Unbounded behaviorally consistent stopping rules

In this article we study behaviorally consistent stopping rules in an unbounded search from a known distribution with no recall and with positive search cost. We show that if the searcher's preferences are quasi-convex in the probabilities, then behaviorally consistent search strategies in the unbounded case are obtained as limits of the corresponding bounded search strategies and are characterized by reservation levels property. Unlike optimal stopping rules under expected utility theory, however, the reservation levels may not be monotonically increasing in the number of permissible stages of the search process, and, in the unbounded case, may not be unique.