Weak identification of forward-looking models in monetairy economics

Recently, single-equation GMM methods have become popular in the monetary economics literature, for estimating forward-looking models with rational expectations. We discuss a method for analyzing the empirical identification of such models that exploits their dynamic structure and the assumption of rational expectations. This allows us to judge the reliability of the resulting GMM estimation and inference and reveals the potential sources of weak identification. With reference to the New Keynesian Phillips curve of Galí and Gertler (1999) and the forward-looking Taylor rules of Clarida, Galí, and Gertler (2000), we demonstrate that the usual 'weak instruments' problem can arise naturally, when the predictable variation in inflation is small relative to unpredictable future shocks (news). Hence, we conclude that those models are less reliably estimated over periods when inflation has been under effective policy control.