Monetary Policy Shocks and Security Market Responses

Abstract

This paper contributes to a recent literature that tries to filter exogenous monetary policy surprises from high frequency (daily) data. The literature uses the fact that monetary policy surprises are realized only on days that the Federal Reserve changes the Federal Funds Target, or on days that the Federal Open Market Committee (FOMC) meets and does not change the target—so-called days. We add to the literature in three ways: (1) we specify a more general model in which security prices respond to two sources of systematic risk (a two factor model)—a common information shock and the monetary policy shock—plus nonsystematic risk—an idiosyncratic shock. (2) We use all of the daily data while other studies use only a small sub-sample of less than 10% of the data. And (3) we use new estimation strategy that gives consistent and more efficient parameter estimates than previous studies. Our empirical results show that efficiently estimating a more general model leads to important differences. Common shocks have an important and statistically significant impact on bond yields at all maturities. Leaving out the common information shocks leads to bad estimates of the impact of monetary policy shocks. Cochrane and Piazzesi's event study found that the yield on a 10 year bond increases in response to a positive Target surprise—which they properly label a puzzle. The factor model solves the puzzle. We get the classic textbook response—a surprise increase in the target rate leads to a decline in the yield on the 10 year bond. Finally we look at equity returns as well as bond yields which introduces a different puzzle. In any model we look at a positive Target surprise causes a large and significant decline in equity returns. A recent event study paper of equity market responses by Bernanke and Kuttner (2003) finds similar results. The question is why?