The value of information in a sealed bidding experiment

Abstract The present study assesses directly the value of perfect information from two different data sources to a decision maker confronted with the sealed bidding task. Predictions about the value of information that eliminates uncertainty and the subsequent bid are derived following Howard's ( IEEE Transactions of Systems Science and Cybernetics , 1966 , SSC-2 , 22–26) analysis of the value of information. A 2 × 4 × 2 factorial design varied (i) the distributions of the data sources, (ii) the kind of information purchasable, and (iii) the order of presentation for blocks of trials where information could or could not be purchased. Each of 80 subjects performed the sealed bidding task for a series of 72 trials. The results show an almost linear relationship between the observed maximum price offered for information and the prescribed value of the information. Furthermore, subjects consistently underbid in all information conditions. Deviations from optimality are interpreted in terms of the subjects' limited information processing capacities and the nature of the decision task.