IMF-从预测中识别新闻冲击（英）-2023.10-78页_1mb

Summary of "Identifying News Shocks from Forecasts" (IMF Working Paper WP/23/208)

Methodology

This paper introduces a method to identify and decompose macroeconomic shocks into anticipated ("news") and unanticipated ("surprise") components in a structural vector autoregression (SVAR) model by incorporating forecast data. The approach assumes that observed time series are influenced by both current and future anticipations, allowing identification through cross-equation restrictions derived from forecast inaccuracies or data generating processes.

Key Findings

• Shock Decomposition: Adding forecasts specifies a VAR that separate each shock into news (anticipated) and surprise (unanticipated) components. For instance, fiscal shocks are largely anticipated, with news components explaining higher volatility and having larger multipliers, while monetary shocks are typically surprises.
• Volatility Contribution: News shocks account for approximately 25% of U.S. business cycle volatility at longer horizons, with fiscal, supply, and inflation variables being more influenced by news compared to output or monetary policy.
• Policy Effectiveness: Coordinated fiscal and monetary policies are substantially more effective than either policy alone in moderating output and inflation volatility, reducing output variance to near zero and significantly decreasing inflation volatility.

Policy Implications

• The identification of news and surprise components enables counterfactual policy analysis. Results show that optimal policies involve fiscal and monetary coordination to stabilize the economy. Passive sticky policies, however, find little support in the data.

Overall Contribution

The method provides a novel approach for identifying structural shocks and policy responses, offering insights into how anticipated outcomes amplify economic fluctuations and improve policy effectiveness.