Antecedents and consequences of heuristic biases: Evidence from individual investors and small business owners.

摘要

Modern financial economics assumes (rather naively) that we trade rationally but we do not. The theoretical models in securities trading that address this paradigm are built on several assumptions, one being that overconfident traders trade too much. Largely using the research of Kahneman and Tversky and the current body of research in the field of critical thinking, this study analyzes the role that seven psychological attributes play in the tendency to trade stocks. Controlling for a host of variables including age, gender, investment experience and education we directly test several hypotheses by regressing several measures of trading volume of individual investors (trading frequency, types of securities traded) with individual heuristic biases scores.; Using survey data comprised of heuristic questions, the Short Form Watson-Glaser Critical Thinking Skills Appraisal (WGCTA), investor trading frequency and types of securities traded as well as demographic data, this study looked at four distinct groups: members of AAII and NAIC, MBA students in an accelerated adult program, small business owners and firefighters. This study tested not only the three overconfidence heuristics often found in the literature but also three heuristics that fall under the rubric of representativeness and the additional heuristic of frame dependence. In addition, this research introduces the added dimension of critical thinking as a possible antecedent explanatory variable used to distinguish those who fall prey to heuristic bias and those who do not. Regression models were developed to predict whether critical thinking as measured by the WGCTA attenuated the likelihood of falling prey to heuristic bias. Controlling for a host of variables including age, gender, investment experience and education we directly test several hypotheses by regressing several measures of trading volume of individual investors (trading frequency, types of securities traded) with individual heuristic biases scores.; Contrary to the predictions of the overconfidence models, we found no evidence that men are not more overconfident than women as women have about the same level of overconfidence and trading frequency as do men and gender is not a useful explanatory variable of trading frequency in a multivariate regression. The series of multiple regression analyses demonstrated that the WGCTA scores explained a statistically significant proportion of variance in respondents' scores on the heuristic biases of miscalibration, illusion of control and the conjunction fallacy. In another surprising turn of events we also find that the most experienced and active traders suffer from the "worst than average" effect when it comes to evaluating their stock picking skills.