Forecasting and Conditional Projection Using Realistic Prior Distributions

Abstract

This paper develops a forecasting procedure based on a Bayesian method for estimating vector autoregressions.The procedure is applied to ten macroeconomic variables and is shown to improve out-of-sample forecasts relative to univariate equations.Although cross-variables responses are damped by the prior, considerable interaction among the variables is shown to be captured by the estimates.We provide unconditional forecasts as of 1982:12 and 1983:3.We also describe how a model such as this can be used to make conditional projections and to analyze policy alternatives.As an example, we analyze a Congressional Budget Office forecast made in 1982:12.While no automatic causal interpretations arise from models like ours, they provide a detailed characterization of the dynamic statistical interdependence of a set of economic variables, which may help in evaluating causal hypotheses, without containing any such hypotheses themselves.