UC Santa Cruz

In this work, I study the behavior of boundedly rational agents in dynamic stochastic settings. The first chapter develops a new laboratory test of the hypothesis that individual investors have an irrational preference for selling winning stocks vs. selling losing stocks. Subjects tend to delay selling losers beyond the optimal point and to sell winners before reaching the optimal liquidation point. Such

behavior is shown to be consistent with a model of reference-dependent preferences.

The second chapter presents the results of a lab experiment on a menu cost model of price adjustments. Subjects' average adjustment points show significant time dependent components in adjustment rules. The data suggest subjects experience substantial cognitive costs from responding to the state, accounting for these patterns.

The third chapter presents techniques for studying the dynamics of evolutionary games that are subject to stochastic disturbances and applies them to morph frequency data from a population of side-blotched lizards. The estimated payoff matrix implies a powerful interior attractor and recover the observed oscillations.