2012年-IMF国际货币组织全球_Global_Bonding_Do_US_Bond_and_Equity_Spillovers_Dominate_Global_Financial_Markets__26页_2mb

Summary of "Global Bonding: Do U.S. Bond and Equity Spillovers Dominate Global Financial Markets?"

Core Content

This working paper by Tamim Bayoumi and Trung Bui investigates the extent of financial spillovers across major global markets—specifically the U.S., Euro area, Japan, and the UK—focusing on government bond yields and equity prices. The authors use a novel structural VAR (Vector Autoregression) methodology that leverages heteroskedasticity to identify contemporaneous causal relationships between financial markets, avoiding the limitations of traditional methods like Cholesky decomposition. The goal is to understand how shocks in one market affect others and to assess the dominance of U.S. spillovers in global financial markets.

Main Views

• U.S. Dominance in Spillovers: U.S. financial shocks have a significantly stronger impact on global markets compared to shocks from the Euro area, Japan, and the UK. The paper finds that U.S. bond and equity market shocks reverberate globally with high magnitude and statistical significance.
• Minimal Inward Spillovers to the U.S.: Conversely, shocks from other markets to the U.S. are minimal and largely statistically insignificant.
• Two-Way Spillovers in Europe: There is evidence of two-way spillovers between the UK and the Euro area, and one-way spillovers from Europe to Japan.
• Uncertainty in Causality: The paper highlights that uncertainty about the direction of causality in contemporaneous correlations is the main source of uncertainty in the estimated impulse response functions, a limitation that other techniques cannot address.

Key Information

Data Overview

• The study analyzes government bond yields and equity prices for the U.S., Euro area (represented by Germany), Japan, and the UK.
• The data spans from January 1, 2000, to December 31, 2009, capturing both the Great Moderation and the global financial crisis.
• Weekly data is used to reduce the impact of different market opening and closing times and to minimize missing data issues.
• Robustness checks include analyzing different sub-periods: 2000–07 (pre-crisis) and 2000–October 2012 (post-crisis).

Methodology

• The authors use a structural VAR approach that estimates spillovers by analyzing the contemporaneous correlation matrix.
• They propose an identification method based on heteroskedasticity, which exploits changes in volatility regimes (high vs. low) to distinguish between structural shocks.
• This method allows for the estimation of standard errors on the coefficients, providing a more reliable measure of significance than traditional techniques.
• The A-inverse matrix is used to normalize the spillover effects, with the diagonal elements set to one to reflect the size of spillovers from a unit shock in any given market.

Results

Bond Market Spillovers

• The A-inverse matrix for bond markets shows that:
  • The U.S. has large and significant outward spillovers to all other markets, with coefficients ranging from 0.51 to 0.69.
  • Inward spillovers to the U.S. are minimal, with none of the coefficients being statistically significant.
  • Two-way spillovers exist between the UK and Germany, though the causality is not clearly determined.
  • Germany's spillovers to Japan are significant (0.27), but UK's spillovers to Japan are minimal (0.03) and insignificant.

Equity Market Spillovers

• The A-inverse matrix for equity markets reveals:
  • The U.S. has large and significant outward spillovers, with coefficients around 0.44–0.68.
  • Japan's spillovers to Germany are significant (0.38), but Germany's spillovers to Japan are less so.
  • Inward spillovers to the U.S. are minimal, with coefficients ranging from 0.09 to -0.02 and mostly statistically insignificant.
  • UK's spillovers to Germany are significant (0.58), but Germany's spillovers to the UK are not as strong (0.34).

Robustness Checks

• The results are robust to changes in the sample period and the inclusion of different volatility thresholds.
• When the sample is restricted to 2000–07 (excluding the crisis), the spillover effects are smaller and less significant.
• When the sample is extended to 2000–October 2012, the Euro area crisis slightly enhances spillovers from Europe to Japan, particularly in equity markets.
• The random walk assumption appears to provide a more accurate estimation of spillovers in equity markets than in bond markets.

Conclusion

• The study concludes that U.S. financial shocks dominate global spillovers, while inward spillovers to the U.S. are minimal.
• European markets (UK and Euro area) show significant two-way linkages, but their influence on other regions is smaller than that of the U.S.
• The methodology based on heteroskedasticity is more robust and reliable than traditional methods in capturing the true size and direction of spillovers.
• The uncertainty in contemporaneous correlations remains a key challenge in estimating spillovers, even with this new method.