

Tirole explains the difference between the first-order conditions for the monopolist vs.

the social planner. "The incentive to provide quality is related to the marginal

willingness to pay for quality, for the marginal consumer in the case of a monopolist and

for the average consumer in the case of a social planner" (Tirole, p. 101).

We note that dividing both sides of Eq. 3-9 by q produces the now familiar

condition expressed in Eq. 3-2 from Dorfman and Steiner. The first-order condition for

optimal choice of quality is no different for the monopolist than for the competitive firm.

A competitive firm cannot influence price by changing quantity, as a monopolist can, but

it can influence price by changing quality. For any given level of quality, the competitive

firm still acts as a price-taker. Implications are that neither the monopolist, nor the

competitive firm, will provide the socially optimal level of product quality, except under

very restrictive conditions.

Hays models a firm facing a discrete choice of quality levels (e.g., whether to

obtain environmental certification-and label its products as such-or not). The firm's

profit function in Hays' model is defined over discrete alternatives as H* = max {lE, HN}.

If the firm chooses to adopt environmental certification and labeling its profit is

(3-12) E =pq-C(q)- F.

If the firm chooses not to adopt environmental certification and labeling its profit is

(3-13) IIN = p,q-C(q)-F,.

The term C(q) is the total variable cost function, and F, is total fixed cost. Hays'

formulation allows one to estimate empirically the firm's adoption decision as the

probability that HE > IN, which depends on the firm's price and cost expectations, as

well as firm characteristics.

Although expected short-run profit is often a primary objective of producers, other

objectives enter their decision-making as well. Risk is often an important factor in the

decisions of a producer or investor. One formulation has the producer maximizing a

utility function defined over expected return and a measure of risk. Risk could be

measured as variance of returns (Markowitz) or mean absolute deviation (Hazell). Isik

and Khanna model a farmer's decision to adopt site-specific technologies (SSTs) and

consider a farmer's preferences defined over the mean and standard deviation of a

stochastic profit function. In their model, the farmer's objective function is:

(3-14) MaxU(rnI + I(n -F -K),c +I(c'-ac)),

where the decision variables X and I refer to variable input level and the choice of

adopting SSTs respectively (I = I if SSTs are adopted; I = 0 if SSTs are not adopted).

The term 17is quasi-rent, a is standard deviation of quasi-rent, and K is the fixed cost of

implementing SSTs. The superscripts c and s refer to conventional practices and site-

specific technologies respectively.

Farmers and firms consider various other factors in their decisions. Philosophical

outlook, environmental and health concerns, risk of liability for environmental damage,

and pressure from various groups can influence the response to voluntary environmental

programs. Empirical studies demonstrate the importance of these factors.

Empirical Research

Henriques and Sadorsky find that firms do in fact respond to pressure from

customers, shareholders, government, and neighborhood and community groups in

formulating environmental plans. Evidence also suggests that public recognition and the

risk of environmental liability provide incentives for firms to adopt voluntary

environmental management systems, trading off short-run returns for expected long run

profitability (Khanna and Damon; Khanna and Anton). The results of a survey of organic

citrus growers, presented in Chapter 6, demonstrate that environmental and health

concerns, as well as the threat of environmental liability, were significant factors in the

decision by some growers to adopt organic practices.

Gobbi collected data on five different types of coffee farms in El Salvador to assess

the financial feasibility of obtaining "biodiversity-friendly" certification. Simulation

results from the study found that mean expected gains in net present value associated

with certification were positive for all five farm types. Despite the favorable results,

Gobbi concludes that the initial capital required to convert to biodiversity-friendly

practices and to obtain certification acts as a deterrent, especially for small growers.

Murray and Abt estimated a compensation function for eco-certified forestry in the

southeastern United States. The function traces out the threshold price premiums

necessary for heterogeneous forest enterprises to convert land to eco-certified practices.

Their simulation results show that a significant portion of forest land would adopt

certification at relatively modest price premiums. The authors caution that environmental

improvement would be modest at low price premiums, because the only adopters would

be forest enterprises that are already using environmentally friendly practices.

Boltz, et al. compared the financial performance of conventional and reduced-

impact logging (RIL) in the Brazilian Amazon. The authors found that reduced-impact

logging is more profitable than conventional and can reduce production costs in some

cases. They suggest that uncertainty about new practices is one reason that RIL is not

more widely adopted.

Bell, et al. identified factors influencing the probability that a landowner would

participate in Tennessee's Forest Stewardship Program. The authors found that a

landowner's attitude towards conservation goals and knowledge of forestry programs can

have a stronger effect on the likelihood of participation than monetary incentives. They

conclude that environmental education, training and technical assistance would be more

effective than financial cost-share incentives at increasing participation in the

environmental management program.

Hays surveyed firms in Oregon to assess their likelihood of participating in an

environmental labeling program to protect water quality. She found that participation

costs and consumer demand expectations were primary motivators. Other factors had

little significance.

Model of Producer Response to Eco-Labels

A producer model provides insight as to the most likely effects of eco-labeling on

supply. Producers want to maximize profits from their production enterprise, although

other objectives may enter their decision-making process. Eco-labeling enables

producers who generate environmental benefits to differentiate their products in the

market with the potential for higher prices and higher returns. In addition to the

incentives created by price premiums for eco-labeled products, producers may be

motivated to adopt environmental certification because of a various other reasons

described previously.

Producer response to environmental labeling and certification can be broken down

into three separate decisions. First, a producer can choose to adopt EMPs or not on a

particular production block. If a producer adopts EMPs, he then has the option to obtain

certification or not. Producers that are EMP-certified have the option to market their

products with an eco-label or sell them undifferentiated through conventional market

channels. A producer with multiple production blocks or plots can make a separate

choice for each production block.

Considering the three related decisions facing a producer for a given production

block, a possible formulation of the producer problem is

(3-15) MaxF = p,q C(q, w, k) c, (w, k, v)q, c, (k, v, p)q, c, (p)q, + p,(p)q,
q'qx ,q, ,q,

q, s.t. q, < qg
q, 5 q,

where I is the producer's expected profit. Decision variables are q, qg, q,, and qe, which

are the quantity of total output, quantity of "green" output, quantity of certified output,

and quantity marketed with an eco-label respectively. The parameters, p, and p, are the

conventional (unlabeled) price and the eco-label price premium (p, = pe -pu)

respectively. The term C() is a standard cost function, and the c() terms are additional

per unit costs (additional to C/q). In particular, g(.) is the additional (average)

production cost associated with using "green" practices, ct() is the additional (average)

cost of obtaining certification, and c,() is the additional (average) cost of marketing with

an eco-label. The parameter w represents a vector of input prices, k is a vector of

characteristics associated with an individual producer and production block, v is a vector

of nonprice incentives created by certification and labeling programs, and p is the

credibility or level of consumer trust in an environmental claim.

Additional clarification is in order, in particular for the non-price incentives, v.

Here we consider two types of nonprice incentives identified in the literature. The first

one is a reduction in costs of complying with EMP ("green") production standards. In

some cases, the development of an environmental certification program improves the

information available to producers. Research and extension activities that disseminate

information in conjunction with a certification program can reduce the costs of green

production. For example, some reduced-impact logging techniques have been shown to

increase efficiency and lower costs (Boltz et al.). Also, best management practice (BMP)

and integrated pest management (IPM) programs have been known to reduce producer

costs, in particular expenditures on pesticides, water, and fertilizers (Leppla; Stamps).

"IPM certification programs have been promoting environmental stewardship, reducing

production and processing costs, improving profits through IPM labeling, and protecting

growers and processors from accusations of pesticide misuse" (Leppla, p. 2).

Development of BMP and IPM programs and related education and extension activities

can help producers become more efficient, increasing grower knowledge of techniques to

reduce costs and environmental impacts, while maintaining yields. Thus the parameter v,

appearing in the cg() function, represents information that reduces costs of using "green"

practices. For some producers, "green" production may be less costly than conventional

production. The cg() function would take a negative value for those producers.

The second type of non-price incentive represented by v relates to a reduction in the

perceived risk of a producer being held liable for environmental damages. This incentive

has been documented or discussed by Henriques and Sadorsky; Khanna and Damon;

Khanna and Anton; Leppla; and WWF. Although not an immediate (short-run) cost of

production, the perceived threat of liability is part of a producer's long-run cost

expectations and could be incorporated in the cost function, C(). If an EMP certification

has legal standing with the government, then obtaining certification could reduce the

perceived risk of liability for a producer. Thus, the parameter, v, in the c(.) function

would have a cost-reducing effect. The full cost of certification for a producer includes a

certification fee paid to the certifying agency and a reduction in expected costs associated

with environmental liability. The net "cost" of certification for a given producer may be

positive or negative.

Another point of clarification regards the inclusion of the parameter p in the

certification and eco-labeling cost functions, c,() and ce(). Steps that increase the level

of credibility of an environmental claim in the eyes of the consumer include on-site

inspection, monitoring and oversight activities by the certifying body, as well as chain-of-

custody and identify preservation measures to keep the certified product separate from

uncertified products through processing, storage, transport, and distribution. In this

sense, additional certification expenses (and therefore fees passed back to the producer)

and marketing costs associated with selling an eco-labeled product must be incurred in

order to increase eco-label credibility, p. Thus, higher certification and labeling costs are

associated with higher levels of credibility. If a producer has a choice between multiple

certifying bodies and eco-labels representing a similar environmental standard, p would

be a decision variable affecting costs and the price premium received for a particular eco-

label. More typically, however, a producer has only the choice of one eco-label for a

particular environmental standard, and must either accept or reject its associated costs

and price premium. In that case, p is an exogenous variable associated with the available

certification and labeling program.

The formulation of the producer problem in Eq. 3-15 requires that the quantity of

output using green practices, qg, is less than or equal to the producer's total output q.

Likewise, certified output, q,, must be less than or equal to qg, and eco-labeled output, qe,

must be less than or equal to certified output.

The Lagrangian for the producer problem generates the following Kuhn-Tucker


(3-16) p, -Cq(q) + < 0

(3-17) -cg(v) + I2 <0

(3-18) -c,(v,p)-Pu2 +/3 <0

(3-19) p,(p)-ce(p)-,P3 0

(3-20) q c 0
ap p ap

where li is the Lagrange multiplier associated with each of the three constraints in Eq. 3-

15, numbered consecutively. Cq is marginal production cost, and cg, ct, and c, represent

additional per unit costs described previously. It is assumed that the producer is a price-

taker with respect to p, and pr, and that the additional average costs, Cg, c,, and c, do not

change with output quantity q.

Given these assumptions, the profit-maximizing producer will choose an output

level that equates marginal cost with the unlabeled output price plus additional net per-

unit benefits that accrue only if the producer chooses to comply with EMPs (assuming

total price is higher than total average cost). If the producer chooses not to comply with

EMPs (qg = 0), then p, = 0, and pu = Cq(q). If the producer chooses to comply with

EMPs (qg > 0), then pI > 0 andpu + U2 cg(v) = Cq(q), where u2 represents any

additional net per unit benefits associated with certification and labeling. If the producer

complies with EMPs, but does not choose to obtain certification and market products

with an eco-label (q,, q, = 0), then p2 = 0, andpu cg(v) = Cq(q). We note that in this

case, the additional "cost" associated with green production for the producer must be

negative, (i.e., this producer associates net benefits with green production over

conventional production). If a producer chooses green production, certification, and

labeling, then

(3-21) pu + Pr(p) Cg(V) c,(v,p) Ce(p) = Cq(q).

The eco-label price (pe = Pu + p,) minus any additional per unit costs associated with

green production, certification, and labeling equals the marginal production cost for the

profit-maximizing producer.

By assuming that the producer is a price taker and that additional per unit costs do

not vary with output quantity, the three constraints in Eq. 3-15 must either hold with

equality or the left-hand-side variable equals zero. This implies that producers face a

discrete choice relative to the three decisions concerning green production, certification,

and labeling. Either all the output from a production block is "green" or none of it is.

Likewise, if the block is managed using "green" practices, either the entire block is

certified or not. Similarly, either the entire output from one production block is labeled,

or none of it is.

To show that our model relates closely to the producer models in the literature, we

simplify the model formulated in Eq. 3-15 by assuming that the producer faces only one

decision, either "green" production, certification, and labeling, or conventional

production. For the marginal producer, indifferent between the two options, the

following conditions must hold

(3-22) p, =C,(q)

(3-23) p,(p)= c,(v)+ c,(v, p) + c,(p)

Equation 3-23 is a discrete choice condition similar to Eqs. 3-2, 3-4, 3-7, and 3-9

identified by Dorfman and Steiner, Rosen, Stigler and Becker, and Tirole. The left-hand

side of Eq. 3-23 is the change in output price received by adopting green practices,

certification, and labeling. The right-hand side of Eq. 3-23 is the change in average cost

associated with adopting green practices, certification, and labeling. Our discrete choice

formulation assumes that the producer does not have control over the credibility

parameter p.

In the case that the producer can choose among multiple certifying agencies and

multiple labels for the same environmental management standard, and that each label has

a different level of credibility in the eyes of consumers, the profit-maximizing producer

would choose a label and the associated value ofp so that Eq. 3-20 holds with equality.

Dividing Eq. 3-20 through by qe = q,, our result is essentially the same as the Dorfman

and Steiner and Rosen results expressed in Eqs. 3-2 and 3-4. When the producer can vary

quality of output, she will increase quality until the marginal change in price with respect

to quality just equals the marginal change in per unit cost with respect to quality.

Likewise, the producer in our model would choose p so that the marginal increase in the

price premium with respect to p just equals the marginal per unit costs associated with

increasing p.

Assuming that other exogenous factors, such as w and k, are constant, producer

responses to the development of environmental certification and labeling programs

depend on the effects of v and p. Based on aggregate producer responses, we can specify

supply functions, S,[p,, pr(p,p), v, p] and Se[pu, p,(p,P), v, p], where p, = pe(P) -Pu.

Nonprice incentives created by dissemination of information to producers, and by the

opportunity to lower the risk of liability for environmental damages, reduce "green"

production costs relative to conventional production. The enhancement of credibility of

environmental claims generated by an eco-label increases the price premium received by

producers of eco-labeled products. Both these price and nonprice incentives serve to

increase "green" supply relative to conventional supply.


Given our conclusions about the direct effects of eco-labeling on demand and

supply for competing products, we now extend the analysis to consider the ultimate effect

of eco-labeling on the environment. After a brief literature review, two theoretical

models are used to identify the connection between the direct effects of eco-labeling on

demand and supply and the ultimate environmental impact. First, a two-product, partial-

equilibrium model is used to identify the effects of eco-labeling on conventional

production quantity under different market conditions. Second, a price-endogenous

programming model is created to simulate the introduction of an eco-label and the

resulting effects on production, land use, and the environment in a producing region.

Equilibrium conditions are identified and the simulation model is run using the General

Algebraic Modeling System (GAMS) to demonstrate hypothetical effects under different

conditions. Lastly, results are summarized and discussed.

Literature Review

The debate over the environmental effectiveness of eco-labeling is highlighted in

three recent articles on shade-grown coffee labeling in the journal Conservation Biology.

Rappole, King, and Vega Rivera (2003a; 2003b) argue that shade-grown coffee programs

have had negative impacts on land-use and biodiversity in Latin America. "If the sole

result of the promotion of shade coffee were to encourage growers to convert sites that

are currently in sun coffee to shade coffee, then there could be few qualms from a

conservation perspective about the campaign so wholeheartedly endorsed not only in lay

publications but scientific journals as well" (Rappole, King, and Vega Rivera 2003a,

p.334). The authors suggest, however, that a common standard protecting biodiversity is

lacking and that "promotion far outstrips certification" (Rappole, King, and Vega Rivera

2003b, p.1848). They contend that promotion of shade-grown coffee has caused

additional clearing of native forests to the detriment of biodiversity in the tropics.

Philpott and Dietsch defend coffee certification and labeling programs for their

conservation successes and for reducing incentives to employ more harmful land-use

practices. These authors contend that coffee overproduction and low prices led to

"widespread environmental and social disasters" (p.1845) and created incentives for

forest conversion to sun coffee, cattle pasture, swidden agriculture, and coca and poppy

production. They point out that leading certification programs require "a diverse canopy

and certain levels of structural diversity" (Philpott and Dietsch, p. 1845) and can provide

an economically viable alternative to activities with far worse environmental impacts.

The key source of disagreement between the two sets of authors, cited above,

concerns alternative land-uses. Is pristine forest, or more intensive agriculture, the

primary land-use alternative to shade-grown coffee? Does promotion of eco-labeled

coffee lead to the clearing of more primary forest land or deter the conversion of land to

sun coffee, cattle pasture, and other environmentally inferior land-uses?

In other literature, one finds a variety of opinions about the potential effectiveness

of eco-labels. Antle provides an optimistic assessment of the potential for labeling and

other information-based policies to affect positive environmental change. Murray and

Abt, and Kiker and Putz, cast doubt upon the ability of environmental certification and

labeling programs to protect the environment. Despite speculation of all sorts, the U.S.

Environmental Protection Agency (1998, p. 59) concluded in 1998 that "to date, the

effectiveness of labels as a policy tool has not been thoroughly studied."

As described in previous chapters, several studies have considered individual

components of an eco-label's effectiveness. Some studies have evaluated consumer

demand and willingness-to-pay premiums for eco-labeled products. Others have

analyzed producer responsiveness to certification and labeling programs. Although these

studies provide empirical evidence regarding important components of an eco-label's

impact on demand or supply under specific circumstances, they do not consider the full

interaction between supply and demand for competing products and the ultimate

environmental outcome.

The Environmental Protection Agency (1994, p. 5) describes the theoretical link

between the effect of eco-labels on consumer demand and environmental outcomes:

With increased acceptance of an environmental label, consumer behavior changes
may take the form of increased demand for products with [eco-labels], or decreased
demand for products carrying negative warning labels. This in turn would effect a
market shift toward products that are presumably less damaging to the environment
than other similar products in their classes. In theory, a market shift of this sort
will reduce the total environmental impacts of consumer products.

The exact linkages between consumer awareness, demand, and environmental impacts

are left rather ambiguous. Several questions remain. Under what circumstances would

an eco-labeling program be most effective at reducing adverse environmental impacts?

Under what circumstances would eco-labeling be least effective? Is it possible that an

eco-labeling program could increase environmentally harmful production in some cases?

These questions are addressed in this chapter.

Few studies analyze the effects of eco-labeling by considering supply and demand

factors for both the eco-labeled and unlabeled product, and the ultimate impact on the

environment. Mattoo and Singh used a graphical analysis to model supply and demand

for an environmentally friendly product and its environmentally unfriendly counterpart.

They produced an interesting result: if the quantity of eco-supply exceeds the quantity of

eco-demand at the undifferentiated (pre-labeling) price, the quantity of environmentally

unfriendly production could increase. Their analysis is limited, however, in that it does

not account for substitution effects, direct impacts on supply (shifters), changes in

marketing costs, or a possible reduction in demand for the environmentally unfriendly

product after the eco-label is introduced.

Sedjo and Swallow use a graphical supply and demand model and consider

substitution effects. They explore the possibility that a price differential may not arise

even when some consumers are willing to pay a premium and identify conditions under

which non-certified producers could lose or gain from eco-labeling. Concerned primarily

with the effects of eco-certification on price differentials and returns to producers, their

analysis does not address effectiveness in terms of changes in production quantities and

ultimate environmental impacts.

Swallow and Sedjo consider the effects of eco-labeling in a general equilibrium

framework. The authors address the issue of unintended feedback effects, and provide an

example that "raises doubts that market changes from certification will lead to large-scale

ecological improvement unequivocally" (p. 33). Their graphical analysis is limited to

mandatory programs that increase producer costs.

In this chapter, we build on the analysis of Mattoo and Singh, Sedjo and Swallow,

and Swallow and Sedjo, and contribute two models that provide a useful framework for

anticipating the market interaction effects caused by eco-labeling programs. The main

objective is to identify explicit conditions that determine changes in production quantities

and the ultimate environmental impact of certification and labeling. A two-product,

partial equilibrium model and a price-endogenous programming model generate

equilibrium conditions that provide a basis for comparative static analysis. The GAMS

software is used to provide simulation results based on the programming model.

Conceptual Background and Assumptions

We consider the case of a voluntary environmental certification and labeling

program designed to protect water quality and preserve water resources in Florida, such

as the expanded environmental management practice (EMP) label envisioned for the

ornamental plant industry. Leatherleaf ferns serve as an example. The analysis assumes

that ferns can be produced using either conventional production methods (without

adhering to EMPs) or "green" production methods (in which EMPs are adopted).

Conventional fern production methods are known to have negative impacts on water

quality, with potential risks for wildlife and human health and the possibility of reducing

recreational values and increasing costs of water treatment and supply (Leppla; Larson

Vasquez and Nesheim; Stamps). Environmental management practices reduce the risk of

negative impacts on Florida water resources (Leppla; Larson Vasquez and Nesheim;

Stamps). It is assumed that fern production using EMPs has negligible impacts on water


Producers using EMPs can apply for certification. A certified producer receives

public recognition and is entitled to sell ferns bearing an eco-label. We assume that eco-

labeled and unlabeled ferns are differentiated in the eyes of the consumer only by the

environmental impact of their production methods, a credence attribute. Eco-labeled and

unlabeled ferns, therefore, are assumed to be close substitutes in demand and supply.

Oversight and monitoring by a reputable third-party certifying agency are needed to

convince consumers of the environmental distinction.

We assume that some producers use green production methods prior to

implementation of a certification program. This is not uncommon. For example, some

ornamental plant nurseries in Florida report using integrated pest management (IPM)

practices even though a certification program is not yet available (Hodges, Aerts, and

Neal; Leppla). In terms of the producer model introduced in Chapter 3, these producers

have a negative cg (i.e., green production entails an additional net benefit over standard

conventional production costs for the producers that adopt EMPs without obtaining

certification or utilizing an eco-label). Also, some producers may advertise their products

as environmentally friendly without utilizing an eco-label.

Furthermore, a limited consumer segment may seek out greener products based on

producer claims, without relying on an eco-label. For example, farmers market shoppers

can talk directly to growers and may trust that their products are organic, pesticide-free,

or produced using EMPs without relying on a third-party seal-of-approval.

We assume that ferns produced using EMPs may be sold with or without an

environmental claim, but ferns produced without using EMPs cannot be sold with an

environmental claim. The possibility of fraud, in which conventionally produced ferns

are sold with an environmental claim, is not considered in the models that follow.

Competitive market equilibrium is assumed (i.e., individual consumers and

producers act as price takers). Information, however, is imperfect. Certification and

labeling serve to improve the information available to producers and consumers.

Information that creates non-price incentives for producers is treated separately from

information embodied in an eco-label and directed at consumers. Marketing costs are

separated from production costs.

Models of producer and consumer behavior provide insight as to the most likely

initial impacts of certification and labeling on supply and demand for the two products.

Ultimate changes in production quantities and resulting environmental impacts, however,

depend also on the interaction between other market factors, such as pre-labeling

conditions, the relationship between own- and cross-price effects, and marketing cost.

Next, we formulate a two-product, partial-equilibrium model to identify the effect of eco-

labeling on production quantities, considering these market conditions.

Two-Product, Partial-Equilibrium Model

A two-product, partial-equilibrium model is used to understand the interaction

between an eco-label's direct effects on supply, demand, and marketing costs, own- and

cross-price elasticities of supply and demand, and initial market conditions. For example,

when the demand curve shifts by a certain amount (at initial prices), the resulting change

in quantity produced is a function of the price elasticities of supply and demand. A

demand-side substitution effect occurs, if demand in one market increases as price rises

in the market of a close substitute, and vice versa. Supply-side substitution implies that

the supply curve in one market shifts in response to price changes in another market.

Based on the consumer and producer models presented in Chapters 2 and 3, the

initial impact of introducing an environmental certification and labeling program is

represented by a change in the parameter p on the demand side, and the parameter v on

the supply side. The parameter p is a measure of the credibility of an environmental

claim. An increase in p is assumed to have a positive impact on demand for the

environmentally differentiated good and a negative impact on demand for the unlabeled

good. The parameter v is a proxy for nonprice incentives associated with short- or long-

run production costs. An increase in v reduces costs of green production relative to

conventional production, increasing green supply and reducing conventional supply, with

farm-gate prices held constant.

In addition, the effect ofp on marketing costs is considered. Improving the

credibility of environmental claims through certification and labeling is not without cost.

Inspection and monitoring of producers, recordkeeping, and identity preservation through

the marketing channel are costly aspects of certification and labeling. Although

certification fees are common, they are often paid or subsidized by government or

nonprofit organizations. For example, certification rebates have been granted to farmers

who obtain organic certification in Europe and the U.S. In the case that certification

costs are borne by the market, these expenses could be considered part of the marketing

costs associated with distributing and selling to the eco-market. Let Me represent the

average cost of distributing and selling to the eco-market (additional to green production

cost), and let Mu represent the cost of distributing and selling to the undifferentiated

market (additional to conventional production cost). If adoption of certification and

labeling (associated with an increase in p) entail costs additional to green production,

then the partial derivative of Me with respect to p is positive (aM, /lp > 0).

Departing slightly from the producer model in Chapter 3, we distinguish between

farm-gate prices, Pc and Pg (for conventional ferns and "green" ferns, respectively), and

retail prices, P, and Pe (for unlabeled ferns and ferns sold with an eco-label or

environmental claim). These prices are considered endogenous in the market equilibrium

model. Other endogenous variables are the aggregate production quantities, Qg

(green)and Qc (conventional). Production quantities are expressed in retail units. The

exogenous variables are p and v, representing the information-based effects of credibility

for consumers and non-price incentives for producers. All other relevant variables are

held constant and omitted from the analysis.

The parameters p and v are treated as continuous variables. Treating them as

continuous facilitates comparative static analysis, and mathematical system models with

continuous flows tend to be more illuminating than those that treat variables as discrete

events (Forrester). "Real systems are more nearly continuous than is commonly

supposed....A continuous-flow system is usually an effective first approximation even

where...discrete decisions and events do occur" (Forrester, p.64-65).

Like negative publicity, advertising intensity, a media event, increased cooperative

extension activities, or other information-related changes, the variables p and v are not

easily quantifiable. They can be estimated, however, ex ante, using survey research, or

ex post, using market data. For example, Teisl, Roe, and Levy use surveys in which

consumers are asked to rate the credibility of different labels, and Teisl, Roe, and Hicks

measure the impact of the dolphin-safe label on demand for tuna. Teisl and Roe discuss

various implications of measuring consumer response to environmental labels.

Comparative static analysis demonstrates how the equilibrium values of

endogenous variables, such as production quantities, change as a result of exogenous

changes in parameters. In particular, the analysis identifies the direction and rate of

change in an equilibrium value of one variable with respect to a change in another.

Comparative statics does not provide insight regarding the path, process, or timing of

adjustment (Chiang).

If the negative environmental effects of conventional production are significant and

the environmental effects of green production and the alternative uses for factors of

production are negligible, the environmental impact would be proportional to the change

in conventional production. A decline in conventional production would result in an

environmental improvement, and an increase in conventional production would cause

more environmental harm. Based on this assumption, we focus the subsequent analysis

on the effect of eco-labeling on the quantity of conventional production, Qc. In a later

section, a price-endogenous programming model is used to consider alternative


Using the partial-equilibrium model, we consider two different market equilibrium

scenarios. In the first market scenario, the entire supply of "green" product is sold with

an environmental claim (on the "eco-market") even before an eco-label is introduced.

This condition would occur if green supply is small relative to eco-demand. Even a small

eco-market (of consumers dedicated to seeking out greener products prior to labeling) is

enough to absorb green production and support a "green" producer price at least as high

as the conventional producer price. In the second scenario, excess green supply is sold

on the undifferentiated market without an environmental claim. This second case

suggests that green supply is large relative to eco-demand. The eco-market is not large

enough to absorb all the green production, and green producers receive the same farm-

gate price whether they sell to the eco-market or the undifferentiated market.

The first scenario is depicted in Figure 4-1. Demand quantity equals supply

quantity for each product, and none of the green supply is sold on the undifferentiated

market. The producer price equals the retail price minus marketing costs for each market.

Constant returns to scale in marketing are assumed (per unit marketing cost does not

change as the quantity marketed changes). The equilibrium conditions for this first

scenario are

S,(P,,P,,v)-Q -O
S,(P, Pc,v)-Q, -0
Sc(P",Pg',p)-Q O

(4-1), p) Q 0
D, (P,, P, p) -Q 0
P, +M,(p)- P,0
P,+M, P O
where Sg and Sc are the green and conventional supply functions, respectively; De and D,

are eco- and undifferentiated demand functions, respectively; and Me and M, are

marketing costs associated with the eco-market and the undifferentiated market,



Undifferentiated Market

Qg Qc

Figure 4-1. First market equilibrium scenario

First considering the effects of eco-label credibility, p, on demand and marketing

cost, the resulting impact on conventional production is represented by the comparative

static derivative, dQ /dp :

[SD,, -SSccDe +S S, -S Sc ]g D

[S,, D,,e S, D,, ] a
(4-2) d = +
dp IJ\

[D,,(SS,,_ -S,Scg) Sc(D,,D,, D,.Du)] aM
+ ap

The mathematical procedures are described in Appendix A. The denominator in Eq. 4-2,

represented by IJ\, is the endogenous variable Jacobian determinant, which is positive in

this case. The bracketed numerator terms in Eq. 4-2 are the partial derivatives of supply

and demand with respect to own price or the other product's price. For example, Sgg is

the change in green supply quantity with respect to a change in green producer price

(aS, l/P, ), and Sgc is the change in green supply quantity with respect to a change in

conventional producer price (aSg, /P, ). The term D,, is the change in unlabeled demand

quantity with respect to a change in unlabeled retail price (aD, I/P, ), and Due is the

change in unlabeled demand quantity with respect to a change in the retail price for the

eco-labeled product (aD, /P, ).

It is assumed that supply and demand functions are well-behaved and that the two

products are substitutes in supply and demand. Thus, own-price supply responses are

positive; cross-price supply responses are negative; own-price demand responses are

negative; and cross-price demand responses are positive. Normal properties of supply

and demand imply that own-price effects outweigh cross-price effects. This result

follows from the negative semi-definiteness of the substitution matrix. Given these

assumptions, the comparative static result in Eq. 4-2 shows that aD, /ap and aM, /ap

have positive effects on conventional production, Qc, whereas the effect of aD, /ap is


The direction of change in conventional production resulting from an increase in

eco-demand depends on the sign of the term S, D,,, S, D,,. If this term is negative,

increasing demand for the eco-labeled ferns has a negative effect on conventional

production as intended. The term is likely to be negative if the rising green price causes

conventional producers to switch to green production (Scg is large), but does not cause

many consumers to switch back to undifferentiated consumption (Due is small). Any

increase in marketing costs associated with labeling, however, offsets the effect of

increasing eco-demand. Conventional production will fall, only if the negative effect

from rising eco-demand outweighs the positive effect of increased marketing costs.

An increase in eco-demand has a positive (unintended) effect on conventional

production, if S,, D, -S D,, is positive. The term could be positive if the increase in

green price does not induce many conventional producers to switch to green practices (Scg

is small), but causes "previously green" consumers to switch to unlabeled consumption

(Due is large). This effect might occur when an eco-label allows access to a foreign

market or large segment of consumers that had previously avoided the class of products

(ferns from the producing region). The resulting rise in price could induce some

"previously green" consumers to buy the unlabeled product instead. An increase in


conventional production caused by the initial impacts of eco-labeling on demand is only

possible if the increase in eco-demand is much greater than the decrease in

undifferentiated demand (i.e., 8DD /apl > aD,, /ap ).

If marketing costs are unchanged and the decrease in undifferentiated demand

equals the increase in eco-demand (i.e., JaD /ap = aDU /apl ), conventional production

declines (regardless of the sign of S, D,,, S D,, ). This result is obtained by comparing

the bracketed terms associated with 9D, /ap and aD /Qp and noting the assumption that

own-price effects are larger in absolute value than cross-price effects. The comparative

static result, presented in Eq. 4-2, highlights the importance of a decline in

undifferentiated demand for reducing conventional production and its harmful

environmental impacts.

Now considering the non-price incentives v, which affect supply, the comparative

static derivative for the effect on conventional production is

dQ [(DD,,.D D,,,D,)+(Sg- D,,, -Sg D,,)] +[Sg ,D,, S D, ]-
(4-3) d
dv J

Again, the denominator term 1J] is the endogenous variable Jacobian determinant and is

positive in sign (Appendix A). Given standard assumptions about supply and demand,

the bracketed term in front of S, la/v in Eq. 4-3, [(DeeDu, DeuDue) + (SgcDe SggDu)],

is positive, and the sign of the bracketed term in front of aS, la v, [ScgD,, SccDue], is


The term in front of aSg l/v in Eq. 4-3 is the same as the term associated with

D, /ap in Eq. 4-2, except that the sign is reversed. An increase in green supply due to

nonprice incentives could lead to a rise in conventional production, if falling green prices

cause many producers to switch back to conventional methods (Scg is large) but do not

cause many consumers to switch to eco-labeled products (Due is small). This positive

effect on conventional production is theoretically possible if many new (and efficient)

producers enter fern production ( aS, / vj > IaS /l v), "previously green" growers

switch back to conventional production as the green farm-gate price falls. This

possibility seems less likely than the previous demand-side case of accessing a new

consumer segment. In particular, one would expect demand for unlabeled ferns to

decline significantly (and demand for eco-labeled ferns to increase) as the eco-label price

falls. This effect would eliminate the possibility that increasing green supply could cause

a rise in conventional production.

If the nonprice incentives associated with certification and labeling cause

conventional producers to switch to green methods, but do not cause significant entry into

the producing sector (i.e., aSg, /av = \S,. /l v ), conventional output will decline

regardless of the sign ofScgD,, SccDue. This result is intuitively clear and can be

verified by comparing the bracketed terms in front of aSc /8v and OSg / v and by

considering the assumption that own-price effects outweigh cross-price effects.

Under this first market equilibrium scenario, our analysis shows that a decrease in

undifferentiated demand unambiguously leads to a decline in conventional production.

Also, nonprice incentives that cause conventional producers to adopt EMPs (without

causing significant entry of new producers into the sector) clearly lead to a decrease in

conventional production. The effect of increasing eco-demand on conventional

production is less certain and may be positive in some cases. The analysis of this first

market equilibrium scenario indicates that an eco-labeling program will be most effective

at reducing conventional production when undifferentiated demand declines significantly,

and producers readily convert from conventional to green production in response to price

and nonprice incentives.

The second market scenario is depicted in Figure 4-4. Excess green supply is sold

on the undifferentiated market, without an environmental claim. In a competitive market,

the farm-gate price for the green product equals the farm-gate price for the conventional

product, although retail prices may not be the same. In this scenario, total supply (green

plus conventional) equals total demand (eco-demand plus undifferentiated demand).


Undifferentiated Market


Figure 4-2. Second market equilibrium scenario

Equilibrium conditions for this second scenario are

Sg (P,, P, v)+ S(,P,,, v)- D, (P,,, p)- D,(P., P,,p) 0
S,(P,,P ,V)- Q =0
( S(P,,Pg,v)-Q, -0
P, +Me(p)-P, =0
P, + M. -P. 0
P, + M, P 0
Again considering the effect of improved credibility from the eco-label, p, on the quantity

of conventional production, Qc, the comparative static result is shown in Eq. 4-5:

d [S+ I +Sg C + aD +[(D,, + D,)(S, + S)] aM
(4-5) =ap
dp (S, + Sg)+(Sc + Sg)- (De, + D,,)-(D, + D,)
The mathematical derivation is provided in Appendix A.
Assuming own-price effects are larger than cross-price effects (in absolute value),

the denominator of Eq. 4-5 is positive. The bracketed numerator term in front of

aD, /ap and aD, /ap, [Scc + Scg], is positive, and the bracketed term in front of aM, /lp,

[(Dee + Due)(Sc + Scg)], is negative. In this scenario, the effect of decreasing demand for

the undifferentiated product is exactly offset by an increase in eco-demand. In other

words, if the increase in eco-demand equals the decrease in undifferentiated demand (i.e.,

9D, /apl = ID, /apl) conventional production quantity will not change, ceteris paribus.

On the other hand, if eco-demand rises more than undifferentiated demand falls (i.e., total

demand increases and \aD, /lpi > OD, l/ap ), and marketing costs are unaffected,

conventional production will increase. This result is the same as the one obtained by

Mattoo and Singh. The simple condition that some green supply is sold on the

undifferentiated market prior to labeling creates a strong likelihood that eco-labeling will

result in an increase in conventional production. Opposite from the first market

equilibrium scenario, an increase in marketing costs associated with labeling reduces

conventional production in this case. Considering the total effect of an increase in p, we

conclude that conventional production will increase if total farm-gate demand (after

accounting for any change in marketing cost) rises.

Nonprice incentives affecting supply result in the following impact on conventional


Hs_ + S)' g +
dQ9 aS a+v v
(4-6) dQc S +
dv Ov (S +Sc)+(Sc +Sg,)-(D,, +D,)-(D,, +D.e)

As in Eq. 4-5, the denominator in Eq. 4-6 is positive, assuming well-behaved demand and

supply functions. The numerator term associated with aSg a v in Eq. 4-6 is negative,

and the combined effect associated with 8S /8 v is positive. This result clearly shows

that any nonprice incentives that increase green supply and decrease conventional supply

will reduce the quantity of conventional production. In other words, programs that

disseminate information about cost-reducing methods of green production or provide

other types of nonprice incentives lead to an unambiguous reduction in conventional


Results generated by the two-product, partial-equilibrium model are summarized in

Table 4-1. An increase in eco-demand is not the best indication of an eco-labels

effectiveness at reducing environmental impacts associated with production. In fact,

when the increase in farm-gate green demand (after accounting for marketing costs)

exceeds any decrease in farm-gate demand for the undifferentiated product, conventional

production and environmental harm may rise. Under the assumption that negative

environmental impacts are positively related to conventional production quantities,

decreases in demand for the undifferentiated product and nonprice incentives that

encourage conventional growers to adopt green practices are most effective at protecting

the environment in the producing region.

Table 4-1. Two-product, partial-equilibrium model results
Initial impact Market conditions Effect on Qc
1st scenario: no excess green supply

loD /,lp = aD,, /ap Negative
DDo pp > > oD. /lapl
8De /Op > O
and SccDue > ScgDuu Ambiguous
D,, /ap < 0 SccDue < ScgDuu Negative
2nd scenario: excess green supply

laD, /ap = 9aD, l/pl No change
JaD, /lpl > aD, /ap\ Positive

1st scenario: no excess green supply

and aS, /a v J = as, /a V Negative
as, lav < o0 la, /I V > las, / vi
SceDue > ScgDuu Negative
SccDue < ScgDuu Ambiguous
2nd scenario: excess green supply Negative

1st scenario: no excess green supply Positive
9M /9p > 0 ^ s *---------- r -----
OM, /p > 0 2nd scenario: excess green supply Negative

Price-Endogenous Programming Model

In this section, a price-endogenous programming model is used to simulate the

effects of introducing an eco-label on land-use and the environment in a producing

region. Price-endogenous programming models allow simulation of market responses to

changes in policies or economic circumstances (McCarl and Spreen; Hazell and Norton).

In particular, they are useful for anticipating the impact of the introduction of new

technologies or market opportunities, for which historical data is not available (McCarl

and Spreen; Hazell and Norton).

Final demand schedules, input supply curves, and intermediate costs are necessary

for building a price-endogenous programming model. Input demands and output supplies

are implicit in the model. Typically in price-endogenous models, demand curves for final

commodities are specified based on elasticities estimated from historical market data. In

the case of introducing an eco-label, a new differentiated product, market data is not

available to estimate a demand function. Instead survey data can be used to provide

estimates of consumer response and demand for a new product.

Green Production -
Figure 4-3. Conceptual diagram of the price-endogenous mConsumers

Alternative Uses
for Factors

Figure 4-3. Conceptual diagram of the price-endogenous model

We consider the example of introducing an EMP certification and labeling program

for ornamental nursery plants, such as leatherleaf ferns, in Florida. In this model, land in

Florida's ornamental plant producing regions may be used for three mutually exclusive

alternatives: conventional fern production, "green" fern production (using EMPs), or a

next best alternative (e.g., housing development or another agricultural crop).

Conventional ferns are sold unlabeled (on the undifferentiated market). "Green" ferns

may be sold with an eco-label (on the eco-market), or without an eco-label (on the

undifferentiated market). A conceptual diagram of the sector is shown in Figure 4-3.

A first step in building the model would be to estimate relevant final demand

curves. We consider three types of consumers, similar to the types introduced in Chapter

2. Consumer Type A currently purchases Florida-grown ferns, but would not pay any

premium for an eco-label. Consumer Type B currently purchases Florida ferns and

would purchase them with an eco-label at some positive premium if she had the option.

Consumer Type C does not currently purchase Florida ferns, but would consider buying

them if they were eco-labeled. This last type of consumer might be in an export market

that requires an eco-label for importation. Consumer Type C could also represent a

serious environmentalist who avoids products that he perceives as harmful to the

environment. Demand from these three consumer groups could be segmented as follows.

Estimates of current demand (prior to introducing eco-label) for Florida ferns are

available from existing market data. Let the inverse form of this demand function be

denoted by Pz(Z), where Pz is the price of unlabeled ferns, and Z is the quantity of

unlabeled ferns sold. This demand schedule from consumer Types A and B represents

implicit demand for the private fern characteristic plus implicit demand for the

environmental impact (a negative value). This explanation is compatible with the theory

and consumer model presented in Chapter 2. The inverse demand for unlabeled ferns

corresponds to Eq. 2-22.

To estimate the new demand created by introducing an eco-label, consumer surveys

would be necessary. In particular, a contingent valuation spike model similar to the one

used by Jensen and Jakus would be appropriate. Targeted at current consumers of

Florida leatherleaf ferns, the survey would first identify those consumers willing to pay a

nonzero premium for ferns bearing the EMP eco-label (i.e., consumer Type B, referred to

as eco-consumers). Then a distribution function could be estimated that relates the level

of the price premium to the fraction of eco-consumers willing to purchase the eco-labeled

ferns at that price premium. If we unit demand is assumed (i.e., quantity units are

expressed as a fixed amount of ferns purchased by each consumer), the number of

consumers willing to purchase at a given premium is equal to the quantity of eco-labeled

ferns demanded (in the per-consumer units). Otherwise the distribution function could be

adjusted to account for variation in quantities purchased among eco-consumers. This

method of estimating demand for quality corresponds to Tirole's model (pp. 96-97)

presented in Chapter 2.

Based on our model presented in Chapter 2, it is known that consumers already

purchasing ferns at the market price will purchase the eco-labeled ferns if their marginal

utility from the eco-label attribute relative to their marginal utility of income exceeds the

price premium. For the consumer at the margin, indifferent between choosing the eco-

labeled and the unlabeled ferns at a given price premium, Eq. 2-24 must hold with

equality. This equation is restated here as

(4-7) PU = P.

Similar to Tirole's model, the distribution function, F(P/p), represents the portion of eco-

consumers unwilling to pay a price premium of size Pr, for the eco-attribute represented

by a particular label with a level of credibility represented by p. For a given eco-label (p

is fixed), the distribution function would relate demand to the price premium, Pr. The

price premium equals the difference between the eco-labeled price and the unlabeled

price of ferns (i.e., Pe P,). The distribution function is show in Figure 4-4.



Figure 4-4. Distribution function representing willingness to pay a premium for eco-label

The ordinary unit demand for the eco-attribute is given by

(4-8) Dl P= N1[ F

and the inverse demand is

(4-9) P,.(E)=pF- EJ

where Ne is the entire eco-consumer segment (those current fern consumers willing to pay

a nonzero premium for the eco-attribute), and E is the quantity of eco-labeled ferns

demanded at premium Pr. In this formulation, demand for the eco-attribute is expressed

as a function of the label's credibility represented by the credibility parameter p. If

inverse demand for the eco-attribute with perfect credibility is PE(E), then the inverse

demand function representing an eco-label with less than perfect credibility is

(4-10) P,(E)= pP,(E).

This demand function for the eco-characteristic could be used in a price-endogenous

programming model, together with the existing demand function for ferns without the

eco-label, P2(Z).

A novel aspect of the model presented here is that final demand (by consumer

Types A and B) is represented by demand for characteristics schedules, instead of

demand for final goods. This approach has practical advantages (allows incorporation of

demand for a new characteristic, estimated using survey data) and is consistent with the

branches of consumer theory described in Chapter 2. As described by Ladd and

Suvannunt (p. 504)

For each product consumed, the price paid by the consumer equals the sum of the
marginal monetary values of the product's characteristics; the marginal monetary
value of each characteristic equals the quantity of the characteristic obtained from
the marginal unit of the product consumed multiplied by the marginal implicit price
of the characteristic....

Ladd and Suvannunt (p. 505) state that "total consumption of each characteristic

can be expressed as a function of quantities of products consumed and of consumption

input-output coefficients." These statements suggest a model in which final demand for

characteristics (marginal implicit price schedules), such as PE(E), can be filled by

products (eco-labeled ferns) in proportion to the amount of a characteristic perceived in a

product (represented by the coefficient p). Equation 4-10 shows this relationship.

Certain aspects of this approach, such as possible interaction effects between

multiple characteristics bundled together in a single product, create difficulties. Also, the

demand specification presented above does not consider possible effects of changes in

price of the base characteristic on demand for the eco-attribute, and vice versa. If it is

assumed that utility is separable and of the Cobb-Douglas form, however, demand for

one characteristic would be independent of other characteristics (Silberberg and Suen). It

could be argued that substitution effects between characteristics would be weak or

negligible. Intuitively, there may be good reasons to assume that two closely related

products, such as eco-labeled ferns and unlabeled ferns, are substitutes. Substitution

effects between demand for characteristics (such as enjoyment of ornamental plants and

water quality), however, are intuitively weaker. Therefore, ignoring substitution effects

between these characteristics schedules does not appear to pose a major shortcoming.

Also, interaction effects between different characteristic bundles would be minimal, since

only two characteristics are considered in this model. Such interaction effects might

present a more significant problem, if several characteristics were contained in the model.

Finally, the potential demand among consumers not purchasing Florida ferns prior

to eco-labeling (consumer Type C) is considered. We call this consumer segment Market

2, to distinguish it from the demands estimated previously (Market 1). Market 2 could

represent an export market. Using survey data or other methods, the demand for eco-

labeled ferns (with the combined characteristics Z and E) could be estimated for Market

2. In inverse form, this demand function is represented by P2(ZE2). We assume that the

demand function in Market 2 is independent of prices in Market 1.

These three demand schedules enter the objective function of the quadratic

program. Other value and cost schedules in the objective function include, a value

function representing demand for land in its next best alternative use (to fern production),

per acre production costs, and marketing costs. Land is a shared input required for both

types of fern production and the next best alternative. The variables, Xc, XG, XA, are the

acreage employed in conventional fern production, "green" fern production, and the

alternative land use respectively. The alternative land value, represented by VA(XA), is a

decreasing function of the amount of land allocated to the alternative uses. Hypothetical

direct per acre production cost schedules, Cc(Xc) and CG(XG), are formulated for

conventional and "green" ferns. These cost schedules are increasing functions of the

amount of land allocated to each activity. Direct inputs are assumed unique to each

production method, allowing the direct cost schedule for each of the production methods

to be independent of production quantities of the other. This simplifying assumption is

fairly strong, since in reality conventional production shares many of the same inputs

with production using EMPs. Per unit marketing costs for each market, C1 and C2,

include all intermediate costs between production and the final consumer. Per unit

marketing costs are assumed constant. Additional to the standard marketing costs,

certification and labeling costs must be incurred for products sold on the eco-market.

These costs are represented by CEIYGEI and CE2YGE2, where CE is a per unit cost and YGE

is the quantity of green production allocated to eco-labeling for each market. It is

assumed that CE includes both c, (certification costs) and ce (eco-label marketing costs)

described in the producer model in Chapter 3. The objective function of the price-

endogenous programming model simulates a competitive equilibrium:

Max JP,,(Z,)dZ, +p ,(E,)dE, +L2 fP2(ZE,)dZE2 + f,(XA)dXA
C,(X,()dX. fC;C(X,)dX, -C,Z, -C2ZE, -L,C.,Y,,, L2C,,2YE2

where Li and L2 are zero-one variables indicating whether an eco-label has been

introduced to Market 1 or 2 respectively. The set of constraints for the price-endogenous

programming model is

X, +X, +X < X
Y,. y,X(, <0
Y,- YX,,<0
Y(;l + y,,, + YX 2 Y< ; < 0
(4-12) + 2 G Y 0
z, Y,, YGEI ( 0
E, Yc;1, < 0
ZE2 Y,2 5 0
EI.X,. + EI,;X, + EIXA EIl

Constraints require that land allocated to the three alternatives is less than or equal

to the total land available in the producing region, X. Conventional fern output, Yc, must

be less than or equal to the number of acres in conventional production times the average

per acre yield, yc. The same constraint applies to green fern output and per acre yield, YG

and yg. Green output may be divided between green-unlabeled, YGu, eco-labeled product

on Market 1, YEI, and eco-labeled product on market 2, YE2. Current fern demand, Zi, is

filled by the sum of conventional output, green unlabeled output, and eco-labeled output

placed on Market 1. Demand for the eco-attribute in Market 1 can only be filled by eco-

labeled product sent to Market 1, and demand for eco-labeled ferns on market 2 is filled

by green, eco-labeled product sent to Market 2. Additionally, an accounting row is added

that calculates the total environmental impact, EIr, (e.g., an index of effluent run-off or

water quality impacts) as a function of the amount of land in each activity and the per

acre environmental impacts associated with the three alternatives, Elc, EIG, and EIA.

Differentiating the Lagrangian for the model, one obtains market equilibrium

conditions. Assuming Xc, XG, XA, Zi, El > 0 and ZE2 = 0, a necessary condition for land

allocation equilibrium is

(4-13) VA = y (Pz, C) C. = (Pz, -C, + pP,, -C,,)- CG

Land is allocated so that the per acre value obtained from conventional ferns sold

unlabeled equals the per-acre value obtained from green ferns sold with an eco-label.

Land values from either type of fern production equal the value of an acre of land in its

next best alternative use, VA. The introduction of an eco-label, represented by an increase

in the parameter p, increases the effective demand for green ferns and the value of an acre

of land producing ferns using EMPs. In the long-run, green acreage increases, while

conventional fern acreage and alternative land-uses decline until equilibrium is restored.

Assuming XG, ZI, El > 0, and green ferns are sold both eco-labeled and unlabeled,

market equilibrium for green fern output requires that

(4-14) VA + = y,(P -C, + pP,, -C,,) = y (P, C,).

The opportunity cost of land plus the direct per acre production cost using green methods

equals the value of an acre's yield sold either labeled or unlabeled. This scenario implies

that all the "eco-label rents" are exhausted, and PPEI = CEI. Only if the price premium no

longer exceeds the additional marketing costs associated with certification and labeling,

and conventional ferns are no cheaper to produce, will ferns produced using EMPs be

sold unlabeled.

Producers will only sell ferns with an eco-label, if the following price equilibrium

condition holds:

(4-15) Pz + pP,, = --(VA +CJ)+C, +CE,

The left-hand side of Eq. 4-15 is the sum of the implicit characteristic values each

multiplied by its associated coefficient. This sum equals the market price for eco-labeled

ferns. This result is consistent with the consumer models described by Gorman, Michael

and Becker, Ladd and Suvannunt, and Stigler and Becker, and is similar to Eqs. 2-2, 2-8,

2-11, and 2-14. The right-hand side of Eq. 4-15 is the sum of all costs associated with

producing and marketing a unit of eco-labeled ferns.

A complete set of necessary first-order conditions can be generated for the price -

endogenous programming model under the two different scenarios considered for the

previous model. If green supply is fully absorbed by green demand (i.e., no green output

is sold on the undifferentiated market), the equilibrium conditions are

yPz, (Z,) yC, Cc (X,.) V, (X, ) = 0
gPz,(Z,)-yC, -C,;(X,,)-V(X )+ pygPEI(E)--yC -0
L2g P2 (ZE2)- yC2 -C, (X(,)- V(XA) 0
X +X. +X,; -X-O
Z, +ZE2 ygX yCX. 0
E + ZE2 yX, 0O

In Eq. 4-16, eco-labeling costs associated with Market 2 are included in C2. When eco-

labeling does not provide access to a new consumer market (Market 2), L2 = 0, ZE2 = 0,

and the third identity in Eq. 4-16 does not apply. If green supply exceeds green demand

at given prices (i.e., some green output is sold on the undifferentiated market), the

equilibrium conditions are

ycPZ1 (Zi) yC, C, (Xc) -VA (XA) -0
ygz I (Z,) y C, C, (X,) V (X ) -0
yPz, (Z,) y C, C (X) VA(X ) + pygPF, (E,) ygCE, 0
(4-17) L2ygP2(ZE2)- gC2 -C( (X(;)-VA (XA)
XA +XC +X -X=O
Z +ZE, ygX, y,Xc 0
E +ZE2 + Yc6 ygX; = 0

Again, eco-labeling costs for Market 2 are included in C2. When eco-labeling does not

provide access to a new consumer market (Market 2), L2 = 0, ZE2 = 0, and the fourth

identity in Eq. 4-17 drops out.

Comparative static analysis can be conducted to identify the direction of change in

the land-use variables Xc, XG, and XA resulting from changes in the parameters p and L2,

or from changes in cost parameters. Instead, however, we create a parameterized version

of the model using General Algebraic Modeling System (GAMS) software and run the

model to simulate effects under different market conditions. The GAMS program shown

in Appendix B is calibrated so that p = 0.5 and Market 2 is not accessible. The value ofp

is built into the shifter parameter on the demand for El. The shift parameters adjust the

units, so that demand quantities are expressed in millions of boxes and land units are in

thousands of acres.

As with the two-product, partial equilibrium model, two different initial market

scenarios are considered. The first scenario has green production small relative to

conventional production. The second scenario has green production acreage large

relative to conventional acreage, and excess green supply is sold undifferentiated. Two

different cases are considered in terms of the relative environmental impacts of the three

land-use alternatives. In the first case, the next best alternative land-use is