美联储-商业周期中的通货膨胀与实际活动（英）-2023.6-53页_1mb

Summary of "Inflation and Real Activity over the Business Cycle"

Key Findings

This paper explores the relationship between inflation and real economic activity (output and unemployment) during the business cycle by introducing the Trend-Cycle Vector Autoregression (TC-VAR) model. Unlike traditional VAR models, which fail to properly distinguish between low-frequency and business-cycle movements, the TC-VAR clarifies how low-frequency variations (e.g., trends) contaminate the analysis of business-cycle dynamics.

Methodology

• TC-VAR model: Decomposes variables into trend and cyclical components using band-pass filters.
• Empirical approach: Analyzes U.S. data from 1955–2019 for inflation, real GDP, unemployment, and interest rates. Fails to reject the New Keynesian framework, which links business-cycle fluctuations in inflation to real activity.

Results

• Business-cycle fluctuations in inflation are correlated with those in real activity (output and unemployment). The identified "unemployment shock" explains a large portion of inflation's cyclical volatility (30–49%) at business-cycle frequencies.
• The TC-VAR reveals that low-frequency trends in inflation (e.g., secular trends) are critical for separating cycle and trend components. Traditional VARs, which rely on fixed parameters, cannot properly identify business-cycle relationships due to contamination by low-frequency movements.

Contribution

The study validates the New Keynesian model empirically while criticizing earlier findings (like those in Angeletos et al., 2020) that suggested low sensitivity of inflation to real activity due to methodological flaws. The TC-VAR framework is recommended for improved inference on the inflation-output relationship.

Robustness Checks

• Alternative TC-VAR specifications and priors on trends yield consistent results.
• Excluding low-frequency trends (via constant trends) recovers VAR-like findings (e.g., low inflation sensitivity), confirming prior econometric limitations of such models.

Policy Implications

This work supports central banks' use of forward-looking models (like DSGE) and emphasizes that accounting for trends and structural breaks (e.g., post-Volcker disinflation) is essential for effective inflation forecasting and policy design.