We investigate claims made in Giacomini and White (2006) and Diebold (2015) regarding the asymptotic normality of a test of equal predictive ability. A counterexample is provided in which, instead, the test statistic diverges with probability 1 under the null.

Treatment for depression is complex, requiring decisions that may involve trade-offs between exploiting treatments with the highest expected value and experimenting with treatments with higher possible payoffs. Using patient claims data, we show that among skilled doctors, using a broader portfolio of drugs predicts better patient outcomes, except in cases where doctors' decisions violate loose professional guidelines. We introduce a behavioral model of decision making guided by our empirical observations. The model's novel feature is that the trade-off between exploitation and experimentation depends on the doctor's diagnostic skill. The model predicts that higher diagnostic skill leads to greater diversity in drug choice and better matching of drugs to patients even among doctors with the same initial beliefs regarding drug effectiveness. Consistent with the finding that guideline violations predict poorer patient outcomes, simulations of the model suggest that increasing the number of possible drug choices can lower performance.

Consider a researcher estimating the parameters of a regression function based on data for all 50 states in the United States or on data for all visits to a website. What is the interpretation of the estimated parameters and the standard errors? In practice, researchers typically assume that the sample is randomly drawn from a large population of interest and report standard errors that are designed to capture sampling variation. This is common even in applications where it is difficult to articulate what that population of interest is, and how it differs from the sample. In this article, we explore an alternative approach to inference, which is partly design‐based. In a design‐based setting, the values of some of the regressors can be manipulated, perhaps through a policy intervention. Design‐based uncertainty emanates from lack of knowledge about the values that the regression outcome would have taken under alternative interventions. We derive standard errors that account for design‐based uncertainty instead of, or in addition to, sampling‐based uncertainty. We show that our standard errors in general are smaller than the usual infinite‐population sampling‐based standard errors and provide conditions under which they coincide.