Evaluating Structural Vector Autoregression Models in Monetary Economies

This paper uses Monte Carlo simulations to evaluate alternative identi…cation strategies in VAR estimation of monetary models, and to assess the accuracy of measuring money instability as a cause of output ‡uctuations. I construct theoretical monetary economies using general equilibrium models with cash-in-advance constraints, which also include technology shocks, labor supply shocks, and monetary shocks. Particularly, two economies are characterized: one is fully identi…ed and satis…es the long-run restriction; another is not fully identi…ed and the portion of temporary technology shocks is mixed with demand shocks when applying the long-run restriction. Based on each theoretical model, arti…cial economies are then generated through Monte Carlo simulations, which allow me to investigate the reliability of structural VAR estimation under various identifying restrictions. Applying short-run, medium-run, and long-run restrictions on the simulated data, I check for the bias between the average VAR estimates and the true theoretical claim. The …ndings show that short-run and medium-run restrictions tend to work better under model uncertainty, particularly because the bias for measuring the e¤ects of monetary shocks using long-run restriction could increase substantially when the underlying economy includes unidenti…ed temporary shocks. This experiment supports the claim that monetary shocks contribute no more than one third of the cyclical variance of post-war U.S. output, and suggests that their contribution could in fact be substantially less. Keywords: VAR Estimation, Monetary Shocks, Business Cycle Fluctuations JEL Classi…cations: E3, E4, C1, C3 Macro-Financial Linkages Unit, Research Department, International Monetary Fund, 700 19th St. N.W., Washington, DC 20431. Email: [email protected]. Home Page: http://binli.economics.googlepages.com/. I am greatly indebted to Robert E Lucas, Jr., Casey Mulligan, Nancy Stokey, and Harald Uhlig for their support, advice and suggestions. I also thank John Cochrane, Timothy Conley, Christopher Erceg, Paul Evans, Luca Guerrieri, Christopher Gust, Lars Hansen, Erik Hurst, Anil Kashyap, Patrick Kehoe, Frederic Mishkin, Rob Vigfusson, and seminar participants of Money and Banking Workshop, Capital Theory Workshop, and Econometrics Workshop at the University of Chicago, Federal Reserve Board, Ohio State University, Southern Methodist University, SUNY Bu­ o, Temple University, UIUC, and IMF for helpful comments. All errors are my own. The views expressed in this paper are those of authors and do not necessarily represent those of the IMF. 1 Introduction Structural vector autoregressions (SVARs) are widely used by economists to study sources of business cycle ‡uctuations. Ideally, if the actual economy follows a well-speci…ed theoretical model that identi…es monetary, preference, and technology shocks, and predicts the economy’s response, we would study the economy simply based on that model. In reality, because we do not have completely successful models of this sort, most evidence for the e¤ects of di¤erent shocks has to come via SVARs. However, the theoretical models, though far from perfect, can then serve as useful lab experiments to check the reliability of structural VARs. In this paper, we use Monte Carlo simulations of reasonably calibrated dynamic general equilibrium models to evaluate SVAR models for estimating contributions of di¤erent shocks to the business cycle ‡uctuations. In particular, I evaluate alternative identi…cation strategies in structural VAR estimation of monetary models, and to assess the accuracy of measuring money instability as a cause of output ‡uctuations. 1.1 Motivation Since the seminal work of Kydland and Prescott (1982), followed by Hansen (1985) and many others, the real business cycle (RBC) economists have claimed a central role for technology shocks driving macroeconomic ‡uctuations in industrialized economies. In these studies, macroeconomic ‡uctuations are interpreted as equilibrium responses to exogenous shocks, in an environment with perfect competition and intertemporally optimizing agents, and the role of monetary policy is assumed to be, at most, secondary. In contrast, the traditional Keynesian monetary models assume a more important role of monetary shocks by emphasizing sticky nominal prices and wage settings, and other market frictions. It is not until more recently that the new generation of small-scale monetary business cycle models with a more extensive range of shocks, (see Christiano, Eichenbaum, and Evans 2005, and Smets and Wouters 2007 among others) demonstrates that monetary policy shocks, as an exogenous source, may contribute only a small fraction of the forecast variance of output at all horizons. Structural VARs have played a fundamental role in advancing research in this area. Many studies, based on SVARs, con…rm that in the postwar United States the exogenous technology shocks measured by the conventional Solow residual account for more than half the ‡uctuations. This is consistent with the classic work of Shapiro and Watson (1988), who employed structural