Discussion of “Robustness and Pricing with Uncertain Growth”

Abstract

1. Why Robustness? Hansen and Sargent, and a number of coauthors have recently initiated a research agenda that introduces the notion of robustness to model uncertainty (or a concern with model misspecification) into macroeconomics and finance. Methodologically they modify key techniques from the robust control literature in applied mathematics in order to allow for recursivity, discounting, and stochastic dynamics, which are obvious ingredients of the conventional modeling paradigm in financial economics. The fundamental idea behind robustness is that economic models are at best viewed as stylized approximations of reality rather than perfectly accurate descriptions thereof. It is then natural to entertain the possibility that economic agents making decisions in these models share the concern for model misspecification that economists and econometricians routinely express. When agents use a particular model as guidance in a dynamic decision problem and worry that this model might be misspecified, one would expect them to insist on considering alternative models in order to obtain decision rules that not only work well when the particular baseline model is correct, but that also perform reasonably well when the model is somewhat misspecified. This is the sense in which the decision rule is designed to be robust. The question of course is how agents go about considering alternative models and obtaining a robust decision rule given these alternative models. In this article, as in other work introducing robustness, the agents are assumed to achieve robustness by considering a least-favorable or worst-case model that is similar to and statistically hard to distinguish from the baseline model.