Vector autoregressive measures of monetary policy: Issues and critiques.

Abstract

This dissertation contains three essays on the empirical measurement of post-war Federal Reserve policy. Each chapter utilizes the vector autoregressive methodology, questioning several results of the current literature. The first examines the price puzzle: a rise in aggregate prices in response to a contractionary monetary policy shock. Conventional wisdom maintains that commodity prices resolve the puzzle because they improve the Federal Reserve's forecast of inflation. However, an investigation of several dozen alternative indicators reveals little correlation between an ability to forecast inflation and resolution of the price puzzle. A sub-sample investigation finds that the price puzzle primarily affects the 1959--1979 period, and that none of the indicators---including commodity prices---resolve the puzzle then. By implication, commodity prices are unlikely to be proxying for supply shocks. These sub-sample results suggest a role for changes in the monetary reaction function. Recent discussions credit Fed policy for the favorable performance of the past fifteen years while blaming it for the 1970s stagflation. But few of these studies attempt to separate policy shifts from other changes in the economy. On the other hand, the VAR literature tends to ignore instability of the endogenous policy rule. Utilizing a common SVAR specification, the second essay finds evidence of a shift in the reaction function around the 1979--1982 Volcker disinflation. However, most variation in the economic dynamics can be attributed to parameter shifts in the non-policy portion of the model. Furthermore, the estimated policy rule differs substantially from a Taylor rule specification. The third essay returns to commodity prices, which usually are treated as exogenous with respect to contemporaneous (if not all) monetary policy actions. Recognizing that commodity prices, as forward-looking flexible prices, react to current and expected future monetary policy, I estimate a structural VAR model of commodity and U.S. consumer price indices. Equiproportional long-run responses of nominal prices to monetary (i.e. nominal) shocks yield identifying restrictions. A significant share of the co-movement of these series is due to (exogenous innovations in) monetary policy, including episodes more commonly attributed to supply shocks.