Substitutability, experience, and the value disparity: evidence from the marketplace

https://doi.org/10.1016/j.jeem.2003.05.001Get rights and content

Abstract

Recent empirical evidence suggests that important disparities exist between willingness to pay and compensation demanded for the same good. These results, which clearly contradict closely held economic doctrines, have led some influential commentators to call for an entirely new economic paradigm to displace conventional neoclassical theory. This study examines the generality of these experimental findings by going to a well-functioning marketplace and examining more than 350 individual decisions across two incentive-compatible elicitation mechanisms. The data suggest that behavior of individuals with intense experience approaches neoclassical predictions: any observed WTA/WTP disparity amongst this group is negligible.

Introduction

A fundamental postulate of neoclassical theory is that preferences between two goods are independent of the consumer's reference point. This basic independence assumption naturally leads to the prediction that with small income effects and many available substitutes, the willingness to accept (WTA) and willingness to pay (WTP) measures of value for a commodity should be roughly equivalent. While most theoretical and applied economic models invoke this assumption, it has been directly refuted in several experimental settings (e.g., [9]).1 Some commentators have argued, however, that these findings represent merely a mistake made by inexperienced consumers and that through time the behavior of those consumers will be more in line with predictions from neoclassical models (e.g., [3], [5], [15]). While this argument has intuitive appeal, critics contend that the available evidence is mixed at best, and that overall the data do not conclusively support the learning premise [13], [14].

In this study, I take a different approach to understanding the market learning process by observing actual market behavior. This approach provides me with an opportunity to observe behavior of agents that have endogenously selected into certain roles within the market, such as being an intense or casual consumer, while simultaneously making use of certain controls afforded in an experiment. This methodology may lead to different results than in an experiment where the roles are exogenously induced by the experimenter (i.e., some subjects are given experience while others are not), but a rigorous examination of behavior in an actual environment is an important step in testing the validity of economic theory.

The specific goals of this study are twofold. First, I compare WTA statements of value with WTP and “endowment-adjusted” WTP values gathered via two incentive-compatible institutions: a random nth price auction and the Becker et al. [2] discrete choice auction.2 Standard neoclassical arguments suggest that the results of several previous experimental tests that found a disparity between WTA and WTP values (e.g., [5], [13]) may be consonant with neoclassical theory after all, since these studies compared value estimates from different indifference curves and thus different marginal rates of substitution applied.3 By comparing WTA and endowment-adjusted WTP, I control for income and substitution effects and therefore provide a more appropriate comparison of value measures. This distinction allows a cleaner test of alternative theories. Second, I push the substitutability argument a bit harder by examining whether exogenous changes in substitutability influence the WTA/WTP disparity.

The empirical results provide some interesting insights. First, I find that intense consumers (i.e., those consumers that have a history of active market participation) display a negligible value disparity in both elicitation institutions. This result is in accord with List [17], [18], and provides a robustness check of his main findings in much different institutional settings. Second, I find that neoclassical theory does an adequate job of organizing the data from experienced consumers, as the comparative static predictions are generally met. For example, (i) endowment-adjusted WTP better approximates WTA than does WTP, and (ii) the estimate of the effect of income changes on WTP provided by the observed WTA/WTP disparity is more reasonable for intense consumers than for casual consumers. Furthermore, if one takes into account informational asymmetries, neoclassical theory provides an adequate explanation for patterns observed in the pooled data. Third, I find some evidence that exogenous changes in substitutability influence the degree of the WTA/WTP disparity.

The remainder of this study proceeds as follows. Section 2 provides theoretical structure to the experimental design. Section 3 summarizes the empirical results. Section 4 discusses limitations of the study. Section 5 concludes with a summary and directions for future research.

Section snippets

Background and experimental design

Ever since the initial experimental findings suggesting that mere ownership of a good may induce a reference-dependent preference structure, neoclassical models have been under heavy scrutiny, as their basic independence assumption is clearly violated in such scenarios. In addition to casting doubt on closely held economic doctrines, influential commentators have used the disparity findings to question the entire basis of welfare economics and common interpretations of indifference curves. From

Results

To put my findings into perspective, I first review the results of List [17], who controlled for Hicksian income and substitution effects by asking subjects to choose between bundles—in one treatment the subject is endowed with good A and has the option to trade it for good B; in a second treatment, a different subject is endowed with good B and has the option to trade it for good A. Pooling his data across more than 300 subjects yields Fig. 5, which makes the trade probability a function of

Potential limitations

While the above results provide insights that may be interesting to economists and policymakers, it is important to recognize the inherent limitations in the experimental design. First, recall that all subjects randomly allocated into the WTP endowment-adjusted treatments received the average offer from the appropriate random nth price auction: each WTPC1 subject received $13.50 and each WTPC2 subject received $16.75. Considering the mean WTA values across subject types (see Table 3), some

Concluding remarks

Substantial-experimental evidence indicates that a non-trivial gap exists between WTA and WTP statements of value. These findings have induced even the most ardent supporters of neoclassical theory to doubt the validity of certain neoclassical modeling assumptions. Now that experimental economists have obtained several unique insights on the value disparity, however, it is important to examine whether these experimental findings generalize to the types of environments that theorists’ models

References (26)

  • W.M. Hanemann

    The economic theory of WTP and WTA

  • D. Kahneman et al.

    Experimental tests of the endowment effect and the coase theorem

    J. Polit. Economy

    (1990)
  • C. Kling, J.A. List, J. Zhao, The WTP/WTA disparity: have we been observing dynamic values but interpreting them as...
  • Cited by (0)

    I would like to thank Don Fullerton, three anonymous reviewers, and the Editor for very constructive comments on an earlier version of this paper. The NBER provided funds to support this research. Colin Camerer, Spencer Banzhauf, Bob Deacon, Shelby Gerking, Michael Greenstone, Danny Kahneman, Gib Metcalf, Jeffrey Racine, Kerry Smith, Mark Strazicich, Robert Sugden, and Aart de Zeeuw provided insightful discussion of the important issues during this research agenda. Seminar participants at the University of Arizona, Cornell University, Iowa State, Pennsylvania University, University of Maryland, North Carolina State, Tilburg University, and the University of South Florida also provided useful comments. Conference participants at the EAERE meetings in Greece, the NBER meetings in Cambridge, as well as the ESA meetings in Tucson also provided useful suggestions. Any errors remain my own.

    View full text