Working Paper: NBER ID: w10916
Authors: William A. Brock; Steven N. Durlauf; Kenneth D. West
Abstract: This paper explores ways to integrate model uncertainty into policy evaluation. We first describe a general framework for the incorporation of model uncertainty into standard econometric calculations. This framework employs Bayesian model averaging methods that have begun to appear in a range of economic studies. Second, we illustrate these general ideas in the context of assessment of simple monetary policy rules for some standard New Keynesian specifications. The specifications vary in their treatment of expectations as well as in the dynamics of output and inflation. We conclude that the Taylor rule has good robustness properties, but may reasonably be challenged in overall quality with respect to stabilization by alternative simple rules that also condition on lagged interest rates, even though these rules employ parameters that are set without accounting for model uncertainty.
Keywords: Model Uncertainty; Policy Evaluation; Bayesian Model Averaging; Monetary Policy; Taylor Rule
JEL Codes: C5; E5
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| model uncertainty (D80) | policy evaluation (H43) |
| Taylor rule (E43) | robustness properties (C52) |
| optimized interest rate rule (E43) | dominates Taylor rule (E43) |
| introduction of model uncertainty (C59) | affect rankings of models (C52) |
| coefficients in model-specific optimal rules (C51) | model complexity (C52) |