Information Processing Biases in Experimental Asset Markets

点击翻译按钮获取中文摘要

Are market prices a good estimate of value? Economists routinely answer yes, but other social and behavioral scientists are less certain. An abundant literature in psychology, political science, and sociology finds judgement bias in the processing of information by individuals. This project examines whether judgement bias about the value of an asset can affect trading behavior in markets. Laboratory experiments are used to test for the existence of judgement bias. A procedure is developed for measuring the cost of judgement bias in a laboratory setting. In practical situations one normally would not be able to control the amount of judgement bias, hence there is a need for methods that measure the cost of judgement bias in the absence of experimental control. These empirical methods are developed and two methods of validation are used. Market prices might be correctly established in the presence of judgement bias as long as some group of traders are sufficiently well-informed and act as arbitragers. This possibility is examined as part of the experimental design. The ability of markets to accurately forecast future events is often taken for granted by economists despite demonstrations by psychologists of the existence of substantial judgement biases by individuals in performing the tasks necessary for the efficient operation of markets. To an economist, the psychological evidence about these biases is often unconvincing. The experimental studies which document their existence frequently use unpaid students who are required to participate to earn course credit. Thus participants have nothing at stake. If these biases do exist at the individual level, can they persist in a market setting where individuals have financial incentives and can gain experience to avoid mistakes? Finally, if these biases do exist are they important in any meaningful sense? Persistent statistical evidence of widespread judgement bias would be of little importance if judgement bias cost an individual only one cent. This project is important because it develops rigorous measures of the cost of judgement bias and tests for the presence of significant judgement bias in experimental asset markets.