林逸軒 Yi-Hsuan Lin

Identification and Testability in Models of Decision-Making under Uncertainty

My research mainly studies identification and testability for various models of choice under uncertainty. Identification is concerned with whether the parameters of a specific model are uniquely recoverable from observable behavior. Testability is concerned with how to test the consistency between a model's predictions and choice data. These two issues are fundamental for any economic model because, without them, we lack the connection between abstract economic theories and real human behaviors.

Specifically, my research contributes to the literature on so-called decision theory, which mainly addresses choice behaviors at a single individual level. My publication "Stochastic Choice and Rational Inattention" provides a non-parametric identification for information cost function, the key ingredient of rational inattention models, widely applied to analyze decision-making under information acquisition at both micro and macro levels. In another publication, "Costly Information and Random Choice," my coauthor and I study two classic types of information cost: additive cost and multiplicative cost. We give necessary and sufficient conditions on random choice for it to be consistent with either of the cases, and we discuss how to identify the underlying information structure as well as its cost. Finally, in "Identification and Welfare Evaluation in Sequential Sampling Models," my coauthor and I consider a decision maker who can learn information through the technology of sequential experiments. In many economic models of dynamic information acquisition, learning costs are usually multiplicative or additive. In either case, we study how random choice with decision time can reveal the learning cost and recover a decision maker's ex-ante welfare without imposing any assumptions on the underlying sequential experiment.