ON THE DEMAND FOR INFORMATION

Most previous research on the economics of information has been based on the expected utility hypothesis and Bayesian learning. Yet there is experimental evidence that decision makers do not always maximize exp ected utility and that human learning is not always consistent with Ba yes' rule. Using a two-period model, this paper examines the valuation of information as well as the demand for information in the broader c ontext of a state preference approach under an ordinal representation of preferences. In particular, it is not assumed that the decision mak er is Bayesian, nor that he behaves in a way consistent with the expec ted utility hypothesis. In this general framework, the value of inform ation is defined and analysed. Also, under active learning, the optima l allocation of information gathering activities is discussed. Behavio ural properties of the demand for information are derived making use o f a compensation function. Implications of the results for measuring t he value of information and for the economic analysis of learning acti vities are discussed.