Default correlation: rating, industry ripple effect, and business cycle

ABSTRACT

For a well-diversified bond portfolio, default risk over the investment horizon is known as the major risk and the risk is largely from correlated defaults. While plenty of theoretical work about default correlation has been developed, empirical studies on default correlation have not made much progress in the past two decades. In this paper, we fill this void in the literature by thoroughly investigating how default correlation changes across different bond ratings, over different time horizons, and across different industries over the sample period of 1970 to 2014. In particular, we examine how rating-based default correlations change before, during, and after recessions. More importantly, we reveal the ‘industry ripple effect’ that default correlations are low within upstream industries but become higher within downstream industries along the structure of the supply chain. Also, default correlations are relatively high between upstream industries and downstream industries.

KEYWORDS:

• Default correlation
• bond rating
• industries
• and recession

JEL CLASSIFICATION:

• G11
• G12
• G32

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Since defaults on Aaa, Aa, and A-rated bonds are rare, we combine the data of these three rating bonds together to get a more reliable estimation.

² Default correlation between investment grade bonds during recessions is only 0.01% higher than that during growth years, as reported by De Servigny and Renault (2004).

³ Aaa-A-rated bonds are not included in this analysis because their default correlations are close to 0 in most of the cases.

⁴ We thank the anonymous referee for helping us engage in the detailed discussion of precursor behaviour of default correlation theoretically, helping understand the empirical finding in depth.

⁵ We follow De Servigny and Renault (2004) to choose 5% to indicate relatively high default correlation for easy comparison.

⁶ We choose this period because the classification of industries is the same as that we study. The industry classification in Input-Output Accounts is different before 1997. Since finance is not listed as a separate industry, we do not have data to position finance industry in the study.