Collaborative Research: Sunspots in the Laboratory

点击翻译按钮获取中文摘要

Economists have developed a theory of "sunspot" equilibrium to describe situationswhere economic fluctuations are not driven by changes in market fundamentals, but areinstead due to extrinsic, non-fundamental factors. This project involves the design of anexperiment that will explore whether and how such extrinsic factors can result in marketfluctuations, and whether the particular market institution also plays a role.The experiment consists of two main treatments. In the first, human subjects participateas buyers or sellers in a computerized "double auction", where goods can be bought andsold continuously over a fixed trading period at mutually agreed upon prices that areobservable as they occur to all participants. In the second "call market" treatment,subjects submit private bids and offers, and a centralized mechanism determines thesingle, market clearing price, as well as the quantities that each participant can buy orsell. Thus, the only information subjects receive in this environment is the resultingmarket price and the quantity they were able to buy or sell at this price.In both treatments, subjects face uncertainty in that they are given two sets of values orcosts for the goods they can buy or sell; these two sets correspond to the"state" of theeconomy, and are referred to as "high" or "low" in reference to the equilibrium price inthe two states. (the equilibrium quantity is the same in both states). Initially, subjects aretold the state of the economy in advance, so that they become familiar with the two statesfor prices. After several rounds, subjects are informed that the true state of the economy,will henceforth be determined expost, after all trading is completed. Furthermore, the rulefor the determination of the state will depend on the median trade price in the doubleauction environment or on the analogous market clearing price in the call marketenvironment. If this price lies within a proscribed domain of the equilibrium low price,then the state is declared to be low and if it lies outside this domain, in that of theequilibrium high price, then the state is declared to be high. Subjects are provided with apotential coordination device (or sunspot realization) which takes the form of a randomannouncement by the experimenter of the forecast for the state of prices in the currentperiod. Subjects are instructed that this announcement has no binding consequences.Preliminary findings suggest that subjects coordinate perfectly on the announcement inthe highly centralized call market environment but fail to do the same in the moredecentralized double auction environment. These experimental findings if replicated,have significant implications for the architecture of real financial markets.