Behavioral New Keynesian Models: An Empirical Assessment

Abstract

This paper estimates and compares three recent Behavioral New Keynesian models: the finite horizon planning model proposed by Woodford (2018), the cognitive discounting model proposed by Gabaix (2020), and the imperfect common knowledge model proposed by Angeletos and Lian (2018). I show that the data prefer the former two behavioral models, while I find less support for the model with imperfect common knowledge. Secondly, I show that cognitive discounting modeled à-la Gabaix (2020) is able to capture psychological elements in a mathematically tractable way and can be trivially introduced in any linearized DSGE model. This is not the case for other behavioral approaches given that most derivations, if not all among those proposed up to today, are algebraically too intensive to be of general applicability. I use this finding to introduce cognitive discounting into the Smets and Wouters (2007) model, hence building and estimating the first micro-founded behavioral medium-scale DSGE model.

Keywords: Behavioral New Keynesian model, bounded rationality, finite horizon planning, cognitive discounting, common knowledge, Bayesian estimation.