AEA RC T Registr y

The American Economic Association's registry for randomized controlled trials

Investor overconfidence helps to make sense of many stylized facts in financial markets, such as excessive trading volume and asset price volatility. The primary goal of this project is to study a process in which investors become overconfident about their ability to invest. Our hypothesis is that investors update their beliefs about their ability mainly by observing realized gains and losses (rather than paper gains and losses). We design and conduct experiments to test this hypothesis.

The learning process described above has the potential to give rise to investor overconfidence. The disposition effect, an empirically robust pattern documented by prior research, says that investors have higher propensities to realize gains than to realize losses. In conjunction with the hypothesis that investors learn about their ability to trade by observing realized gains and losses, the disposition effect implies that investors tend to become overconfident after a sequence of investment choices.

Our study includes two experiments. In Experiment 1, subjects first go through an investment task that asks them to make investment choices over multiple periods. After each period, the computer sells a stock from subjects’ portfolios. Between subjects, we exogenously manipulate whether the computer sells winning stocks (Selling Gains treatment) or losing stocks (Selling Losses treatment). After this investment task, we elicit subjects’ beliefs about their investment ability to test whether our treatments affect subjects’ level of confidence. We then ask subjects to perform a second investment task to investigate whether the treatments from the previous task influence subjects’ subsequent investment choices.

Experiment 2 has the same design as Experiment 1, except that subjects do not make active investment choices in the first investment task. Rather, subjects get stocks randomly assigned to their portfolio and merely observe price changes of these stocks. In this setting, our treatment task (the first investment task) does not depend on subjects’ own investment ability. Experiment 2 tests whether the manipulation of gain or loss realizations in such a passive setting affects subjects’ beliefs about the quality of the observed assets as well as their subsequent investment choices.

The results from these two experiments will contribute to a large literature in economics that studies the role of overconfidence in financial decision making.