Experimental Economics: Financial Markets, Auctions, and Decision Making

Attitudes to inequality and risk can be interpreted in a number of ways. A standard approach is to think in terms of strength of inequality aversion or risk aversion. There are several studies which examine these questions from the standpoint of economic orthodoxy in these two fields, and some which investigate risk and inequality jointly within the standard theoretical framework for analyzing income distributions.¹ However such an approach presumes acceptance of the basic premises of this framework. In this paper we examine a deeper question that addresses the basis for statements about inequality or risk comparisons without incorporating a priori the key assumptions that impose structure on inequality indices, measures of risk or other distributional criteria. Instead of posing the question “what is the degree of inequality (risk) aversion”

Here, I would like to raise a couple of questions with respect to methods commonly used in Experimental Economics.

The experimental asset market literature provides strong evidence concerning the robust convergence of transaction prices to the competitive equilibrium price and the high percentage of gains from trade exhausted in double auction (DA) markets. However, the designs of previous experimental asset market studies have incorporated trading period durations that are constant and known to the trading participants. Friedman (1984, p. 71) suggests that the predetermined, known time at which trade will cease is one of a number of institutional features of experimental DA markets that enhance the informational and competitive efficiency of observed outcomes. The intention here is to extend previous work (Duxbury, 1997) by conducting a series of experiments designed to determine the importance of trading period duration on observed market performance. To this end, markets are conducted with either constant and known (CK), variable and known (VK) or variable and unknown (VU) trading period durations. The VK treatment represents a novel feature of the experimental design, allowing the analysis to differentiate between the effects of variable trading period duration and uncertain trading period duration. The few theoretical models of bid, ask and transaction price behavior in experimental DA markets developed to date rely on assumptions concerning the time remaining until the end of trade (see for example Friedman, 1991).

The traditional assumption in economics is that decision makers are (perfectly) rational and that, at least when game theory is involved, this is commonly known. Rationality means to react optimally to one’s decision environment. The orthodox interpretation assumes that material success is maximized an assumption that has been convincingly falsified, partly by experimental evidence. Recently more general notions of preferences are propagated which allow to capture non-material rewards like emotions, social concerns or intrinsic motivation. In any case decision alternatives are compared according to their present and future consequences. The past only plays a role in so far as it determines structural aspects, e.g. via state variables like in dynamic programming or in dynamic games. Thus rational decision making is purely forward looking.

It has been standard in economic theory that agents are modeled as self-centered individuals pursuing their own material interest only. In recent years increasing evidence, however, from economic experiments in particular, suggests that subjects are also motivated by concerns for others. Although fairness has been taken to be a main explanatory variable for certain observed deviations of behavior from game theoretic predictions (cf. Fehr and Schmidt, 2001), fairness is not defined in a consistent way in the economics literature. Fehr and Schmidt (1999) modeling fairness as self-centered inequity aversion assume inequality to be the relevant notion of inequity. Rabin (1993), in turn, employs beliefs about opponents’ intentions in characterizing fairness, extending the notion of fair behavior to include reciprocal fairness (Camerer, forthcoming; Fehr et al. 1996, Fehr etal. 2000). Many authors, on the other hand, fail to specify how fairness is defined. Equality is often implicitly assumed to be the reference point for fairness judgments (cf. the survey by Fehr and Schmidt, 2001). Sometimes, however, determining a reference point is regarded as more complicated, and authors ask their experimental subjects to state their perception of fairness (e.g. Gächter and Riedl, 2001; Babcock at al., 1995; Kahneman et al., 1986 a, Kahneman et al., 1986 b; Yaari and Bar-Hillel, 1984).

Sequential prisoners’ dilemma games can be regarded as formalizations of many everyday situations involving trade-offs between private and collective interest. Examples of such situations are donating to fund-raising campaigns in which donors can observe what others have donated earlier; efforts to provide collective goods of society by listing volunteers, or team production where workers observe their co-workers’ efforts.

This paper is an experimental study of a repeated first price sealed bid auction with private values. Values are drawn independently each round and are private information. In each round of the game, players are allowed to send simultaneously messages about their current value. These messages are cheap talk and have no direct payoff consequences. Moreover, communication is anonymous (players do not know the identity of their opponent) and structured (the set of possible messages is limited). Players are allowed to stay away from bidding, while retaining the possibility to bid in later rounds; in this way the good can be allocated to the bidder with the highest value at the lowest possible bid. After the object is allocated, the winner can make a sidepayment to the loser; sidepayments can be used to provide incentives for a low value player to stay out from bidding in favor of a high value player.