To better understand individual decision-making, economists have increasingly recognized the importance of subjective expectations and the necessity of disentangling beliefs and preferences. This exercise is useful in settings where behavior appears puzzling--for example, entrepreneurship. The high failure rate of new businesses is well-documented and yet individuals continue to forego paid employment to open their own business. This project uses laboratory experiments to disentangle beliefs and preferences to understand why individuals tend to overestimate the probability of high payoff outcomes. In settings, such as entrepreneurship, it is common to assume that an individual overestimates the probability of success because he overestimates his abilities. However, it is also possible that he is simply optimistic and believes good things are more likely to occur. We isolate beliefs about performance from optimism to test the prevailing assumption that the two are independent and to ask to what extent overestimation of the probability of success is driven by performance beliefs versus optimism. We find that both optimism and performance overestimation drive individuals to overestimate the probability of success; however, women are driven more by optimism than men. We also find that performance overestimation and optimism are highly correlated--individuals who are optimistic are also more likely to overestimate their performance. Next, we will examine the relationship between preferences for risk-taking on optimism and performance overestimation, as well as test whether individual optimism varies as the stakes involved change.
We employ a 2x2 (Payment x Performance) within-subject experimental design, where subjects are asked to report their beliefs about probabilistic events with known distributions and are incentivized for their accuracy via the quadratic scoring rule. To identify optimism, the payment treatment maps outcomes of a random draw from the distribution to monetary payoffs, varying whether specified outcomes are payoff favorable. A subject is optimistic if a he believes an outcome is more likely to occur when the outcome is payoff favorable than when it is not. To identify performance overestimation the performance treatment varies whether the subject's performance on IQ questions can affect the distribution or whether the distribution is exogenously given by the experimenter. When a subject answers an IQ question correctly, without feedback, he is facing a better distribution in the sense of stochastic dominance. The extent to which a subject overestimates his performance is measured by the weight he puts on the probability of being in a better distribution than his performance actually warrants. Because the quadratic scoring rule is only incentive compatible assuming risk-neutrality and preferences given by expected utility, we will correct for violations of these assumptions by eliciting risk attitudes and jointly estimating risk preferences and subjective probabilities. We will also examine the nature of the relationship between optimistic beliefs and preferences by varying the intensity of the induced preference in the payment treatment.
