Estimating the Extent of Trade Under Incomplete Information: The Case of HIV
19 Pages Posted: 1 Feb 1997
Date Written: Undated
This paper discusses problems and prospects in the empirical analysis of the effects of incomplete information on mutually beneficial activities when direct measures on information are available from traders. In particular, we focus on sexual behavior in the presence of information asymmetries about HIV status, which, we argue, shares many features with other markets involving mutually beneficial activities under asymmetric information. We argue that the misreporting incentives between partners that cause the asymmetric information between them may also induce measurement error when sampling them; this motivates the sampling of only one of the partners. This leads us to discuss what aspects of the allocation of information can be identified by such one-sided sampling. The paper shows how well the joint distribution of information within a couple can be identified under one- sided sampling by deriving the bounds within which the joint distribution must fall. We apply these bounds on beliefs delivered by one-sided sampling to a dataset on 5,397 couples in San Francisco in 1988-89 and find that there is virtually no trade taking place under asymmetric information; all couples either being fully informed or mutually uninformed. This substantive conclusion is robust with respect to the other identifying restrictions we impose, such as those of standard models of signaling.
JEL Classification: I1
Suggested Citation: Suggested Citation