the economics of labeling: an overview of issues for...

The Economics of Labeling: AnOverview of Issues for Health andEnvironmental Disclosure

Mario F. Teisl and Brian Roe

During the last two decades, product labeling has become an increasingly used policy tool,

particularly with respect to the provision of health and environmerrtd information. Theory

holds that the flow of information among market participants plays a critical role in the

efficient operation of markets. This paper explores the role of product labeling policy in

ameliorating two potential market deficiencies: asymmetric information and costly search

behavior, Practical considerations for the design and implementation of labeling policy and of

labeling research are explored,

During the last two decades, product labeling hasbecome an increasingly used policy tool, particu-larly with respect to the provision of health andenvironmental information. As a result, product-labeling policy is a topic of growing interest andpublic debate at both the federal and state levels.An example at the federal level includes USDA’scurrent rulemaking on organic labeling of foods.Examples at the state level include measures toimpose price and environmental labeling of elec-tricity, emissions labeling of automobiles, and in-gredient labeling of cigarettes. The labeling debateis largely about information and the processing anduse of that information by consumers. The debatecenters on questions such as how much informa-tion to supply to consumers to facilitate effectivechoice and how that information should be sup-plied.

A tenet of economic theory holds that the flowof information among market participants plays acritical role in the efficient operation of markets.However, when the flow of information is associ-ated with explicit costs (e.g., costly search [Dia-mond 197 1], cost] y revelation of arbitrage oppor-tunities [Grossman and Stiglitz 1980]) or implicitcosts (adverse selection or moral hazard [Akerloff1970]), markets can lose the neoclassical promise

Mario F. Teisl is an assistant professor, Department of Resource Eco-nomics and Policy, University of Maine, Orono. Brian Roe is an assistmrtprofessor, Department of Agricultural, Environmental and DevelopmentEconomics, Ohio State University, Columbus. This manuscript was writ-ten while Roe was a staff fellow with the Center for Food Safety andApplied Nutrition, U.S. Food and Drug Administration, Washington,D.C. Maine Agriculture and Forest Experiment Station Publication num-ber 2242,

of efficiency (Stiglitz 1996). In this paper, we areinterested in exploring the role of informationallabeling in the operation of consumer product mar-kets. We present an economic framework thatmight be used to estimate the economic benefitsassociated with labeling policies and explore aplethora of practical considerations that must beaddressed before labeling policies are imple-mented. Throughout the paper we turn to both eco-nomic and psychological theory to explore severalof these implementation issues.

Labeling Defined

We begin our discussion of labeling issues by firstdefining what we mean by labeling or a labelingprogram. We define product labeling as any policyinstrument of a government or other third partythat somehow regulates the presentation of prod-uct-specific information to consumers. This infor-mation might describe use characteristics of theproduct, such as price, taste, and nutrition, or non-use characteristics, such as the environmental im-pact or moral/ethical elements surrounding theproduct’s manufacturing process,

Labeling policy can differ along three majorcontinua: compulsoriness, explicitness, and stan-dardization. First, a labeling policy can vary in itsdegree of compulsoriness, the degree to whichfirms are required to provide product information.At one extreme, labeling restrictions are manda-tory: certain pieces of information are required tobe displayed on the product. At the other extreme,

Teisl and Roe Economics of Labeling 141

labeling restrictions are voluntary: firms chosewhat information, if any, will be displayed. Mostthird-party certification programs (e.g., the GreenSeal, Good Housekeeping Seal) fall into the vol-untary category. An intermediate example isclaims-based labeling, wherein firms are requiredto disclose particular items of information, often ina specified format, only when a certain type ofclaim is made about the product elsewhere on thelabel or in product advertising. Under these typesof labeling policies, firms have at least some con-trol over the information presented on their prod-uct.

The second major component of labeling poli-cies is explicitness, the degree of information detailpresented to consumers. Current ISO 14000 nego-tiations provide a useful nomenclature here. ISO14000 identifies two types of labels (Type I andIII) that are differentiated by the level of informa-tion detail. 1 Type I labels provide the least amountof detail concerning attribute values. With a TypeI label, the information about a vector of attributelevels is condensed into a one-dimensional scoreby an agreed-upon scoring algorithm. Products re-ceiving a score above a predetermined thresholdmay present a seal of approval or certification. Atthe other extreme are Type III labels, which pro-vide the most detailed information. Information isdisclosed about several of the products attributes(e.g., nutrition labels on food), and the disclosuretypically involves continuous or categorical infor-mation about each element (e.g., grams of fat,high/mediurn/low risk). Type 111labels are gener-ally considered the most objective of the label cat-egories, while Type I labels are often consideredthe most normative.

The third major component of labeling policiesis standardization, the degree to which the regula-tion requires the information to be provided in apresentation format that is standardized and uni-form across products. At one extreme, a labelingpolicy can make presentation format requirementsquite explicit, where the firm has no discretionover the presentation. For example, warning labelson cigarettes and alcoholic beverages have word-ing, font size, font typeface, and message locationprescribed by regulation. Alternatively, the contentof the information may be regulated but the firmhas some discretion over the presentation of theinformation (e.g., health claims on food labels).

Justifications for Labeling

To economists, an obvious next question is <‘Whatis the economic justification for labeling pro-

grams?” As we pointed out in the introduction, theefficiency of markets is eroded as the flow of in-formation among market participants is impeded oras information flow becomes costly.

Simply stated, labeling policies can circumventthese market inefficiencies by making the informa-tion initially held by the firm also available to theconsumer, This removal of information asymmetryor subsidization of search costs is clearly beneficialto consumers as they are now more informed as tothe exact attributes of the product. Choices will bemore closely in line with preferences, and uncer-tainty regarding the nature of product attributes isminimized during the choice process. Firms thatproduce goods with desirable attributes also gain,as they are rewarded for marginal improvements inthe quality of various attributes, Even firms sellinglow-quality products may gain. Consider a case ofsevere adverse selection (e.g., Akerloff 1970)where consumers know only the average quality ofgoods and have no other means to infer quality. Insuch a case, labeling of product attributes is onemethod of creating a market that would otherwisecollapse because of adverse selection. Hence firmsselling low-quality goods will be able to sell goodsand benefit from the existence of the market.z

However, the costs and benefits of labeling arelikely to depend upon the type of attributes con-sidered. Caswell and Mojduszka (1996) outlinethree types of product attributes: search, experi-ence, and credence.3 Search attributes are thosethat can be assessed prior to purchase via researchand inspection (e.g., color and size). Experienceattributes are those that are assessed only after pur-chase and use (e.g., convenience). Credence attrib-utes are those that cannot be easily verified evenafter uurchase and use but whose value effects util-hy (e~g., the nutritional component of food).

Caswell and Mojduszka suggest that labelingplays an increasingly beneficial role as attributesprogress along the spectrum from search to expe-rience to credence.4 The movement along this spec-trum is roughly correlated to the cost or difficulty anindividurd would face while trying to independentlyovercome the information asymmetry. For example,identifying the color and size of a certain object(search attributes) can be done rather quickly orcheaply by most individuals, while verifying that afirm used 75% postconsumer recycled material tocreate the package (credence attribute) would betremendously difficult. In fact, the individual mayhave no legal authority to undertake such an inves-tigation.

Conversely, as one progresses along the spec-trum of attributes from search to credence, thecomparative advantage of a government or other

142 October 1998 Agricultural and Resource Economics Review

third-party organization in providing the informa-tion services becomes more obvious. One obviousadvantage is that such an organization has econo-mies of scale in verifying, monitoring, and dis-seminating information. Another advantage is thatsuch organizations have means to require the re-lease of the asymmetric information and to penal-ize firms that they find to be misrepresenting suchinformation. For example, governmental organiza-tions may use force of law to require disclosure orto penalize deceptive firms, while third-party or-ganizations may have economies of scale in harm-ing deceptive firms’ reputations by organizing pro-tests or coordinating lawsuits.

However, labeling is not costless, There is oftenthe fear that labeling will cause “excess inertia” or“lock-in” in the Farrel and Saloner (1985) sense.That is, for a particular product, a labeling regimemay include only certain attributes or may use acertain criterion for the awarding of a seal of ap-proval. If there are changes in the marketplace,with regards to either technology or consumerpreferences, appropriate changes in the labelingscheme may be greatly lagged because of institu-tional bureaucracies and coordination difficulties.Also, there are the obvious costs of gathering, veri-fying, and monitoring the needed information andadministrating the labeling program. These costsare likely to be positively correlated to the privatecosts mentioned above.

Given these cost and benefit considerations, onemight imagine that labeling policies are least likelyto be justifiable for search attributes. Often it isassum~d that the market communications suppliedby firms through voluntary labeling and advertis-ing will provide a sufficient degree of informationabout search attributes to make the market workefficiently to the benefit of all participants. Even inthe case of search attributes, however, labelingpolicies may offer benefits often overlooked in tra-ditional economic analyses. We argue that withoutstandardization of information disclosure, somesearch attributes effectively resemble experienceor credence attributes. This occurs because certainsearch attributes, such as the pricing structure forsome services (e.g., phone and electricity), are verycomplex and difficult to judge ex ante. We viewconsumers as boundedly rational (Simon 1955;Conlisk 1996): they face not only budget con-straints during the decision process but also timeand cognitive constraints. Consumers facing theseconstraints may optimally make decisions based onrelatively poor information about the search attrib-utes. The true nature of these search attributes maynot become evident until after repeated use of theproduct (i.e., until one receives several representa-

tive bills), and the true ranking of products in termsof this attribute may never become obvious.

Consider the most fundamental of search attrib-utes—product price. It is often implicitly assumedthat firms reveal the product’s price and that con-sumers understand it with precision and appropri-ately incorporate it during the decision-makingprocess. However, consumers must use scarce timeand cognitive resources to rank products or ser-vices on price and must determine price differ-ences so that marginal tradeoffs with other attrib-utes can be made. If different firms communicateprice via different disclosure structures, compari-sons among competing firms rapidly becomes dif-ficult, For example, Estelami (1997) showed con-sumers’ inaccuracies in judging the price of asingle product increased as the complexity of thepricing scheme increased. From the unit pricingliterature, Russo (1977) showed that consumers’ability to find the “best deal” was significantlyhindered when only one price existed but productsizes varied. Even though formats were identical,information was still difficult to compare becausethe unit of comparison was nonstandard.

Psychological studies provide some insight tothe comparison difficulties that arise when con-sumers encounter nonstandardized information.Kleinmuntz and Schkade ( 1993) outlined three ma-jor components of informational displays that af-fect the processing tactics consumers employ and,hence, the time and cognitive costs incurred andthe accuracy derived from these efforts. Thesecomponents are form (e. g., numeric/graphic/verbal), cross-product organization (matrix, para-graph, hierarchical), and sequence. Schkade andKleinmuntz (1994) reported experimental resultssuggesting that all three elements affect the diffi-culty and accuracy of judgment, but the largesteffect was due to cross-product organization. Thissuggests that standardizing the display across prod-ucts would provide the largest benefit to consum-ers, while improving the sequence and actual dis-play format is next. Coupey (1994) provided addi-tional insight to this finding. She found that, whenconsumers are faced with information that is notstandardized across products, they often take pro-cessing shortcuts, such as eliminating certain at-tributes, eliminating certain products, and roundingnumbers. Hence, standardizing information formatacross competing products may increase the num-ber of products or attributes considered duringchoice and may provide more precision intradeoffs made by the consumer.

Other research confirms that standardization ofthe label format can reduce the cognitive costs ofextracting information, thus facilitating the con-

Teisl and Roe

sumer’s primary uses of a product label, i.e., tomake cross-product comparisons of attributes andto verify firm-provided claims made elsewhere onthe label. For example, when Winneg et al, (1998)asked consumers to compare two hypotheticalelectricity services in terms of price, 82910couldcorrectly identify the service that cost five dollarsless per month when both products displayed astandardized format, while only 6090 could do sowhen the products used different formats.

Beyond standardization, labels may help im-prove the credibility of firms’ privately sponsoredcommunications. Often consumers are highlyskeptical of the honesty of firms’ advertising (Cal-fee and Ringold 1988); the existence of labelingallows consumers to verify claims made by adver-tisers and may hinder some firms from overstatingproduct qualifications.

It is unclear whether individual firms would everfind it privately beneficial to organize and displayproduct information in a standardized format thatwould also benefit consumers.5 Disincentives to doso may stem from a fear that standardized pricedisplays would promote price competition in mar-kets that were previously monopolistically com-petitive, hence driving down profit margins. Thereare also the more straightforward costs of coordi-nation that must be overcome by all the partici-pants. These include both the logistical costs andthe costs individual firms expend while lobbyingfor disclosure formats that highlight their product’scomparative advantage, In some cases, such as theproducts reviewed in Consumer Reports, marketopportunities arise and a third party may profit byproviding such standardized information. But foritems such as household electric service, whichvaries greatly from region to region, the marketmay not offer such opportunities.~

Determining the Effectiveness of Labeling

How can economists contribute to the design andimplementation of labeling programs? This ques-tion can be answered in two ways, depending onthe audience. Among policy makers the questionusually means “How can economists help in de-signing labeling programs that effectively movecurrent consumer behavior toward some target be-havior?” Among economists the question appearsas “How can economists help in designing label-ing programs to maximize net social benefit?”We’ll discuss both of these questions, starting withthe first.

While labeling has been the focus of majorpolicy initiatives in the last few years, little em-

Economics of Lubeling 143

pirical economic research has attempted to under-stand the market effects of different labeling poli-cies.’ Presumably, one reason for this lack of re-search is that policy makers have only recentlyshown interest in labeling programs. However, an-other potential reason is the relative difficulty inisolating the economic consequences of changes ininformation policy.

Most empirical studies have basically used oneof three methods to identify the behavioral effectsof changes in information:

1.

2.

3.

They use time-series data to estimate the de-mand for a commodity or a group of com-modities as a function of some index of in-formation.They estimate demand as a function of indi-vidual awareness or knowledge.They assume demand shifts are due to a Dar-ticuiar information change and use dumm~ed-trend variables in the demand specification todenote the change in information.s

A problem with the first approach is that usingan information index as a determinant of marketbehavior implicitly assumes information translatesrelatively cleanly into consumer awareness andconcern. Psychology and marketing studies indi-cate that this assumption is often tenuous. Al-though studies that incorporate health-relatedawareness are valuable in linking changes in de-mand with changes in awareness, they are lessvaluable in linking changes in demand with policy-relevant instruments. Finally, the use of trend vari-ables is problematic because trend variables, ag-glomerating all time-varying factors, do not allowidentification and measurement of changes in de-mand due to changes in information. This isparticularly troublesome because label-inducedmarket changes may take months or years beforesome consumers notice or incorporate the newinformation (e.g., Levy et al. 1985; Levy andStokes 1987; Schucker et al. 1983; Teisl, Roe, andHicks, n,d.).

Ultimately, though, if economists are to makesignificant contributions to the design of labelingpolicies, the research question is more than wheth-er labeling programs can work, for there is alreadyresearch that indicates that labeling can make sig-nificant changes in both consumer behavior (e.g.,Teisl and Levy 1997; Levy et al, 1985; Ippolito andMathios 1990, 1996) and producer behavior (e.g.,Frazao and Allshouse 1996). What is needed isresearch that develops understanding of what theconditions need to be for a labeling policy to beeffective. That is, what characteristics of the inter-


action between the label, the consumer, and theproduct affect the impact of information?

For example, the effectiveness of environmentallabeling is partially dependent upon the altruisticpreferences of consumers and their knowledge ofpathways to satisfy their preferences. Furthermore,the degree of altruism exhibited by an individualmay depend on how information concerning anenvironmental attribute is presented, suggestingthat different wording approaches may induce dif-ferent warm-glow effects (Andreoni 1995),9 In theenvironmental arena, this opens the debate to Sci-tovsky’s (1986) delicate question—can resourcecosts be reduced without reducing consumer satis-faction? In our context we ask another question—might labels shape preferences by framing the in-formation so that consumers give additional weightto attributes associated with environmental costs?What little market evidence there is does not sup-port this idea; label information does not seem toalter consumer behavior in terms of gross “con-sumption” of product attributes but may simplyallow consumers an increased ability to substituteacross products (Teisl and Levy 1997). Thus, la-bels may increase consumer welfare but may notfulfill the goals of health and environmental poli-cymakers.

The effectiveness of labeling efforts may alsodepend on the product category. For example, Stra-hilevitz and Myers (1998) find that a firm’s offer tomake a charitable contribution upon product pur-chase is more effective with products classified as“frivolous luxuries” (e.g., ice cream sundaes) thanwith “practical necessities” (e.g., detergent). It isunclear whether attribute labeling, which may beviewed as more objective and less persuasive,would also be product sensitive,

This lack of knowledge regarding the marketeffectiveness of labeling policy characteristics isparticularly evident, and potentially troublesome,with respect to environmental labeling. Environ-mental labeling programs are widespread; govern-ment and nongovernment organizations haveimplemented various environmental labeling pro-grams that cover thousands of products in morethan twenty countries (U.S, EPA 1993), Theirwidespread use suggests that environmental label-ing is perceived as an effective method of alteringconsumer and producer behavior; however, re-search concernirw its effectiveness is limited andquantitative resu~s are rare. Much of the researchhas measured effectiveness either by identifyingchanges in consumer awareness after exposure tolabel information (Hartwell and Bergkamp 1992;Hashizume 1992; Yakami 1992) or by asking con-sumers whether labeling programs would affect

their purchase behavior (Chase and Smith 1992).However, a change in awareness does not neces-sarily translate into a change in behavior (Bonne-ville Power Administration 1985; U.S. EPA 1989),and consumers do not necessarily follow their ownpurchasing assertions (Bailey and Eastlick 1993;Gutfeild 1991). Market-based research investigat-ing the effectiveness of other types of labels (e.g.,nutrition labels) may not be applicable to environ-mental labeling because these other labeling pro-grams provide information about the use charac-teristics of the product. Environmental labels oftendifferentiate products with respect to non-use char-acteristics.

Environmental certification programs providean example of the lack of knowledge surroundingthe market effectiveness of different labeling char-acteristics. Currently, firms are spending substan-tial amounts of money and are altering productionmethods to obtain environmental certification la-bels, and organizations like the World Bank sup-port the use of these Type I labeling programs.However, focus group research (Teisl, Halverson,and Holt 1997) indicates that consumers may reactnegatively to these Type I environmental labels,and experimental research (Winneg et al. 1998)indicates that these labels may be ineffective inaltering consumer behavior. There seem to be twopossible reasons for the counterintuitive results.First, consumers may view Type I labels as toovalue laden. Consumers seem to prefer the moredetailed Type III labels, which allow them moreflexibility in applying their own value judgments,The second reason consumers may not like Type Ilabels is that they may doubt the veracity of thecertifying organization. That is, consumers may in-voke a <‘schemer’s schema” (Wright 1986) inwhich they try to guess the true motivation behindthe communication effort. Consumers may viewthe certifying organization as just another market-ing gimmick employed by the firm and may there-fore disregard the information.

The Welfare Effects of Labe[ing Policy

Now we turn to the economists’ version of ourearlier question—how can economists contributeto the design of labeling programs to maximize netsocial benefit? In application, this question pro-vides some interesting welfare and equity issues.For example, consumer research indicates thatstandardizing the presentation of information canreduce the cognitive costs of information process-ing. However, we also know that consumers varyin their ability to process information. As a result,


one question facing makers of information policyis the level of standardization that should be im-posed. Policymakers cannot provide labels that sat-isfy everyone’s information desires while simulta-neously catering to consumers’ cognitive and timeconstraints. As a result, policymakers need to un-derstand how different information policies affectdifferent sectors of the consumer population.

Economists, using cost/benefit analysis, may beable to provide a framework to help answer suchquestions. However, relatively little work has beendone to develop a method of benefit analysis forinformation changes. For example, current prac-tices in valuing the economic benefits of nutritionlabel changes take the position that the value of alabel change should be judged by the degree towhich it provides “nutritionally correct” behavior(e.g., Zarkin et al. 1991). This perspective, whichpresumes that the “correct behavior” can be pre-judged by the researcher, is atypical for econo-mists, who usually assume that the individual is thebest judge of her/his well-being.

The empirical literature validates potential be-havior changes in the face of label-provided infor-mation. Therefore, consumers prevented frommaking such adjustments by lack of informationwill be worse off. However, conceptualizing thewelfare effect of improved information is some-what perplexing. If one attempts to measure thewelfare effect of information change by usingchanges in the consumer surpluses behind affecteddemand curves, then a paradox results. Consider agood, Z, whose production is found to decreaseenvironmental quality. Dissemination of informa-tion on this link may lead to backward shifts in thedemand for Z. The consumer surplus associatedwith the consumption of Z shrinks and, using thismeasure, the consumer appears to be worse offwith the information than helshe was without it.

Foster and Just (1989) address this problem: ce-teris paribus, individuals are viewed as worse off ifinformation exists that would help them make bet-ter choices but, because they are ignorant of theinformation, they cannot optimally adjust their be-havior. When comparing utility and behavior undertwo states of information, the choice made withless information can be viewed as restricted againstthe context of better information. While individu-als make optimal decisions conditional on the in-formation set available, utility differences underdifferent information states can be measured as thedifference between the optimal decision under bet-ter information and the restricted decision underbetter information (where the restricted decision isdefined as the optimal decision under poorer infor-mation). To measure the welfare effects of provid-

ing information, one must assume that the indi-vidual eventually obtains better information anddoes not remain in ignorance. Ultimately, welfareeffects must be measured against the correct infor-mation; ignorance is bliss only if the individualnever learns the correct information.

To provide a modeling framework to measurechanges in consumer behavior and welfare due tochanges in label information, one needs to knowhow attribute information enters an individual’sutility function (here defined in terms of a purchaseoccasion or decision). The utility evaluation can berepresented by the indirect utility function

Vs = V(AS, q, Y, p),

where As denotes a vector of attribute-related qual-ity assessments fgr m products given informationset S (i.e., As = Y(A~, . . . A;), q denotes a vectorof other (search) quality characteristics (e.g., tasteor texture), p is a corresponding vector of prices, andY denotes income. Vs is increasing in q and Y, de-creasing in p.

The technology that extracts and translates labelinformation into an assessment of a product’s at-tribute-related level of quality can be viewed as a“household production” process by which an in-dividual combines her prior knowledge, cognitiveabilities, time, and the information presented dur-ing the purchase decision. Thus, we could modelthe assessment process during the purchase deci-sion as

A: = f(SJ, G, $; 6),

where A; denotes the (subjectively) assessed at-tribute-related quality level of good j given infor-mation set S, Sj is the attribute information dis-played about product j at the point of purchase, Gdenotes the consumer’s prior stock of related in-formation (which may include information fromnews accounts or firm-provided advertising), and ~denotes the time that the individual devotes to pro-cessing Sj.

The objective level of the attribute representedby the information variable S is denoted by 6. Forexample, if S represents a dolphin-safe claim on acanned tuna label, then (3denotes that the produc-tion of the tuna led to no actual dolphin deaths. 8is separate from the assessment function becausethe individual does not observe it at the time ofpurchase except through the variable S. Although 0may be unobservable to the consumer at the timeof the purchase decision, we include it within thediscussion to distinguish between a factor that af-fects consumer decisions, S, and one that causesany health or environmental impacts, 0.

146 October 1998

We begin by presenting expenditure functionsunder two alternative states of information. Denote

e(AO(.), q, U, p)

as the expenditure function when the product at-tribute is at level O and the information about theattribute accurately reflects that level (S = O andreflects the state of 0). Likewise,

e(A1(.), q, U, p)

is the expenditure function when the product at-tribute is at level 1 and the information about theattribute reflects the new level (S = 1 and reflectsthe state of (3).

Although not commonly framed as such, com-pensating variation (CV) measures the change inindividual welfare when the qualities of a goodchange (denoted as a move from 8 = 0° to 6 = 01)along with a corresponding change in informationabout the quality change (S = O to S = 1). CV canbe expressed as

CV = e(A1(.), UO, p“) – e (AO(.), UO, p“),

where UO denotes the initial utility level (q isdropped for simplicity).

CV is an appropriate welfare measure when agood’s attribute changes and the individual is pro-vided with information reflecting the change.However, CV is not appropriate when a good’sattribute does not change but the individual is pro-vided with new information so that she adjustsconsumption to avoid possible welfare losses (dueto a decrease in assessed quaIity) or to obtain pos-sible welfare gains (due to an increase in assessedquality). Foster and Just (1989) suggest that a use-ful welfare measure in this case is the cost of ig-norance.

The cost of ignorance (COI) measures thechange in individual welfare when the quality of agood does not change but the information about thegood’s quality changes. Following Foster and Just,the consumer’s choice without new informationcan be viewed as being restricted in terms of theallowed choice of x (the elements of the vector xmeasure the quantities purchased of different prod-ucts). Foster and Just note that COI can be definedas the difference between a restricted and an unre-stricted expenditure function

COI = e (p”, UO, Al(.)1 X“)) - e(p”, UO, A1(.)),

where s(.) is a restricted expenditure function be-cause the individual’s consumption is restricted tothe original bundle, X“ (that is, the quantities ofgoods chosen under the old information set). Notethat the COI measure assumes that the individualeventually obtains better information. That is, the

Agricultural and Resource Economics Review

restricted expenditure function reflects the situa-tion where the individual has obtained the betterinformation but the individual’s choice is restrictedto what it was when the consumer was ignorant.

COI can be measured by finding a price vector,pl, such that the point represented by the restrictedchoice (given p“) is represented by an unrestrictedchoice (given pi). Using this approach, COI can beredefined as

COI = e(p”, UO, A’(.)) - e(pl, UO, A1(.))+ (pl – p“) X“,

where pl is the price vector such that the originalconsumption bundle, X“, would be freely chosen.Note that COI (for a normal good) is alwaysgreater than or equal to zero; COI denotes theamount of money an individual is willing to giveup to gain better information about (3so that she isfree to alter consumption.

Foster and Just’s (1989) approach provides avaluable method of estimating the welfare effectsof changes in information about product attributes,However, there is a lot yet to learn. First of all, toour knowledge, Foster and Just’s paper is the onlypublished attempt at empirically measuring thewelfare effects of information change. However,their application was limited to information pro-vided by the news media and focused on one prod-uct. How to determine the welfare effects of label-ing when the program may cover a wide variety ofproducts is less clear. Furthermore, possible differ-ences in consumer welfare due to different con-sumer, label, or product characteristics have notbeen studied.

Interactions with Other Policy Instruments

Compared with command-and-control options orperformance standards that directly affect the at-tributes firms produce, clear labeling of attributesmay m-ovide a-less intrusive and often less exuen-. .

‘ 10sive method of improving market outcomes.Thus, a costfbenefit analysis may support a label-ing policy to promote environmental or healthgoals, whereas another type of policy may not passthe cost/benefit test. However, given that environ-mental labeling may focus on attributes that havepublic good aspects, the potential for free-ridershipin product choice may indicate that a labeling pro-gram by itself may not provide the socially optimallevel of tmblic goods. Because it is unlikelv thatlabeling ‘will re~lace other policy instrume~ts, apotentially rewarding avenue for economic re-search is to determine the theoretical and empirical


interactions between labeling programs and otherpolicy instruments.

Such interaction between labeling and otherpolicy instruments is at least partially predicatedupon the existence of altruistic preferences by con-sumers. Johansson (1997) shows that the sociallyoptimal externality-correcting tax may be less than,greater than, or equal to the standard Pigouvian taxfor cases in which agents have altruistic utilityfunctions. The direction and magnitude of this de-viation critically depends upon the type of altruismexhibited by the agents, 11For example, Johanssonshows that the optimal Pigouvian tax is still so-cially optimal in the presence of impure altruism(Andreoni 1990). Hence, the optimal policy pre-scription requires knowing how many agents ex-hibit each type of altruism-a very difficult (if notimpossible) task, both methodologically and politi-cally.

Dynamic Implications of Labeling onR&D Investment

A last area of potential research interest is in termsof the dynamic implications of labeling programs.That is, how do labeling programs affect firm in-vestment in the development of new products? Re-search indicates that nutrient and health-related la-beling of food products has made significantchanges in the supply of new products by provid-ing firms an incentive to improve the nutritionalquality of those products (Ippolito and Mathios1996; Levy and Stokes 1987; Frazao andAIIshouse 1996). Similar supply-side changes maybe expected to result from the issuance of the neworganic standard, However, the effects of setting oraltering labeling standards on industry researchand development expenditures are difficult togauge because of the many subtle and potentiallycontradicting incentives that may exist.

Consider a product category in which no itemcurrently meets the criteria to qualify for a particu-lar certification. If substantial rewards would go tothe first firm to develop a product reformulationthat meets these criteria and qualifies for the cer-tification, then substantial research and develop-ment dollars might be devoted to producing such afacilitating technology. Now suppose the labelingcriteria were changed such that several differentproducts qualified for the certification without anyreformulation whatsoever. Firms that qualify forthe claim may have little if any incentive to con-tinue research to improve the quality of their prod-ucts. Firms that still do not qualify may simplyadopt or license the technology used by the firms

who qualify for the claims and drop their own re-search efforts. However, the existence of the new,“lower bar” may induce some firms to improvetheir product quality, whereas the old, “higherbar” may have been viewed as unachievable andmay also have resulted in no research or productimprovement. Questions concerning the supply--side and general-equilibrium effects of alteringcertification standards on research and develop-ment incentives in nutritional quality are difficultto answer and may require substantial theoreticaland empirical analysis in order to improve our un-derstanding.

Conclusions

Policymakers are increasingly using product label-ing as a tool to alter the behavior of market par-ticipants. Furthermore, economists have long heldthat the flow of information plays a critical role inthe efficient operation of markets. It is thereforesomewhat surprising to find that there has beenlittle in the way of empirical economics research todetermine the market effects (policy effectiveness)of labeling programs. Of the little research that hasbeen done, most has focused on nutrition labelingof food products. Although these studies providevaluable insights into the potential role of labelingin correcting market inefficiencies, empirical re-sults from these studies may not be applicable toenvironmental labeling. Although current demandmodels can easily be adapted to analyze the marketeffects of labeling, data sets that actually allowisolation of label effects are probably rare. Experi-mental data sets, generated both in the lab and inthe marketplace, may provide policy-relevant em-pirical results.

Beyond measuring the policy effectiveness oflabeling programs, economists can contribute tothe design and implementation of these programsso that they improve social welfare. A stated ob-jective of the consumer research leading to the de-sign of the current nutrition label was to enable thepublic to make more informed food choices byproviding as complete information as possiblewhile presenting the information in an easily pro-cessed form (Levy, Fein, and Schucker 1996). Thissuggests an implicit tradeoff between the cost ofinformation acquisition and information accuracy.The type and volume of information presented ona label, as well as the format of the label, willaffect the amount of time and effort an individualmust supply in order to assess product attributes.After a point, simplified information that is easierto process can be obtained only at the cost of less


precision. However, simply increasing the amountof information on a label may actually make anygiven amount of information harder to extract(Chaffee and McLeod 1973). This may cause in-dividuals without the time or ability to process in-formation to ignore it (Heimback 1982; Achterberg1990; Jacoby, Chestnut, and Silberman 1977;French and Barksdale 1974), leading to less opti-mal purchasing decisions (Magat, Viscusi, and Hu-ber 1988).

Unfortunately, the optimum level of simplicityand detail is likely to be different for different in-dividuals and for different products. Determiningthe optimal form of labeling program is an impor-tant welfare question because, given the unequaldistribution over the population of cognitive abili-ties, consumer desires, and values of time, labelingregulations will have equity and distributional im-plications, However, only one published study hasexamined the welfare implications of informationpolicy. To understand these implications, policy-makers need to know the characteristics of labelingprograms preferred by different sectors of the con-sumer population. Economists, using costfbenefitanalysis, may be able to provide a framework inwhich to answer such policy questions.

References

Achterberg, C. 1990. Information presented at the Workshop on

Consumer Understanding and Use of Food Labels, Com-

mittee on the Nutrition Components of Food Labeling,

Food and Nutrition Board, Institute of Medicine, Wash-

ington, D.C. March 13.

Akerloff, G.A. 1970. “The Market for ‘Lemons’: Quality Un-

certainty and the Market Mechanism. ” Quarterly Journal

of ~CO~OWliCS 84:488–500.

Andreoni, J. 1990. “Impure Altruism and Donations to Public

Goods: A Theory of Warm Glow Giving,” Economic

Journal 100:464-77.

_. 1995. “Warm-Glow Versus Cold-Prickle: The Effects

of Positive and Negative Framing on Cooperation in Ex-

periments.” Quarterly Journal of Economics 110(1):1-21.

Bailey, W., and M.A. Eastlick. 1993. “Study Shows Many

Claiming to Be Green Don’t Buy Green.” Northwest Ar-

kansas Times, November 12.

Bockstael, N.E. 1984. “The Welfare Implications of Minimum

Quality Standards, ” American Journal of Agricultural

Economics 66:466–7 1,

Bonneville Power Administration. 1985. Report on Market Re-

search and Program Recommendations: Bonneville Power

Administration Regionwide Promotion of Energy-E@cienc

Appliances. Portland, Ore,: B.G. William Lesh Inc.

Brown, D.J., and L. Schrader, 1990. “Information on Choles-

terol and Falling Shell Egg Consumption.” American

Journal of Agricultural Economics 72:548-55.

Calfee, J. E., and D.J. Ringold. 1988, “Consumer Skepticism

and Advertising Regulation: What Do the Polls Show?” In

Advances in Consumer Research; ed, M.J. Houston, 15:

244–48, Provo: Association for Consumer Research,

Brigham Young University.

Capps, O. Jr., and J.D. Schmitz. 1991. “A Recognition ofHealth and Nutrition Factors in Food Demand Analysis.”

Western Journal of Agricultural Economics 16:21-35.

Caswell, J.A. and E.M. Mojduszka. 1996. <‘Using Informa-

tional Labeling to Influence the Market for Quality in Food

Products. ” American Journal of Agricultural Economics

78(5):1248-53.

Chaffee, S.H., and J.M. McLeod. 1973. “Consumer Decisions

and Information Use. ” In Consumer Behavior: Theoreti-

cal Sources,” ed. S. Ward and T.S. Robertson. Englewood

Cliffs, N.J.: Prentice Hall.

Chaug, H. S., and H. Kinnucan. 1991. “Advertising, Informa-

tion and Product Quality: The Case of Butter. ” American


Chase, D., and T.K. Smith. 1992. “Consumers Keen on Green

But Marketers Don’t Deliver.” Advertising Age, June 29.

Conlisk, J. 1996, “Why Bounded Rationality?” Journal of

Economic Literature 34:669–700.

Coupey, E. 1994. “Restructuring: Constructive Processing of

Information Displays in Consumer Choice,” Journal of

Consumer Research 21( 1):83-99.

Darby, M. R., and E. Karni, 1973, “Free Competition and the

Optimal Amount of Fraud.” Journal of Law and .Econom-

ics 16:67–88.

Diamond, P, 1971. “A Model of Price Adjustment.” Journal of

Economic Theory 3: 156–68.

Estelami, H. 1997. “Consumer Perceptions of Multi-Dimensional Prices. ” In Advances in Consumer Research,

vol. 24, ed. M. Brucks and D. Machrnis. Provo: Associa-

tion for Consumer Research, Brigham Young University.

Farrel, J., and G. Saloner. 1985. “Standardization, Compatibil-

ity, and Innovation.” Rand Journal of Economics 16:70–

83,

Foster, W., and R.E, Just. 1989. “Measuring Welfare Effects ofProduct Contamination with Consumer Uncertainty. ”

Journal of Environmental Economics and Management

17:266–83,

Frazao, E., and J.E. Allshouse. 1996. Size and Growth of the

Nutritionally Improved Foods Market. Agriculture Infor-

mation Bulletin no, 723. Washington, D. C,: Economic Re-

search Service, U.S. Department of Agriculture,

French, W, A., and H.C, Barksdale. 1974, “Food Labeling

Regulations: Efforts toward Full Disclosure.” Journal of

Marketing 38:14-19.

Gehrig, T., and P.J. Jest. 1995. “Quacks, Lemons, and Self

Regulation: A Welfare Analysis, ” Journal of Regulatory

Economics 7:309–25,

Grossman, S., and J,E. Stiglitz. 1980, “On the Impossibility of

hrformationally Efficient Markets,” American Economic

Review. 70(3):393-408.

Gutfield, R. 1991. “Shades of Green: Eight of 10 Americans

are Environmentalists, or at Least Say So.” Wall S’treet

Journal, August 2.

Hartwell, R.V., and L. Bergkamp. 1992. ‘‘Eco-Labeling in Eu-

rope: New Market-Related Environmental Risks?” BNA

lntemational Environmental Daily, October 20.

Teisl and Roe m.. , r,. ,,m

Hashizume, S. 1992. Environmental Labeling in Japan: The

Eco-Mark. Tokyo: Japan Environment Association,

Heimbach, J.T. 1982. Public Understanding of Food Label In-

formation. Washington, D. C.: Food and Drug Administra-

tion.

Ippolito, P. M., and A,D. Mathios. 1990. “Information, Adver-

tising and Health Choices: A Study of the Cereal Market. ”

Rand Journal of Economics 21:459-80,

— 1996. “Information and Advertising Policy.” A Study

of Fat and Cholesterol Consumption in the United States,

1977–1990, Bureau of Economics staff report, Federal

Trade Commission. September,

Jacoby, J., R.W. Chestnut, and W. Silberman. 1977. ‘‘Con-

sumer Use and Comprehension of Nutrition Information, ”

Journal of Consumer Research 4: 199–228.

Johansson, O. 1997, “Optimal Pigovian Taxes under A1tru-

ism.” Land Economics 73(3):297–308.

Kleinmuntz, D, N., and D.A. Schkade 1993. ‘‘Information Dis-

plays and Decision Processes, ” Psychological Science 44;

221-27.

Kuhre, W.L. 1997. ISO 14020s Environmental Labelling-

Marketing. Upper Saddle River, N. J,: Prentice Hall PTR.

Levy, A, S,, and R, Stokes. 1987, ‘‘Effects of a Health Promo-

tion Campaign on Sales of Ready-to-Eat Cereals. ” Public

Health Reports 102(4):398-403,

Levy, A. S,, S.B, Fein, and R,E, Schucker. 1996. ‘‘Performance

Characteristics of Seven Nutrition Label Formats. ” Jour-

nal of Public Policy and Marketing 15(1):1 – 15,

Levy, A. S., 0, Matthews, M. Stephenson, J.E. Tenney, and R.E.

Schucker. 1985. “The Impact of a Nutrition Information

Program on Food Purchases,” Journal of Public Policy

and Marketing 4(l): 1–13,

Magat, W,, W,K. Viscusi, and J. Huber. 1988, “Consumer

Processing of Hazard Warning Information. ” Journal of

Risk and Uncertainty 1(2):20 1-32.

Nelson, P, 1970. ‘‘Information and Consumer Behavior. ” Jour-

nal of Political Economy 78:3 11–29.

—. 1974. “Advertising as Information. ” Journal of Po-

litical Economy 81 :729–54.

Putler, D.S. 1987, The Effect of Health Information on Shell

Egg Consumption, Department of Agricultural and Re-

source Economics Working Paper 448. California Agricul-

tural Experiment Station.

Russo, E. 1977. “The Value of Unit Price Information. ” Jour-

nal of Marketing Research, May: 198–201.

Schkade, D. A., and D,N. Kleinmuntz. 1994. “Information Dis-

play and Choice Processes: Differential Effects of Orga-

nization, Form, and Sequence, ” Organizational Behavior

and Human Decision Processes, 57:3 19–37.

Schucker, R. E., R.C, Stokes, M,L. Stewart, and D,P. Hender-

son. 1983. ‘‘The Impact of the Sacchsuin Warning Label

on Sales of Diet Soft Drinks in Supermarkets. ” Journal of

Public Policy and Marketing 2:46-56.

Scitovsky, T. 1986. Human Desire and Economic Satisfaction.

New York: New York University Press,

Simon, HA. 1955. “A Behavior Model of Rational Choice,”

Quarterly Journal of Economics 691:99-118,

Smith, M. E., E,O. van Ravenswaay, and S.R. Thompson. 1988.

“Sales Loss Determination in Food Contamination Inci-dents: An Application to Milk Bans in Hawaii. ” American


cconomlcs OJumeang 14?

Stiglitz, J.E. 1996. Whither Socialism? Cambridge: MIT Press,

Strahilevitz, M., and J.G. Myers 1998. “Donations to Charity as

Purchase Incentives: How Well They Work May Depend

on What You Are Trying to Sell. ” Journal of Consumer

Research 244:434-46.

Swartz, D. G., and LE. Strand. 1981. “Avoidance Costs Asso-

ciated with Imperfect Information: The Case of Kepone. ”

Land Economics 57(2): 139-50.

Teisl, M. F., and A,S, Levy, 1997. ‘‘Does Nutrition Labeling

Lead to Healthier Eating?” Journal of Food Distribution

Research 28(3):1 8–27.

Teisl, M, F,, L, Halverson, and E, Holt, 1997. Information Dis-

closure and Electricity Marketing: Results from Focus

Group Research—West Coast. Report to the Regulatory

Assistance Project and the Department of Energy, Gar-

diner, Me.

Teisl, M. F., B. Roe, and R.L. Hicks. n.d. “Can Eco-Labels

Tune a Market?: Evidence from Dolphin-Safe Labeling.”

U.S. Environmental Protection Agency (EPA). 1989. Environ-

mental Labeling in the United States: Background Re-

search, Issues and Recommendations, Applied Decision

Analysis. Washington, D.C.

—. 1993. Status Report on the Use of Environmental La-

bels Worldwide. Washington, D.C,

van Ravenswaay, E. O., and J.P. Hoehn. 1991. “The Impact of

Health Risk and Food Demand: A Case Study of Alar and

Apples.” In Economics of Food Safety, ed. Julie Caswell.

New York: Elsevier Science Publishing.

Winneg, K., M,J. Herrmann, A,S, Levy, and B, Roe. 1998,

Label Testing: Results of Mall Intercept Survey. Report to

the National Council on Competition and the Electric In-

dustry. Regulatory Assistance Project. Gardiner, Me,

Wright, P. 1986, “Schemer Schema: Consumers’ Intuitive

Theories about Marketers’ Influence Tactics. ” In Ad-

vances in Consumer Research, ed. Richard J, Lutes, 13: l–

3. Provo: Association for Consumer Research, Brigham

Young University.

Yamaki, S. 1992, Critical Analysis of the Vermont Household

Hazardous Product Shelf-Labeling Program on Consumer

Awareness: A Survey Evaluation,

Zarkin, G. A., N, Dean, J,A, Mauskopf, and D.M. Neighbors.

1991. Estimated Health Benefits of Nutrition Label

Changes: Final Report—Volume 1, RTI no, 233U-3972-

05. Durham, N. C.: Center for Economics Research, Re-

search Triangle Institute.

Zuo, J., and W.S, Chern. 1996. “Health Information and Con-

sumer Participation: The Case of Fresh Milk Consumption

in the U, S.” Presented at the annual meeting of the Ameri-

can Agricultural Economics Association. San Antonio,

Tex.

Notes

1. 1S0 14000 also designates a Type II label,which differs from a Type I or III label in that aType II label is a se~-declared, single-criterionclaim. Thus, Type II claims include “labels, state-ments and many other forms of marketing” (Kuhre1997, p. 15), such as a ‘’40% recycled content”


label on a paper product or a “dolphin-safe” labelon a can of tuna.2. Note that other policy interventions, such asminimum quality standards, often remove theselow-quality items from the market, thus deprivingfirms and consumers from beneficial trades (Bock-stael 1984).3. These three categories follow Nelson’s (1970,1974) and Darby and Karni’s (1973) classificationsof products.4. For example, the attribute of taste lies betweensearch and experience. If one searches enough, onecould taste-test a sample of many foods at a super-market somewhere.5. Gehrig and Jest (1995) note that, in some cases,firms may find it beneficial to impose some type ofself-regulation of information provision. However,their results focus on experience attributes; cre-dence and search attributes may have unique as-pects that affect self-regulation incentives.6. Consumer Reports has remarked that it does notplan to offer any summaries in the household elec-tric service market.7. Most studies have focused on studying the be-havioral effects of information provided throughthe news media (Swartz and Strand 1981; vanRavenswaay and Hoehn 1991; Smith, van Raven-swaay, and Thompson 1988) or the medical litera-ture (e.g., Brown and Schrader 1990; Chang andKinnucan 1991; Capps and Schmitz 1991; Putler1987; Zuo and Chern 1996).

8. In an exception, consumer researchers at theFDA (Levy et al. 1985; Levy and Stokes 1987;Teisl and Levy 1997) used a controlled-market ex-periment to determine the effect of nutrition label-ing on consumer purchase behavior. Treatmentstores exhibited shelf labels carrying nutrient in-formation whereas, during the same time period,control stores did not carry the nutrient informationon the shelf labels. The studies indicate that labelinformation can significantly affect market behav-ior.9. For example, Andreoni (1995) found that ex-perimental participants were more likely to con-tribute to public goods when the experimental in-structions implied that individual action wouldhelp rather than hurt others, even though the re-ward structures accompanying both sets of instruc-tions were identical.10. The labeling of attributes may allow somegoods to exist that would otherwise be eliminatedunder a command and control policy (see note 2),11. Johansson (1997) outlines four types of altru-ism: (1) paternalistic-an agent’s value increasesin the amount that other agents consume a particu-lar good; (2) ncmpaternalistic or pure—the utilityderived by others enters an agent’s utility function;(3) impure or warm glow—an agent’s own utilityincreases only because a transfer is made to others,not because others increase consumption or utility;and (4) genuine—an agent acts like a social plan-ner when optimizing.

the economics of labeling: an overview of issues for...

Documents