Are Stated Preferences Good Predictors of Market Behavior? Maria L. Loureiro, Jill J. McCluskey, and Ron C. Mittelhammer ABSTRACT. Using an economic experiment in conjunction with a survey, we analyze whether consumers’ hypothetical willingness-to-pay re- sponses are effective predictors of actual mar- ket behavior. We model revealed preferences as a function of socio-demographic characteristics and instrumental variables that represent the in- tensity of stated preferences. Our ndings show that consumers who state that they are willing to pay a premium, which is equal to or greater than a positive lower bound, have a higher likelihood of actually buying the product in question. This implies that consumers’ actions in the economic experiment validate their survey responses. ( JEL Q13, Q26) I. INTRODUCTION Policymakers often must make decisions based on non-market valuation estimates ob- tained from either revealed preference (RP) or stated preference (SP) techniques. Ques- tions may vary widely, such as: How much money should be spent to avoid global warming? How much are Americans willing to pay to keep Lake Tahoe blue? 1 How much money should be spent on cleaning and maintaining ocean beaches? RP techniques, such as hedonic price analysis and the travel cost method, use actual consumer decisions to model consumer preferences and exploit the fact that consumer decisions reveal pref- erences for goods, in both market and non- market contexts. SP techniques, such as con- tingent valuation, contingent behavior, and choice experiments, ask people questions that are intended to elicit their preferences for a good or amenity, without requiring that the consumer act accordingly. If the RP and SP results are consistent, then from a policy perspective with limited budget resources, data can be collected in the Land Economics February 2003 79(1): 44–55 ISSN 0023-7639 ã 2003 by the Board of Regents of the University of Wisconsin System least costly way and still provide results that are consistent with other more costly meth- ods. Further, sometimes it will only be possi- ble to collect SP data. This will be the case when obtaining data on the valuation of cer- tain endangered species. Cross validation of approaches provides support for policy deci- sions that are based solely on results from SP data. SP methods are commonly criticized be- cause of the hypothetical nature of the ques- tions and the fact that actual behavior is not observed (Cummings, Brookshire, and Schulze 1986; Mitchell and Carson 1989). Adamowicz, Louviere, and Williams (1994) 2 criticize RP methods on the basis that the models of behavior developed constitute a maintained hypothesis about the structure of preferences that may not be testable. They also point out that RP methods can suffer from collinearity among attributes, preclud- ing identi cation of the marginal impact of factors that affect choice. Given these criti- cisms of both RP and SP approaches and that both are commonly used for policy formation and as evidence in courts of law, it is surpris- ing how relatively few studies analyze the The authors are, respectively, assistant professor, Department of Agricultural and Resource Economics, Colorado State University; assistant professor, Depart- ment of Agricultural and Resource Economics, Wash- ington State University; and professor, Department of Agricultural and Resource Economics and Program in Statistics, Washington State University. The authors wish to thank, without implicating, Phil Wandschneider, John Loomis, Paul Barkley, Tom Schotzko, and an anonymous reviewer for helpful ad- vice. The authors gratefully acknowledge the nancial support from the Federal-State Marketing Improvement Program (FSMIP) of the U.S. Department of Agricul- ture. 1 ‘‘Keep Tahoe Blue’’ is a popular bumper sticker referring to the degradation of the formerly crystal clear blue waters of Lake Tahoe on the California-Nevada border. 2 This study uses a multinomial logit, in which the structure of preference is far more complex.