Comment on Poirier: Operational Bayesian Methods in Econometrics

Abstract

[This paper responds to “Frequentist and Subjectivist Perspectives on the Problems of Model Building in Economics,” by Dale J. Poirier, in this same issue.] My purpose in this comment is to illustrate that economists can indeed act upon the agenda implied by Poirier's five pragmatic principles of model building. Many economic researchers have had a subjective view of the world for a long time. But now, the argument that Bayesian approaches do not lead to operational methods, often decisive one or two decades ago, is becoming irrelevant in rapidly increasing numbers of applications as cheap and massive desktop computing power spreads. For the foreseeable future there will be substantial returns to developing appropriate numerical methods of Bayesian inference in econometrics, and to using these methods in empirical work. Many econometric problems that are messy and intractable from a classical perspective can become elegant and straightforward from a Bayesian perspective. I am not interested in defending this statement as an abstract proposition. Rather, I am interested in providing solutions to problems that have been stubbornly immune to treatment by classical methods. I will take up two of the scores of such problems here.