Optimal Search

From MaRDI portal





This paper presents a general model of search in which the searcher maximizes the expected utility of search by choosing not only the number of periods of search but also the size of sample to be taken each period. Both no-recall and full-recall cases are allowed and the cost of search includes not only financial but also psychic cost. The optimal search rule is shown to require the searcher to compare the net benefits of stopping to those of continuing search at each t. The optimal search rule is shown to be nonunique but exists under conditions which ensure that the expected present value of search is always finite. The paper shows that the fixed-sample-size (FSS) rule is optimal if the marginal search costs or discounting are sufficiently large, so that either the value of continuing search in the second period is not greater than the currently available utility or no observation will ever be taken after the first period. The sequential search rule is optimal if the decision horizon is infinite, the searcher enjoys full recall and has zero time preference, the fixed costs of search are zero, the marginal search costs are nonnegative and increasing, and the utility function is strictly increasing with wealth. However, in general the optimal search rule dominates the above two well-known search rules. Therefore, these familiar rules may not be good approximations to an optimal policy.




Cited in
(55)








This page was built for publication: Optimal Search

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3688055)