2017年-IMF国际货币组织全球_Sovereign_Bond_Prices_Haircuts_and_Maturity_37页_1mb

Summary of IMF Working Paper: Sovereign Bond Prices, Haircuts and Maturity

Core Content

This IMF Working Paper challenges the conventional assumption that creditor losses ("haircuts") in sovereign debt restructurings are symmetric across debt instruments, regardless of their maturity. The authors analyze a comprehensive dataset of 28 sovereign debt restructuring episodes (18 external, 10 domestic) between 1999 and 2015, involving private creditors, and find that haircuts on shorter-term debt instruments are systematically larger than those on longer-term debt. This asymmetric haircut pattern is robust to different measurement methods and is explained by the behavior of bond prices and the term structure of default risk.

Main Findings

1. Asymmetric Haircuts by Maturity

• Stylized Fact 1: Haircuts on short-term bonds are larger than those on long-term bonds.
• This pattern holds across various restructuring strategies (preemptive vs. post-default), types of debt (external vs. domestic), and exchange methods (single menu vs. different menus).
• The result is confirmed using two distinct measures of haircuts:
  • SZ Recovery Rate: Based on net present values of old and new debt instruments.
  • Exchange Recovery Rate: Based on observed market prices of old and new instruments.

2. Bond Price Convergence

• Stylized Fact 2: In the run-up to a debt restructuring, the price difference between short-term and long-term bonds decreases.
• Short-term bond prices tend to be higher than long-term bond prices before the restructuring, and they converge to the long-term bond prices as the restructuring approaches.
• The convergence is observed across both external and domestic debt, and is not specific to any particular restructuring strategy or exchange method.

Key Data and Methodology

• Dataset: The authors compile a new dataset covering 449 instruments for SZ haircuts and 111 instruments for exchange haircuts.
• Data Sources: Offering memoranda, government press releases, financial databases (Bloomberg, Datastream, Dealoric, JP Morgan Markit), and reports from the IMF and other researchers.
• Time Window: The exchange recovery rate measure is calculated from 6–9 months before the restructuring announcement to the exchange date.
• Regression Analysis:
  • Cross-sectional regressions confirm that maturity positively affects recovery rates, with a 10-year bond having a recovery rate 3–12 percentage points higher than a 1-year bond.
  • Panel regressions show that short-term bond prices converge to long-term bond prices over time, with a significant 9 percentage point reduction in price differential over 6 months.

Theoretical Model

• The authors build on a standard arbitrage-free risk-neutral pricing model to explain the observed stylized facts.
• First Result: Short-term bond prices exceed long-term bond prices before a restructuring, assuming strictly positive default risk after the maturity of short-term bonds.
• Second Result: As default risk becomes increasingly short-term in the run-up to restructuring, the price differential between short- and long-term bonds narrows, leading to smaller haircut differentials.
• Interpretation: The convergence of bond prices reflects the time variation in the term structure of default risk. Short-term bonds, which are perceived as having higher default risk, experience larger haircuts due to their higher pre-restructuring prices.

Default Probability Term Structure

• The paper derives an estimated term structure of default probability.
• This structure shows that short-term bonds face higher default risk than long-term bonds, which aligns with the observed price behavior and haircut patterns.
• The model supports the idea that the asymmetry in haircuts is a natural implication of the term structure of default risk and its evolution over time.

Contribution to Literature

• Empirical Literature: The paper contributes to the study of creditor losses in sovereign debt restructurings by providing instrument-specific haircut data for a large sample of episodes, unlike previous studies that only reported average haircuts.
• Theoretical Literature: It builds on the classic framework of Eaton and Gersovitz (1981) and extends it to examine the maturity structure of sovereign debt.
• The findings are consistent with other empirical evidence on bond price dynamics and default risk perception.

Conclusion

The paper presents two novel stylized facts:

1. Shorter-term debt instruments experience larger haircuts than longer-term ones.
2. Short-term bond prices converge to long-term bond prices as restructuring approaches, reducing the ex-post haircut differential.

These findings are robust across different measures and are explained by the term structure of default risk and its time variation. The results have important implications for understanding the pricing of sovereign debt and the distribution of losses in restructuring episodes.