Structural scenario analysis with SVARs

Abstract

Macroeconomists constructing conditional forecasts often face a choice between taking a stand on the details of a fully-specified structural model or relying on correlations from VARs and remaining silent about underlying causal mechanisms. This paper develops tools for constructing economically meaningful scenarios with structural VARs, and proposes a metric to assess and compare their plausibility. We provide a unified treatment of conditional forecasting and structural scenario analysis, relating them to entropic tilting. A careful treatment of uncertainty makes our methods suitable for density forecasting and risk assessment. Two applications illustrate our methods: assessing interest-rate forward guidance and stress-testing bank profitability.