BIS国际清算银行-Effects-of-eligibility-for-central-bank-purchases-on-corporate-bond-spreads_33页_353kb

Summary of BIS Working Paper No. 894: Effects of eligibility for central bank purchases on corporate bond spreads

Core Content

This working paper evaluates the causal effect of the European Central Bank's (ECB) Corporate Sector Purchase Programme (CSPP) on the yield spreads of corporate bonds in the primary market. The study uses a regression discontinuity design (RDD) to assess whether the program permanently altered the yield spreads of eligible bonds relative to noneligible ones.

The CSPP, which ran from June 2016 to December 2018, involved the purchase of 180 billion euros worth of corporate bonds. The program aimed to provide additional monetary policy accommodation and raise inflation, while also improving financing conditions for the real economy. The eligibility criteria for the CSPP were based on the highest bond rating exceeding a threshold, specifically BBB- or equivalent.

Main Findings

• The CSPP did not have a statistically significant causal effect on the yield spreads of eligible corporate bonds relative to noneligible bonds during its entire duration.
• The stock effect of the program, i.e., the effect due to the permanent reduction in eligible securities in the private sector, was not observed.
• The findings suggest that central bank purchases do not permanently alter the yield spreads of the securities they acquire, even when those securities are relatively illiquid.
• The results complement previous studies that found a temporary negative effect on eligible bond spreads, reinforcing the idea that the effects of central bank asset purchase programs are felt across a broader range of securities, not just those directly purchased.

Key Information

Program Overview

• Announced on March 10, 2016, and implemented on June 8, 2016.
• Eligible bonds: Rated investment-grade (BBB- or higher), with a maturity between 6 months and 30 years.
• Purchase scope: Both primary and secondary markets.
• Impact on market: Approximately 85% of eligible bonds issued after the program's announcement were purchased by the ECB, compared to 60% of those issued prior.

Methodology

• Regression Discontinuity Design (RDD) is used to estimate the local average treatment effect (LATE) of the CSPP on bond yield spreads.
• Ordered probit model is employed to estimate the propensity score for bond eligibility, which is used as a surrogate running variable.
• Two estimators are used:
  • Simple weighting estimator
  • Doubly robust augmented weighting estimator
• The propensity score is calculated based on issuer and bond characteristics, including:
  • Coupon rate
  • Original maturity
  • Profitability
  • Interest coverage
  • Solvency
  • Size

Data Description

• The study uses data from Bloomberg and S&P Capital IQ.
• The dataset includes:
  • Yield spreads (measured by option-adjusted spread (OAS))
  • Credit ratings from Standard & Poor’s, Moody’s, Fitch, and DBRS
  • Issuer financial data for the 2015 fiscal year
• Sample size: 1,654 bonds, of which 1,058 are used in the analysis.
• Bond types:
  • 635 callable bonds
  • 364 bullet bonds
  • 52 convertible bonds
  • 1 putable bond
• Excluded bonds: Convertible and putable bonds due to the unavailability of OAS data.

Eligibility and Propensity Score

• The eligibility threshold is based on the highest rating of a bond.
• Bonds with a maximum rating of BBB- or higher are eligible.
• The propensity score is used to estimate the probability of a bond being eligible for purchase, which is crucial for the RDD methodology.
• The model includes:
  • Coupon rate (cpn)
  • Original maturity (mat)
  • Profitability (prof)
  • Interest coverage (cov)
  • Solvency (solv)
  • Size (log of total revenue)
  • Quadratic terms for size and other variables

Results

• The estimated propensity scores for eligible bonds (BBB- and BBB) are above 0.5 for 93% and 97% of the units, respectively.
• For bonds just below the threshold (BB+ and BB), 38% and 70% of the estimated propensity scores are below 0.5, respectively.
• The absence of a permanent effect on yield spreads implies that the stock effect of the CSPP is not significant.
• The findings suggest that central bank purchases may not have long-term impacts on the prices of the securities they acquire, relative to close substitutes.

Conclusion

The study contributes to the understanding of how central bank asset purchase programs affect financial markets. It shows that the CSPP did not lead to a permanent change in yield spreads for eligible corporate bonds, highlighting the importance of considering long-term effects when evaluating such programs. The methodology and findings provide a robust framework for future research into the impact of central bank interventions on corporate bond markets.