In this paper we report the results of a laboratory experiment, in which we observed the behavior of agents in a simple macroeconomic setting. The structure of the economy was only partially known to the players which is a realistic feature of our experiment. We investigate whether subjects manage to approach optimal behavior even if they lack important information. Furthermore, we analyze subjects' perceptions of the model and whether their behavior is consistent with their perceptions. The full information model predicts changes of employment correctly, but not the level of employment. In the aggregate, subjects have correct perceptions, although individual perceptions are biased. We finally show that deviations from the full information solution are due to optimization failures than than misperceptions.