On the Identification of the Costs of Simultaneous Search

This paper studies the identification of the costs of simultaneous search in portfolio problems (Chade and Smith, 2006). We show that market shares data from a single market do not provide sufficient information to identify the search cost distribution in any interval, even if utilitydistributions are known to the econometrician. We then show that by pooling data from markets wherethe alternatives that similar decision makers confront vary, the search cost distribution and theutility parameters of the logit demand model can be identified.