Some Unorthodox Thinking and Observations

In this talk, I will share two of my recent studies on very different subjects. In the first study, I try to connect the behavioral bias in economics with the bias-variance tradeoff in statistics via the James-Stein Shrinkage. Specifically, James–Stein shrinkage demonstrates that, by aggregating unrelated tasks and leveraging supposedly irrelevant information, the decision maker may improve an unbiased decision by “shrinking” it toward an arbitrarily chosen reference point. This revelation points to the difference between probability theory and statistical inference, and it leads to novel interpretations of “rational vs. irrational” and testable hypotheses for estimation problems. In the second study, motivated by an online platform that deliberately limited its set of suppliers, we develop an analytical model that addresses the underlying market frictions. We identify circumstances where the classic reputation mechanism is insufficient and demonstrate that by accepting only a subset of suppliers, the platform can support a welfare-enhancing equilibrium even if all the suppliers are homogeneous and the customers can transact with suppliers outside the platform. Our work enriches the market frictions–based perspective by showing how it can guide platform design in governing economic exchanges.