University of Massachusetts Amherst Department of Resource Economics Working Paper No. 2003-1 Information Policy and Genetically Modified Food: Weighing the Benefits and Costs Mario F. Teisl1 and Julie A. Caswell2 Abstract: The labeling of genetically modified foods (GMFs) is the topic of a debate that could dramatically alter the structure of the U.S. and international food industry. The current lack of harmonization of policy across countries makes GMF labeling an international trade issue. The U.S. and Canada do not require GMFs to be labeled unless the GMF is significantly different than the conventional food or the GMF presents a health concern. However, many other countries are requiring GMFs to be labeled. This paper discusses empirical work on the sources and magnitude of benefits and costs from labeling programs, with particular emphasis on the impact of the design of the labeling program on benefits and costs. Keywords: GMOs, biotechnology, labeling, benefits, costs JEL Classification: F1, L5, Q10 1 Mario F. Teisl, Department of Resource Economics and Policy University of Maine, 200 Winslow Hall Orono, Maine 04469-5782 E: [email protected] P: 207-581-3162 F: 207-581-4278 2 Julie A. Caswell, Department of Resource Economics University of Massachusetts, 215 Stockbridge Hall, 80 Campus Center Way Amherst, MA 01003 E: [email protected] P: 413-545-5735 F: 413-545-5835 Information Policy and Genetically Modified Food: Weighing the Benefits and Costs3 Introduction The labeling of genetically modified foods (GMFs) is a topic of debate — a debate whose outcome could dramatically alter the structure of the U.S. and international food industry. Recent polls have emphasized that a majority of U.S. consumers desire GMFs to be labeled (CNN 1999, Time 1999, Pew 2001a, Hallman and Metcalfe 1995) and legislation has been entered at both the federal and state levels. For example, H.R. 3377 and S. 2080 the “Genetically Engineered Food Right to Know Acts” were introduced into the U.S. House of Representatives and Senate, respectively. In addition, at least seven states are currently debating nine different labeling and marketing requirements for GMFs (Pollack 2001). The current lack of harmonization of policy across countries also makes GMF labeling an international trade issue. The U.S. and Canada do not require GMFs to be labeled unless the GMF is significantly different than the conventional food or the GMF presents a health concern. However, many other countries are requiring GMFs to be labeled. For example, the EU, Australia, and New Zealand require labeling if a food contains more than one percent GM ingredients (with important exceptions for some foods, e.g., foods served in restaurants). Japan’s policy is similar except its threshold before labeling is required is five percent (Bernauer 2001). Currently, Taiwan and Hong Kong are moving to implement labeling rules similar to Japan’s and China recently issued regulations that appear to require all GM-foods to be labeled (AgBiotech Reporter 2001). GMF labeling is a prime example of a quick moving policy area where individual countries are not willing to take the time necessary for development of international consensus on the best approaches. The strategy is to regulate now and worry about harmonization later. The record of discord and gridlock in the relevant Codex Alimentarius committees reinforces the "everyone for themselves" approach. 3This paper was originally presented at the 2nd World Congress of Environmental and Resource Economists, Monterey, California, June 2002. 2 The debate surrounding the labeling of GMFs is largely framed as how much information should be supplied to consumers to facilitate effective choice and how that information should be supplied. The U.S. Food and Drug Administration’s (FDA) position on how much information should be supplied occupies one end of the spectrum in this debate. It says that the Food, Drug, and Cosmetic Act (by which FDA obtains legal authority for the labeling of foods) requires food labels to “disclose information that is material to representations made or suggested about the product and consequences that may arise from the use of the product” (CFSAN 1999). Under its substantial equivalence approach, FDA has interpreted this to mean that labeling is not required for most GMFs and, furthermore, that voluntary labeling of the nonuse of biotechnology may be misleading to consumers (CFSAN 200x). This policy can be characterized as a ‘need to know,’ or perhaps more accurately a ‘do not need to know,’ policy. At the other end of the spectrum is the “right to know” position where all products using biotechnology are required to be labeled. Under H.R. 3377, for example, GMFs would be required to exhibit a label saying: THIS PRODUCT CONTAINS OR WAS PRODUCED WITH A GENETICALLY ENGINEERED MATERIAL Several countries including the EU, Japan, Australia, and New Zealand are using this approach. Proponents of this alternative usually state that a consumer has a ‘right to know’ that a food is genetically engineered. For example, H.R. 3377 explicitly states, “consumer’s [sic] have a right to know whether the food they purchase contains or was produced with genetically modified material”. The right-to-know position focuses not on an intrinsic attribute of the product, or necessarily on the private effects of product consumption, but on the process attribute of how the product was produced. The flow of information among market participants plays a critical role in the efficient operation of markets. In a broad sense, labeling and marketing have the ability to convert a market in which all goods feature an attribute that consumers cannot observe (e.g., was biotechnology used in producing this product?) into one that can be learned prior to purchase. From a policy perspective, labeling allows consumers to make choices match their preferences. From a business perspective, labeling allows firms 3 that use (or don’t use) particular techniques to gain market share and maximize any value-added rents. Most fundamentally, disagreement about labeling policy is about whether increased information on the use of biotechnology will improve consumer choice and market function and at what cost. Here, we explore the operation of possible labeling approaches to GMFs and use a cost-benefit framework to evaluate alternatives. We argue that both the substantial equivalence (FDA) and right-to- know positions have limitations because each focuses on only a particular attribute of the food. Given the multi-attribute nature of many genetic modifications, and the relatively low degree of knowledge and understanding of genetic engineering, consumers are likely to desire information about the range of possible benefits and risks associated with GMFs. What GMF Labeling Does and Doesn’t Do The central function of labeling is to communicate to consumers about the quality attributes of a product. The use of biotechnology in food production can have multidimensional effects on product quality. For example, it affects intrinsic process attributes (e.g., how was this product produced) and may affect search attributes such as color and size, experience attributes such as use characteristics, and other credence attributes such as nutritional content or presence of pesticide residues (Caswell 2000). Consumers may want or regulators may believe they need to know about some or all of the changes in product attributes brought about by the use of biotechnology. Studies of consumer acceptance of GMFs tend to identify two groups: those that accept or are indifferent to GMFs and those who reject them. In terms of quality information needed, those consumers who wish to reject or select products based solely on the use of biotechnology need only labeling that allows them to make this distinction. Other consumers may have a generally rejecting or accepting view of GMFs but their purchase decision will be influenced by other quality attributes associated with competing products and price. Yet other consumers have no concerns about GMF status and will focus only on comparing other quality attributes between GMFs and conventional products. For the latter two 4 groups, a labeling program that simply communicates whether biotechnology was used will likely provide inadequate information or be irrelevant. The primary function of GMF labeling to date has been to turn a credence attribute regarding the process of production into a yes/no search attribute that can be learned before purchase. The consumer’s ultimate purchase decision will be influenced by the yes/no GMF labeling, any additional explanatory labeling provided, direct observation of product attributes, and prior experience with use of the product. The yes/no message about the use of biotechnology can be delivered to consumers through different means. So-called “positive” mandatory labeling requires companies to tell consumers when biotechnology has been used in production or when cross-contamination from bioengineered products is above a defined threshold. In this case, the absence of labeling may communicate that a product is not a GMF. In contrast, so-called “negative” voluntary labeling allows companies to tell consumers that their product is non-GMF, again if it meets standards for such a claim. In this case, the absence of a label may indicate that the product is GMF. The qualifier may is necessary in the above sentences because unless labeling is symmetric (i.e., absence and presence are both labeled) what consumers assume about the unlabeled product is a matter for empirical inquiry. Expanding labeling to communicate beyond the yes/no level gives rise to a further array of issues regarding what is communicated and how consumers perceive the message. The central questions for labeling programs are two-fold. First, to what extent do they provide benefits by improving consumer decision making, and by extension consumer welfare, and for which consumers? Second, what is the cost of providing those benefits given that any labeling program must be supported by a set of standards, actions to meet the standards, certification of the actions, and governmental enforcement of the program? The balance of benefits to costs is affected by technology (e.g., testing techniques) and may change over time. For example, it appears that one of the reasons some consumers want GMFs labeled is due to uncertainties about long-term health and environmental effects (Teisl et al. 2002). With time, consumers’ perceived uncertainty about these long-term effects (and, presumably, the associated benefits of labeling) may decline. 5 What Are the Benefits of Labeling? The benefits of providing product specific information can be measured by its ability to inform consumers as to the positive and negative attributes of the product. When product information is well understood and credible, then consumers’ choices match their preferences and dollars spent are in-line with any underlying willingness to pay for the bundle of attributes received. Firms that produce goods with desirable attributes also gain as they are rewarded for providing those attributes. However, the benefit of labeling depends upon the type of attributes considered. In general, labeling is increasingly beneficial, as attributes become more costly or difficult for a consumer to independently assess (Caswell and Mojduszka 1996). For example, most individuals can identify the color of a product rather easily while verifying that a food was not genetically modified would be difficult. The aggregate benefit of a labeling policy will also depend upon the relative importance the information has to consumers. In general, benefits are maximized if either 1) the information is important to a large number of consumers, even if the information may be of relatively small importance to each consumer or 2) the information is extremely important for even a small number of consumers (e.g., disclosure of peanut ingredients to those who have a life-threatening allergy to peanuts). Policies that allow consumers to make purchase decisions match their preferences are inherently desirable, whether the attributes concern end-use characteristics (e.g., the consistency of flour used in baking) or process attributes (e.g., whether genetically modified grain was used in making the flour) – so long as these policies are not too costly. Thus, there is no a priori reason for FDA to limit its labeling policy, as it currently does, to only product attributes and the private consequences of product consumption. In fact, other federal labeling programs focus on disclosing information about process attributes (e.g., organic labels, irradiation) or the public consequences (e.g., dolphin-safe labeling of canned tuna) of product consumption. Until recently, there has been little empirical evidence identifying the benefits of labeling GMFs. Golan and Kuchler (2000) and Golan, Kuchler, and Crutchfield (2000) use economic intuition and 6 estimated supply and demand parameters from previous studies4 to estimate the welfare impacts of GMF labeling. In their analysis, they assume consumers are differentiated by whether they want to avoid GMFs (or are indifferent between non-GMFs and GMFs). They also assume the genetic modification only provides benefits to consumers by lowering GM food prices (the genetic modification reduces the costs of production). They then compare market and surplus changes due to the institution of a labeling program in two different situations. In the first situation, the labeling program is costless; in turn, it is not surprising that instituting a labeling program is necessarily welfare enhancing. Labeling leads to improvements in consumer surpluses for those consumers who don’t care to avoid the technology (because they enjoy greater price reductions once labeling is imposed). In addition, labeling increases the welfare of individuals who want to avoid GMFs by allowing them to do so. Under this scenario they estimate a net consumer welfare gain of $76 million for the introduction and labeling of genetically modified soybeans. In the second situation, the labeling program is no longer costless. Golan and her coauthors provide a list of examples showing where labeling and product differentiation may be costly and note that these are all due to the ‘production externalities’ caused by having GMFs in the market. Their list of externalities include such items as: non-GM crops being cross-pollinated by neighboring GM crops, increased resistance development in non-targeted insects and weeds, maintenance of separate storage and transportation facilities, and certification/testing costs.5 They then assume that all of these costs fall solely on the non-GMF market.6 Under this situation labeling leads to improvements in consumer surpluses for those consumers who are indifferent to the technology. However, because labeling is now costly, the introduction of GMFs and the imposition of labeling can lead to reductions in the welfare of individuals 4 Falck-Zepeda, Traxler and Nelson (2000); and Moschini, Lapan and Sobolevsky (2000) looked at changes in welfare due to the introduction of GM soybeans; they did not estimate welfare effects due to labeling. 5 Although the first two items are externalities the second two are not. They are just the cost of instituting a labeling program; these latter costs can occur with any labeling program, even those without externalities. 6 This assumption is not realistic. Note that some externalities may be faced by both GM and non-GM producers (e.g., increased resistance in weeds) and others will depend upon the property rights of growers (e.g., who will be forced to incorporate buffer strips to reduce cross-pollination). Further, who pays the labeling program costs will be dependent upon the structure of the program. 7 who want to avoid GMFs. This is due to these consumers having to pay more for the same non-GM foods that they initially obtained at a lower cost. Further, as the proportion of consumers who prefer non-GM foods grows, the relative size of the consumer welfare loss due to higher non-GMF prices begin to outweigh the gains due to the lower GMF prices. They indicate that if 25 percent of the market prefers non-GMFs, then losses to consumers may outweigh gains. Under this scenario they estimate a net consumer welfare loss of $21 million. Bullock and Desquilbet (2001) obtain results similar to those of Golan, Kuchler and Crutchfield (2000), in that they indicate that anti-GMF consumers are made worse off by the appearance of GM technology if labeling is imposed. Although they are able to avoid consuming GMFs the segregation and identity preservation costs cause them to pay more for food than if GMFs had never appeared. However, unlike Golan, Kuchler, and Crutchfield, individuals who are indifferent between GMFs and non-GMFs may also lose because they may pay higher food prices than they would otherwise, due to the increased costs of labeling and segregation. The above results hinge critically on several factors. First is the assumption made about the structure of the labeling program, which is assumed to be voluntary. Second, the structure of the problem assumes that there are only two types of consumers, those who want to avoid GMFs and those who are indifferent. This is tied to the assumption that the genetic modification only provides benefits to consumers by lowering prices; i.e., the studies do not look at situations where the individuals may derive a non-price benefit. However, the bottom line of these two studies support the central thesis here that the imposition of a labeling program for GMFs should depend upon a careful weighing of the benefits and costs associated with a specific program. Lusk et al. (2001) and Huffman et al. (2002) used experimental auction methods to elicit consumers’ willingness to pay for GMFs. The Lusk study used a sample of students at Kansas State University and found they were unwilling to pay a premium for non-GM corn chips. Huffman et al. used a randomly recruited sample of individuals from two urban areas (Des Moines, IA and St. Paul, MN) and found that consumers were willing to pay about a 14 percent premium for food items (vegetable oil, 8 tortilla chips, and potatoes) they perceived as non-GM (the GMFs were labeled as being genetically modified). McClusky et al (2001) and Boccaletti and Moro (2000) used survey approaches (contingent valuation) to estimate the willingness to pay for or to avoid GMFs. McClusky et al. used an intercept sample of Japanese consumers and, similar to Lusk et al. and Huffman et al., studied consumer reactions to GMs that only reduced the price of the food (as opposed to altering another product attribute). They found that Japanese consumers would be willing to pay to avoid GMFs; that is, consumers would only be willing to purchase GM noodles or tofu if there was a 60 percent discount. The Boccaletti and Moro study is different, in that they looked at Italians’ willingness to pay for GM foods with enhanced attributes (reduced use of pesticides, improved nutritional or organoleptic characteristics, or longer shelf life). They found that a majority of consumers would be willing to pay up to 10 percent more for these GM foods. The experimental and survey results again are not necessarily surprising. Consumers are not willing to pay more for a new technology (that because it is new is almost necessarily associated with long-term effects that are, from the consumers point of view7, uncertain), without being provided an incentive to do so. That is, consumers faced with a potential long-term cost will reject a technology unless they perceive themselves as obtaining some sort of benefit. Again, the bottom line of these studies supports the central thesis here. What Are the Costs of Labeling? As with benefit estimates, the costs of labeling programs are highly dependent upon the particular labeling program being considered. Labeling programs require standard setting, private compliance and certification efforts, and public enforcement oversight. Typically the costs of the actual physical labeling (e.g., label design and printing) are a tiny fraction of the costs of compliance and certification (supply chain costs), particularly if the transition time to the new labeling regime is at least a period of months. 7 Teisl et al (2002) indicate that many focus group participants, when notified of the prevalence of GMFs, were comforted; participants combined the fact that GM-foods are prevalent with the notion that they had not heard or known of anyone getting sick as positive news. 9 Other costs may include the costs to public authorities of monitoring and enforcing compliance with labeling regimes, the costs of trade impacts from labeling programs, costs arising from possible changes in market structures, and costs from the dilution of information already included on labels (n/e/r/a/ 2001). As an example of the latter potential cost, labeling may impose cognitive costs on some consumers. Simply increasing the amount of information content on a label may actually decrease the consumer’s ability to process other more important label information (Scammon 1977, Roe, Levy, and Derby 1999). In addition, requiring specific information to be placed on a label imposes a cost in that the limited space on the product label could have been devoted to other, potentially more useful and important information. For example, almost twice as many Americans say they would rather have information about pesticide residues placed on a food label rather than information about GMFs (CSPI 2001). Because information content is competing for valuable space on the label, labeling requirements should be justified in terms of the importance of the required information. A prescription such as “more information is better” does not necessarily characterize an optimal labeling policy. Costs are difficult to measure, particularly because many labeling policies are newly implemented, in the process of being implemented, or at the proposal stage. Perhaps as importantly, the introduction of GMFs is taking place in a dynamic international market in which shifts in sources of supply and the move from a commodity to an identity preserved orientation are ongoing. This makes it challenging to isolate the impacts and costs of labeling programs, and associated attitudes toward GMFs. Work by Ballenger et al. (2000) indicates the type of price and trade flow analysis necessary to begin to disentangle these effects. As a result of this difficulty, most studies focus on particular supply chain costs of labeling. The exception is the n/e/r/a (2001) study done in the United Kingdom that attempts to measure a broad range of benefits and costs of different options for extending the EU’s current labeling program. It is important to acknowledge that many believe that much more is at stake in program adoption than the incremental costs, however significant, of a particular labeling regime. They think that labeling policy, particularly mandatory labeling of GMFs, fronts for an entire agricultural, food, and trade agenda 10