Essays on Identification of Monetary Policy Shocks in Vector Autoregressive Models: Alternative Identification Schemes and Lag Structures.

Abstract

This dissertation is primary concerned with the sensitivity of the effects of monetary policy shocks across alternative identification schemes and lag structures. The four widely-cited identification schemes of Christiano, Eichenbaum, and Evans (1994; 1996), Strongin (1995), Bernanke and Mihov (1998), and the long-run restrictions approach pioneered by Blanchard and Quah (1989) are used. Also, three types of lag structures---symmetric, Keating-type, and Hsiao-type asymmetric lag structures---are employed. The first essay focuses upon a closed economy framework. The results indicate that impulse response functions for macro variables are often sensitive to identification schemes and lag structures. For a given lag structure, the Strongin, Bernanke and Mihov, and long-run restrictions schemes generate similar results, while the Christiano, Eichenbaum, and Evans scheme often yields different responses from others. This essay also illustrates that the Christiano, Eichenbaum, and Evans and long-run restrictions schemes are relatively insensitive to the type of lag structures. The second essay examines the effects of lag structure misspecification within a Monte Carlo framework. It is shown that the lag structure of a VAR model does matter when assessing the effects of monetary policy shocks. For most horizons, t-statistics support for the hypothesis that the responses from the misspecified VARs are significantly different from the assumed 'true' responses. The dissertation is completed by the third essay in which the model is extended to an open economy framework. In general, the contemporaneous restriction schemes give reasonable results, but the magnitude and timing of the effects differ across identification schemes. By contrast, the long-run restrictions approach is found to be not suitable for a relatively large system like our open economy framework. Also, in this essay, the responses for the open economy are contrasted with those for the closed economy. The results indicate that the quantitative effects are different, despite the similarity in the general patterns of the responses. In particular, all identification schemes considered in this essay showed either some degree of the 'price puzzle' or weaker price effects than in a closed economy framework, even in the presence of commodity prices and the exchange rate.