Controlling for initial endowment and experience in binary choice tasks

Learning literature typically assumes that initial attractions to choose each possible alternative are given exogenously. However, evidence shows that current behaviour depends on past experiences. In this paper, we design an experiment to control for the initial experience in decisions from experience by providing decision makers with an exogenous history (successful vs. unsuccessful) prior to initiating the decision task. Moreover, varying the initial endowment level for fixed histories we investigate the income effect. We are also interested in analysing the duration of both effects (history and income). We find significant treatment effects in the sense that more risk taking behaviour is associated with good histories and with low income levels. According to previous literature, our results confirm the transitory nature of both effects, although the duration of the income effect doubles the duration of the history effect. In the long run, risky choice behaviour converges across different treatments.