Does Money Matter? The Robustness of Evidence from Vector Autoregressions
批评传统宏观计量模型依赖主观假设，介绍Sims提出的向量自回归（VAR）方法，认为其限制少、估计简单，能避免传统方法的虚假设定问题。
TRADITIONALLY, EMPIRICAL MACROECONOMIC RESEARCH begins with the use of theory to construct a highly restricted structural econometric model. The determination of the nature of those restrictions is fraught with difficulty and thus many specification decisions turn out to be largely ad hoc and of dubious validity. Once these specification issues have been (probably inadequately) addressed and a structural model constructed, it is then estimated using complicated econometric techniques requiring sophisticated computer programs. Recently, an alternative approach to modeling macroeconomic time series has come into wide use. This alternative, introduced by Sims ( 1980a, 1980b), suggests the use of vector autoregression (VAR) models to analyze time series relationships among macroeconomic variables. The fact that the use of this approach has proliferated widely in a relatively brief period of time is due not only to the very persuasive criticisms of traditional methods made by Sims (see especially Sims 1980a) but perhaps even more to the perception that specification and estimation problems are greatly simplified. The user of the VAR approach imposes few restrictions and can generally use ordinary-least-squares estimation procedures. The VAR approach is thus believed to be largely free of the spurious specification assumptions and consequent specification errors necessitated by traditional macroeconometric procedures.