Take your model bowling: forecasting with general equilibrium models
介绍DSGE-VAR方法，将DSGE模型作为先验约束VAR参数，提升预测表现，尤其适用于货币政策变量（产出增长、通胀、联邦基金利率）的预测，结果优于无约束VAR，与明尼苏达先验VAR相当。
During the past two decades, dynamic stochastic general equilibrium (DSGE) models have taken center stage in academic macroeconomics. Nonetheless, these models are still rarely used in policy-making and forecasting. ; This article describes the workings of the DSGE-VAR, a procedure that combines DSGE models and vector autoregressions (VARs). The procedure uses DSGE models as priors to restrict the VAR?s parameters. Since the VAR?s parameters are imprecisely estimated unless a very long time series of data is available, using DSGE priors can improve the VAR?s forecasting performance. Moreover, the Lucas critique implies that DSGE priors can be particularly useful when forecasting the impact of policy changes. ; The authors assess DSGE-VAR?s forecasting performance in terms of three variables that most interest monetary policymakers: real output growth, inflation, and the federal funds rate. Their results show that the DSGE-VAR forecast is superior to that of unrestricted VARs and comparable to that of VARs with Minnesota priors. ; The article also discusses how DSGE-VAR can be used to identify the fundamental shocks that hit the economy and to forecast the impact of changes in the policy rule followed by the monetary authorities. ; Perhaps in the not-too-distant future, practitioners and policymakers will be able to use a full-fledged DSGE model for both forecasting and policy assessment. In the meantime, the authors argue, DSGE-VAR may provide a viable alternative to the models currently used.