What do consumer surveys and experiments reveal and conceal about consumer preferences for genetically modified foods?

Abstract

Assessing consumer perceptions and willingness to pay for genetically modified (GM) foods has been one of the most active areas of empirical research in agricultural economics. Researchers over the past 15 years have delivered well over 100 estimates of consumers' willingness to pay for GM foods using surveys and experimental methods. In this review, we explore a number of unresolved issues related to three questions that are critical when considering the sum of the individual contributions that constitute the evidence on consumer preferences for GM foods.

Keywords: :

• GM foods
• GM labeling
• experimental auctions
• surveys
• consumer preferences
• consumer valuation
• WTP for GM foods

Introduction

Assessing consumer perceptions and willingness to pay for genetically modified (GM) foods has been one of the most active areas of empirical research in agricultural economics. Researchers over the past 15 years have delivered well over 100 estimates of consumers' willingness to pay for GM foods using surveys and experimental methods. Studies have spanned a wide spectrum of consumer categories, countries, food products, and GM technologies (see Table 1). Two meta-analysis papers, Lusk et al.¹ and Dannenburg,² along with a review paper by Lusk³ provide useful summaries of some of the insights that have been revealed by the substantial research on consumer GM attitudes. In particular, three general findings have been consistently supported by the literature.

Table 1. Consumer GM Food Studies by Country

Country	Study
Australia	Burton and Pearse,^10 James and Burton^18
Brazil	Gonzalez, Johnson, and Qaim^73
Canada	Gao, Veeman, and Adamowicz,^89 Hu et al.,^16^,^81 Hu, Veeman, and Adamowicz,^17 Huang et al.,^87 Larue et al.,^20 Volinkskiy et al.,^27 West et al.^28
Chile	Schnettler, Sepúlveda, and Ruiz^102
China	De Steur et al.,^84 Hu,^15^,^103 Li et al.^76
France	Noussair, Robin, and Ruffieux^53^,^54
Germany	Dannenberg, Scatasta, and Sturm,^46 Hartl and Herrmann^14
India	Anand, Mittelhammer, and McClusky^71
Ireland	O'Connor et al.^104
Italy	Boccaletti and Moro^80
Japan	Peterson and Burbidge,^25 Hu,^103 Terawaki,^85 McClusky et al.^78
Kenya	Kimenju et al.,^88 Kimenju and De Groote^75
Malta	Zammit-Mangion, Al-Qallaf, and Vella^105
New Zealand	Jaeger and Harker,^49 Kaye-Blake, Bicknell, and Saunders^19
Norway	Grimsrud et al.^74
Phillipines	Depositario et al.^47
South Africa	Vermeulen et al.^106
South Korea	Onyango and Govindasamy,^23 Chung, Boyer, and Han^12
Taiwan	Chen^107
Uganda	Kikulwe, Wesseler, and Falck-Zepeda,^108 Kikulwe et al.^109
United Kingdom	Burton et al.^9
United States	Bernard et al.,^110 Buhr et al.,^44 Bukenya and Wright,^72 Colson and Huffman,^45 Colson, Huffman, and Rousu,^4 He and Bernard,^48 Loureiro and Bugbee,^82 Loureiro and Hine,^83 Lusk,^77 Lusk et al.,^21^,^51 Onyango, Nayga, and Govindasamy,^24 Rousu et al.,^55Wachenheim and Vanwechel^56
Multiple Countries (Asia, EU, and/or US)	Chern et al.,^11 Lusk, Roosen, and Fox,^22 Lusk et al.,^50^,^52 Moon and Balasrubramanian,^79 Tonsor et al.^26

Finding #1: Consumers' willingness to pay for GM foods is lower than for non-GM foods

Studies consistently show consumers are willing to pay a premium for non-GM food products or would pay less if the product is genetically modified. In the Lusk et al.¹ meta-analysis, their review of 25 studies found a premium of almost 29% for non-GM food products (excluding an outlier). Dannenberg² finds a similar premium in her meta-analysis: an 18% median premium and a 45% mean premium for non-GM food products. The finding of preferences in opposition to GM foods is largely consistent across countries, elicitation mechanisms, and food types.

Finding #2: US consumers are more accepting of GM foods than European consumers

This is found both in individual studies and the reviews by Lusk et al.¹ and Dannenberg.² In Lusk et al.¹ they state “Europeans are willing to pay premiums for non-GM food that are 92 percentage points higher than the ones American consumers are willing to pay.” This finding is also consistent with public policy. While European countries have mandated labeling of GM food products, public support in the US has failed to shift US policy from voluntary to mandatory labeling. A recent ballot initiative in California, which has a population more sympathetic toward required labeling of GM foods than much of the US, failed in November, 2012 to force labeling rules more in line with European requirements.

Finding #3: The magnitude of consumers' discount for GM foods depends upon the type of genetic modification, the type of food product, and how the genetic modification alters the final product

Dannenburg's² meta-analysis examined whether consumer preferences differed based on whether the genetic modification was on an animal or on a plant. She found that consumers were less accepting of genetic modification when dealing with animals relative to plants. Lusk et al.¹ found that consumers are more accepting of genetically modified foods when the product has a “direct tangible benefit” to consumers such as improved nutrition or reduced pesticide use. Colson, Huffman, and Rousu⁴ and Lusk and Rozan⁵ found that consumers distinguish between GM foods developed using genes within the same species compared with foods developed using genes from different species.

Overall, while it appears that the myriad of consumer GM food studies have converged on these three primary findings, it is important to view the supporting evidence through the appropriate lens and consider potential confounds and missing elements relevant for characterizing consumer attitudes and guiding public policy. In this review, we explore a number of unresolved issues related to three questions that are critical when considering the sum of the individual contributions that constitute the evidence on consumer preferences for GM.

• Question #1: Why do willingness to pay (WTP) estimates vary substantially across studies?

• Question #2: What do consumer WTP studies actually reveal, given many consumers have low levels of awareness and knowledge about genetic modification?

• Question #3: What is the relationship between (1) WTP estimates from surveys and experiments and (2) consumer valuation and behavior in real-world food purchasing settings?

To explore these questions, we first provide an overview and critical discussion of the variety of preference elicitation mechanisms that have been used by researchers in the literature. Second, we consider the evidence on the influence of the information environment in which consumers' preferences are elicited and the confound uncertainty presents for the interpretation of GM food valuation studies. Finally, before concluding, we consider potential issues related to a common element across GM food studies—they involve researcher constructed decision situations.

Methodological Factors Affecting WTP Estimates

Given the dearth of traditional retail- or individual-level shopping data covering actual purchases of GM and non-GM foods (e.g., scanner data, food diaries), researchers have employed an array of survey and experimental approaches to qualitatively and quantitatively gauge consumer attitudes, valuations, and market decisions. As evidenced by the meta-analyses by Lusk¹ and Dannenberg,² there are a number of study-specific factors that can in part explain the variation in consumer valuations for GM across different studies. Among the variety of explanations for the variation in willingness to pay estimates (e.g., study subject pool, food type, type of genetic modification), evidence suggests that the method and format of the elicitation mechanism used in a study is one of the most significant drivers of differences. In this section we review the primary approaches that have been used in the WTP for GM foods literature and the method-specific factors that in part explain why it has been difficult for the literature to consistently coalesce upon a consensus measure of consumers' WTP for GM vs. non-GM foods.

Choice experiments

Across the economics, marketing, and transportation literatures, discrete choice experiments (CE) are among the most commonly utilized methods for assessing consumer preferences. Choice experiments, which are economically rooted in random utility theory,^6-8 are relatively simple to implement with consumers and have a number of attractive features. In practice, consumers are presented with a series of decision situations consisting of a discrete set of alternative products (or descriptions of products) that are characterized by their attributes and price. In each choice scenario, which mimics the decision situation consumers are familiar with in typical retail settings, consumers are asked to choose their most preferred alternative. In the GM valuation literature, choice experiments have been applied in a number of studies.^9-28

Although there is strong support for choice experiments capturing consumer preferences,²⁹ there are a number of concerns that have been raised with the methodology that are relevant. First, in contrast to experimental auctions, choice experiments only reveal consumers WTP indirectly and relies upon a researcher-assumed functional form.³⁰ Second, choice experiments have been shown to potentially be sensitive to a number of seemingly innocuous design elements such as the number of choice sets presented to participants, the number of alternatives listed for each product alternative, the ordering of product attributes, and the range of attribute levels.^31-34

Third, and perhaps most controversial and consequential, is the hypothetical nature of the majority of choice experiments that have been conducted for GM food products. While there are exceptions,^26,27 the preponderance of GM choice experiments do not involve any financial consequences for participants' decisions within a CE; thus, raising the specter of potential hypothetical bias. While in the contingent valuation literature the emerging consensus appears to be that hypothetical bias is indeed a significant problem (for a review, history, and discussion see refs.^35-37), in the choice experiment literature the evidence is sparse and less clear.³⁸ For example, while Carlsson and Martinsson³⁹ and Lusk and Schroeder³⁰ find no statistically significant difference between marginal willingness to pay (i.e., attribute valuation) in hypothetical and non-hypothetical CEs, Lusk and Schroeder³⁰ and Alfnes and Steine⁴⁰ do find a statistically significant difference in total willingness to pay. Overall, while it is not clear if or to what extent the hypothetical nature of most choice experiments biases estimates of consumer WTP for products and product attributes, this is an issue that remains a concern when interpreting GM valuation estimates.

Experimental auctions

The second most common approach to eliciting consumer WTP for GM foods is non-hypothetical experimental auctions such as the Vickrey 2nd-price auction⁴¹ or the random nth price auction.⁴² The BDM mechanism⁴³ is technically not an auction, but shares enough traits with the random nth price auction that it is typically considered an experimental auction, as we will here. Experimental auctions avoid two of the primary concerns surrounding choice experiments: they are theoretically incentive compatible due to their non-hypothetical nature and they directly elicit a measure of individuals' WTP without any researcher assumption of functional form. Given these properties, auctions have been utilized in a significant number of GM studies.^4,44-56

Despite the attractive theoretical properties of experimental auctions, a number of concerns have emerged in the literature that may in part explain why (a) there is a disparity in valuations across different studies using auctions and (b) there is a significant difference in valuations between auction studies and studies using non-auction methods. First, despite the theoretical incentive compatibility of a variety of alternative auction mechanisms, there is mixed evidence whether different types of auctions (e.g., Vickrey vs. BDM) indeed yield common bidding results.^57-60 Second, in practice there are substantial variations in the specific protocols researchers use when implementing their auction research study; many of which have been shown to have consequential effects on agent bidding behavior. Some of the protocols shown to impact bidding behavior are whether the auction uses one round or multiple repeated rounds of bidding,⁶¹ whether prices are posted,^62,63 whether participants are endowed with products,⁶⁴ and the use of practice rounds.^65,66 Third, the incentive compatibility of auctions creates a potential problem in the cases where the bidders' optimal strategy is no longer to bid their true value for a product due to the availability of outside the auction substitutes in the marketplace. As explored in Alfnes,⁶⁷ Colson, Corrigan, and Rousu,⁶⁸ and Harrison, Harstad, and Rutström,⁶⁹ in the presence of outside substitutes, auction bids should theoretically be censored from above at the market price, thus rendering individual bids no longer a true measure of a consumer's product valuation for some consumers.

Overall, while there is an emerging consensus of “best practices” for conducting experimental auctions,⁷⁰ it is clear that the specific procedures of the auction can affect bidding behavior. Given this fact, it is important to keep this feature in mind when comparing studies using experimental auction methods.

Other methods

While choice experiments and experimental auctions are the most common methods for eliciting consumer values for GM foods, a variety of other approaches have been used in the literature. Examples include dichotomous choice valuations,^71-79 payment cards,^79-83 and open-ended survey questions.^84,85 As with choice experiments, these approaches have the potential to elicit biased value estimates due to their hypothetical nature. In addition, these methods are subject to a similar criticism as experimental auctions in that the decision setting presented to respondents is dissimilar to the traditional choice situation consumers face when making food purchases in a retail setting. Although it is not clear what specific feature of these methods drive the disparity, the literature review of GM studies by Dannenberg² finds a substantial difference in valuations using these methods. Dannenberg² estimates that the premium for GM food in studies using dichotomous choice methods and payment cards or open-ended questions is 27% and 40% lower, respectively, compared with studies using choice experiments. No statistically significant difference in valuations between choice experiment studies and experimental auctions were found when other study characteristics were controlled for.

WTP of Consumers with Low Information in an Uncertain Environment

One of the most prominent potential confounds that must be considered when evaluating the substantial evidence that has accumulated assessing consumer WTP for GM vs. non-GM foods is the role of uncertainty. Genetic modification and the emergence of commercialized first-generation GM commodities represented a significant shock to the food market, affecting consumers and producers globally. As with all new technologies, the emergence of GM foods entails uncertain risks and benefits. However, given that food is a fundamental component of all human life, the scale of the potential risks and benefits to humanity is monumental. In settings involving uncertainty evolving over time, information can play a key role in determining market outcomes and ultimate welfare realizations. Further, as objective information of the risks and benefits of GM foods emerges, this should reduce the level of uncertainty (regardless of whether the information is positive or negative), and economic theory dictates that the market will react.

However, when considering the issues of information and uncertainty in the GM food debate, it is clear that a number of questions remain unanswered that are vital for appropriately assessing the conclusions that can be drawn from consumer valuation studies. From the literature on WTP for GM foods, two stylized facts have seemingly emerged: (1) consumers are willing to pay a substantial premium to avoid GM foods and (2) consumers are largely uninformed about genetic modification technologies, benefits, risks, and regulations. With regards to stylized fact (1), Lusk et al.¹ and Dannenberg² estimate respectively in their review of the existing valuation literature that consumers are willing to pay a 42% and 45% premium for non-GM foods (when adjusted to account for the number of observations in each study, the premium drops to about 25%). While over 51 primary studies with 114 valuation estimates lead to this conclusion that consumers are WTP a premium for non-GM foods, stylized fact (2) raises the question of what exactly this premium is based upon. In the US, a 5-year annual survey of households between 2001–2006 found that only 32–44% of respondents had seen, read, or heard recently about genetically modified food that is sold in grocery stores.⁸⁶ Similar levels of awareness about GM food have been uncovered globally. In China, 67% of respondents stated that they had heard of GM foods,⁸⁷ and only 38% in Kenya.⁸⁸ Even when information is made readily available and nearly costless to acquire, as in a study by Gao, Veeman, and Adamowicz⁸⁹ where consumers completing an online choice experiment were able to click on a link to more information about GM, less than half of the Canadian participants accessed the information.

Given the combination of high consumer premiums for non-GM food and seemingly low levels of awareness and knowledge about GM technology in general and GM food in particular, three natural questions emerge. First, if consumers place a significant premium on non-GM food, why is there not extensive marketing and labeling of food products as such? Second, are consumers actively seeking information about food products in order to ensure that they obtain the substantial premium they seemingly place on GM foods? Third, in a market environment characterized by low information consumers, what role does information play in market outcomes and improving welfare? While the evidence regarding the first question remains unclear from the existing literature and would benefit from market-level sales data evidence on GM and GM-free product label introductions, there is evidence for the latter two questions. In a study of French consumers, Noussair, Robin, and Ruffieux⁹⁰ found that French consumers expressed strong opposition to GM foods with 79% of respondents agreeing that GMOs should simply be banned. However, in French consumer experimental auction studies of food products with GM ingredients, Noussair, Robin, and Ruffieux^53,54 found that bids for GM foods were only affected when they explicitly alerted bidders to the GM ingredients listed on the product they were bidding for. Viewing this result in the larger context of GM valuation studies, it casts concerns on whether choice experiments, experimental auctions, or other methods are appropriately capturing actual consumer behavior and valuations. Either, consumers do indeed value non-GM foods, but simply choose not to inspect their products prior to purchase or believe that they are non-GM (and hence no inspection is necessary), or perhaps consumer aversion to GM is not nearly as strong as valuation studies indicate. Better understanding the connection between experimental valuation studies and actual consumer retail-setting behavior remains an important area of research.

Experimental evidence on the impact of information

Tied to this issue of how consumers make their food choices in an uncertain environment is the question of what role information can play in improving consumer welfare by moving consumers from an uninformed to an informed state. Anand, Mittelhammer, and McCluskey⁷¹ find that consumers in India have a WTP for GM wheat that is substantially affected by information. Most notably, they find that when negative health information is provided, the average respondent was not willing to buy GM wheat at any price. In a multinational study (England, France, and US) using experimental auctions, Lusk et al.⁵² find that in three different information treatments (environmental, health, and world benefit of GM) each significantly decreased the money consumers required to consume a GM cookie in the US and England, but had little impact on French consumers' willingness to accept GM. In a study of US consumers, Rousu et al.⁵⁵ find that consumers' premium for non-GM foods is affected by positive, negative, and verifiable 3rd-party information. Furthermore, using the bid prices under differing information treatments, they estimate that there is a small per-person, per-product value of verifiable information ($0.051, $0.054, and $0.045 per bag of tortilla chips, bottle of vegetable oil, and bag of potatoes, respectively). Although small, when considering the number of consumers and products sold on an annual basis, the public value of verifiable information is substantial. Using a similar methodology, Colson, Huffman, Rousu⁴ and Colson and Huffman,⁴⁵ find that US consumers' willingness to pay for intragenic and transgenic GM labeled foods with enhanced nutritional attributes are significantly impacted by the information available to participants. In an online choice experiment with Canadian consumers, Hu et al.⁸¹ provided voluntary access (via a simple mouse click) to specific information about GM (e.g., environmental effects, evidence on GM food safety). The study indicates that in this voluntary information access setting information significantly impacts utility from GM.

Polling evidence on the impact of information The case of California proposition 37

In November, 2012, Californians voted on a ballot initiative which would have required all GM foods to be labeled as genetically modified. This was known as Proposition 37. It was eventually defeated, but it provides great non-experimental insights on how powerful information on GM foods can be in shaping consumer preferences.

While the proposition was eventually defeated by a 51.5–48.5% margin, it was not certain this would be the case. Figure 1 shows the support for Proposition 37 among Californians between July and late October. For most of the polling period, the support in favor of Proposition 37 was well over 60%. The actual vote likely would have been much higher, as the poll found many “undecided” voters. With a relatively even split of the undecided voters, the bill may have received over 70% of the vote if it were held in July. As the election drew nearer, however, the television ad campaigns picked up. Opponents of Proposition 37 raised $45 million to defeat the bill, while supporters raised about $6.7 million.⁹¹ Most of this money was spent on television ads providing information to the public about what the bill would mean (either in favor of it, or opposed, depending on the group spending the money). The change in support in the last month and a half of the campaign was stunning. The Proposition went from well over 60% support to a final telephone poll tally of 40% support. The final vote ended above this amount, at 48.5% (Fig. 1).

Figure 1. Percentage Support for Proposition 37 Among Californians in Telephone Polls.¹⁰¹

WTP Estimates from Surveys and Experiments vs. Real-World Markets

One of the motivations that have driven the burgeoning literature assessing consumers' WTP for GM foods is the need to help inform agribusinesses, farmers, and policymakers of the nature of the demand-side of the market for GM and non-GM foods. Socially optimal labeling policies and profit maximizing production and R&D decisions in the agricultural sector hinge in part on the value consumers place on GM foods and the factors driving these consumer valuations.^92,93 When considering the insights and implications that can be drawn from the totality of the empirical consumer WTP for GM literature, it is critical to consider the setting and factors in which the valuation estimates were drawn. Each of the GM food studies described in this review relied upon decision situations that were artificially created by researchers and then presented to consumers. None of the studies reviewed in this chapter or the most complete meta-analysis by Dannenberg² involve market transactions by consumers in a non-experimental market.

This raises two pertinent questions. First, how confident are we that the value estimates obtained through surveys and experiments capture consumers true WTP for GM foods and behavior outside of experimental markets? Second, are there important factors not captured in experimental GM studies that may influence consumers’ WTP and their ultimate purchase decisions in real-world markets? In this section we discuss some of the evidence pertaining to these two questions and a number of open questions that are important for future research.

What do WTP estimates for GM food from surveys and experiments capture?

Across the array of studies assessing the value of GM foods, a number of findings are now commonly accepted. As discussed earlier, these include that consumers are willing to pay less for GM foods, the magnitude of this discount varies across countries, and the magnitude differs based on the type of genetic modification and how it affects the food product. As well, research has revealed that there is substantial heterogeneity among consumers in terms of their preferences for GM foods. Socio-economic factors such as education levels, income, and exposure to agriculture (e.g., urban vs. rural residence) are significant factors in explaining preferences. Furthermore, the literature has revealed that consumer preferences are conditional on their information state; that is, that consumer exposure to information has a significant influence on their preferences at a given point in time. This, coupled with the widely revealed aspect that most consumers are largely uninformed regarding the scientific, economic, environmental, health, and political facts related to GM, indicates that public attitudes and the market for GM foods are still evolving and able to be influenced by interested parties.

Although these results appear to be robust across the empirical literature, it is important to consider the context in which they were derived—surveys and experiments, not real-world markets. In order for the literature on consumers' WTP for GM food to inform policy and supply-side decision makers it is necessary for surveys and experiments to not only accurately capture consumer decision making within the artificial experimental environment, but also for there to be a correspondence with real-world consumer decision making. Based upon the previous literature, there is support for a wide array of methods that have been employed by researchers assessing consumers' WTP for GM foods. The evidence of a relationship between real-world market outcomes and experiments include studies focusing on auctions,⁹⁴ choice experiments,⁹⁵ framed field experiments,⁹⁶ discrete choice and payment cards,⁹⁷ and surveys.⁹⁸ This methodological evidence helps give credence to the transferability of GM valuation estimates from experimental settings into actual market decisions.

What Do WTP estimates for GM food from surveys and experiments not capture?

When considering the evidence on GM foods it is important to recognize that a wide array of factors that may affect consumer valuations and ultimate purchase decisions have not been captured in the WTP for GM food literature. For example, it is unknown how consumers’ WTP evolves over time and what roles information and product experience plays in the evolution. Since the experiments reviewed in Dannenberg² typically involve measuring consumers’ WTP at a single-point in time, it is not possible to extrapolate how consumers arrived over time to that valuation or what the future may hold. This is particularly problematic when trying to project the impact of positive and negative information on consumers’ willingness to purchase GM foods. With only a measure of consumers' WTP at a single point in time after being provided with information, it is not known how long-lasting the impacts of information are or how actual repeated product-experiences may come into play. Tied to this is the open question of how social networks and interactions affect consumers’ preferences for GM foods. Are individuals’ willingness to purchase GM foods linked to the attitudes of their family, friends, and society as a whole and if so, how does this bode for the potential of GM foods in voluntary vs. mandatory labeling settings? Even at the household level, it is unclear how the individual preferences of family members aggregate to determine the GM preferences and purchase decisions for the household.

The issue of whether availability of GM foods in a country affects preferences is also an issue that surveys cannot easily address. In the European Commission framework document on GM foods, they found that despite widespread survey/experimental findings of Europeans rejecting GM foods that they will purchase GM foods. They find specific evidence looking at when Polish and UK consumers are in North America, finding that they do not generally avoid GM foods when in North America.⁹⁹

Although, as discussed in the previous section, there is evidence that gives support to surveys and experiments appropriately capturing consumers’ valuations, research assessing the validity of experimental measures of the impact of information on preferences is lacking. Several studies^4,81,100 have assessed the impact of information on participants' WTP for GM and non-GM labeled foods and revealed that information can shift consumer preferences. However, it is important to recognize that the methodologies employed for these studies are constrained in three key dimensions. First, the information is received by consumers in a manner different from how information is normally received (i.e., the information is within a survey or experiment and not from a traditional information source). Second, the format of the information might be different than what the consumer would otherwise receive. Third, the experiment can't completely (or accurately) assess consumers' prior beliefs and attitudes. Given the importance of information, this is a major limitation of the ability of surveys and experiments to not only capture current attitudes, but to assess future attitudes as the information environment changes.

Concluding Thoughts and Recommendations

With more than 100 different estimates of WTP for GM foods from over 20 countries, the preferences of consumers for GM foods have been extensively studied. Research has consistently found that (1) consumers WTP is less for GM foods, (2) US consumers are more accepting of genetic modification than European consumers, and (3) attitudes toward genetic modification depend upon the type of product, genetic modification, and benefits. Despite the general agreement of the literature with regards to these three results, a number of important questions remain, and in particular three questions that we have focused on in this paper.

• Question #1: Why do willingness to pay (WTP) estimates vary substantially across studies?

• Question #2: What do consumer WTP studies actually reveal, given many consumers have low levels of awareness and knowledge about genetic modification?

• Question #3: What is the relationship between (1) WTP estimates from surveys and experiments and (2) consumer valuation and behavior in real-world food purchasing settings?

Researchers conducting studies examining GM food preferences have found wildly different results. These WTP estimates vary dramatically often because of different methods used by researchers to collect data (surveys vs. experiments), different products under consideration (i.e., type of GM), and different participants (e.g., Europeans vs. Americans). When considering the insights that this literature has revealed, it is critical to recognize the limitations and intricacies of the survey and experimental methods that have been employed to measure consumer preferences. Although there is a growing body of research indicating that surveys and experiments not only accurately capture consumer WTP within the context of the artificial decision environment of the research study, but also consumer behavior in real-world markets, there are still concerns and drawbacks surrounding these methods. Their static one-time assessment nature does not facilitate a robust assessment of the dynamics of consumer preferences toward GM nor how information, product experience, and social interaction shape future decisions.

Ultimately, despite the already substantial literature assessing consumer WTP for GM foods, there is a need for more research to help our understanding and aid researchers and policymakers in their efforts. Given the critical role consumer GMO preferences play in determining optimal agriculture, labeling, and trade policies, efforts to further refine and improve our understanding will help improve policymaking, markets, and research and development efforts. Better capturing consumer prior information and beliefs is critical for assessing the impact of information on WTP. Moving away from one-time valuation elicitation mechanisms toward repeated measurements of consumers WTP would aid in understanding the stability and dynamics of consumer preferences for GM. This is particularly important for developing countries where policymakers and consumers face a rapidly changing agricultural and information landscape and are confronted by the crucial question of whether to allow domestic GM production. Better understanding of differences in preferences across demographic groups and the distribution of the benefits and costs of GMO policies is also needed. Finally, and perhaps most importantly, moving away from artificial surveys and experimental environments toward assessments of consumer behavior and outcomes in actual markets would be tremendously beneficial.

Disclosure of Potential Conflicts of Interest

No potential conflict of interest was disclosed.