Paying is believing: the effect of costly information on Bayesian updating

The supposed irrelevance of historical costs for rational decision making has been the subject of much interest in the economic literature. In this paper we explore whether individual decision making under risk is affected by the cost of information. To do so one must distinguish the effect of cost from self selection by individuals who value information the most. Outside of the laboratory it is
difficult to disentangle these two effects. We thus create an experimental environment where subjects are o¤ered additional, useful and identical information on the state of the world across treatments. We find a systematic effect of sunk costs on the manner in which subjects update their beliefs. Subjects over-weigh costly information relatively to free information, which results in a ‘push’of beliefs towards the extremes. This shift does not necessarily lead to behavior more attuned
with Bayesian updating. We find that an intensification of representativeness bias due to cost is the most likely explanation of our results.