2012年-IMF国际货币组织全球_Assessing_Dsge_Models_with_Capital_Accumulation_and_Indeterminacy_35页_660kb

Summary of "Assessing DSGE Models with Capital Accumulation and Indeterminacy"

Core Content

This paper analyzes New Keynesian Dynamic Stochastic General Equilibrium (DSGE) models that incorporate capital accumulation and the potential for indeterminacy. It evaluates the dynamic properties of these models and their empirical performance when estimated on U.S. data from 1960:I to 2008:I.

Main Points

1. Model Structure

• The model includes sticky prices, capital accumulation, and monetary policy rules.
• Households maximize utility over consumption, leisure, and real money balances.
• Firms operate in a monopolistically competitive market and face Calvo-style staggered pricing.
• The New Keynesian Phillips Curve is derived from the firm's profit maximization under sticky prices.
• The monetary policy rule is specified with interest rate smoothing and responses to inflation and output.

2. Indeterminacy and Monetary Policy

• Indeterminacy arises when monetary policy is passive (i.e., $\psi_{\pi} < 1$), while determinacy is associated with active monetary policy (i.e., $\psi_{\pi} > 1$).
• The paper shows that indeterminate models perform better in empirical estimation than determinate ones.
• The stock nature of capital introduces persistency in economic variables, which is crucial for explaining the observed dynamics in U.S. data.

3. Empirical Estimation

• The model is estimated using Bayesian methods with state-space decomposition and Kalman filter.
• Prior distributions are used to maximize the overall likelihood of the model.
• The pre-Volcker period (1960-1979) is assumed to have passive monetary policy, while the post-1982 period is assumed to have active monetary policy.
• The Metropolis-Hastings algorithm is used for inference.

4. Model Simulations

• The model with passive monetary policy generates higher volatility in interest rates and inflation compared to the active version.
• Variance decomposition shows that in the passive version, preferences shocks explain most of the volatility, while in the active version, marginal cost shocks are more significant.
• Autocorrelation coefficients for consumption and capital are substantial, exceeding 0.85 in both versions, indicating strong persistence.
• The interest rate has a stronger effect on output and capital in the active version, whereas in the passive version, it is positively correlated with output and capital, reflecting inadequate policy response.

5. Comparison with Canonical Models

• Canonical New Keynesian models typically fail to generate sufficient persistence in inflation and output.
• The inclusion of capital accumulation in the model significantly improves the persistence of economic variables.
• The simulated results align with the observed behavior of U.S. data during the pre-Volcker period, which was characterized by passive monetary policy and higher volatility.

Key Information

• Key variables: Consumption, interest rate, inflation, output, and capital.
• DSGE model features: Sticky prices, capital accumulation, and monetary policy rules with interest rate smoothing.
• Empirical data: Quarterly U.S. data from 1960:I to 2008:I, including GDP, CPI, Federal Funds Rate, consumption, and capital.
• Methodology: Bayesian estimation, Kalman filter, and Metropolis-Hastings algorithm.
• Findings:
  • Indeterminate models outperform determinate ones in empirical fit.
  • Capital accumulation increases the persistence of economic variables.
  • Passive monetary policy is associated with higher volatility and indeterminacy.
  • The response of monetary policy to inflation and output significantly affects the dynamics of the model.

Implications

• The paper challenges previous findings by showing that passive monetary policy can be consistent with indeterminacy and that capital accumulation plays a critical role in model dynamics.
• It provides new insights into the stability and persistence of New Keynesian DSGE models and their empirical relevance.
• The results have important implications for monetary policy analysis and DSGE model specification.