Abstract
Despite intense criticism, agency credit ratings are still widely used in regulation and risk management. One possible alternative is to replace them with quantitative default risk measures. For US data, I find that systemically relevant losses from corporate defaults are mostly smaller if risk-taking in portfolios is limited with the help of default probability estimates from the Credit Research Initiative rather than through Moody’s ratings. The results continue to hold when investors follow a regulatory arbitrage strategy that tilts portfolios toward issuers with high systematic risk. I further show that combining information from both measures can lead to a systemic risk profile that is more favorable than can be achieved by using only one.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Cf. the information at http://www.naic.org/cipr_topics/topic_rating_agencies.htm.
Portfolio adjustments could matter if they lead to systemically relevant liquidity spirals. This issue will be discussed in Section 4.
National University of Singapore, Risk Management Institute, CRI database. Available at http://rmicri.org.
I refrain from conducting studies for other countries because the number of rated non-US firms is still relatively low and was even lower in the 1990s and 2000s.
Related to the distinction between real-world and risk-neutral default probabilities, note that the CRI distance-to-default measure is risk-neutral. Since the model is then calibrated on actual default data, however, the final CRI default probability should be a real-world one again.
A possible reason why the lagged stock return is insignificant in the CRI model but not in CHS (2008) or other models is that the CRI model includes not only the level of distance to default and size but also their one-year trend.
On its web pages, Moody’s provides industry-specific descriptions of rating methodologies. They do not mention stock market prices and volatilities of issuers.
If the procedure leads to ambiguous matches—for example, because a name is associated with two different identifiers in the Moody’s database—I use the match that produces a longer series of matched ratings.
Throughout the paper, I apply the correction T/(T-K) to the standard Newey and West estimator in order to mitigate small sample biases, as is done, for example, in STATA.
Contrary to the results here and in other papers, Hilscher and Wilson (2017) find that agency ratings also underperform over longer horizons. Their default definition, however, includes delistings and other events that are not part of the agency’s default definition, thus putting agency ratings at a disadvantage relative to their model, which is trained on the default data generated with the authors’ default definition.
Acharya et al. (2017) study the systemic risk contributions of financial institutions and suggest estimating MES as the average equity return of an institution that is conditional on the stock market return being below its 5% quantile.
The continuing importance of the investment grade boundary is documented in Baghai et al. (2018). They show that more than 80% of fixed-income fund prospectuses mention the investment grade threshold.
A minimum of 20 months is required for inclusion. Accordingly, the first formation month is not December 1990 but August 1992, which explains why the firm months from Panels B and C do not add up to half of the firm months reported for Panel A.
I also used other methods of estimating systematic risk: (i) logarithmic changes of default probabilities for the dependent variable, (ii) correlation coefficients between transformed default probability changes and market returns, (iii) and the past 60 months to determine the mean of default probability changes conditional on the market return being below its 5% quantile. The resulting spreads between strategies are mostly lower than the ones from Table 3, and the results do not necessitate a change in conclusions.
I select firms assigned to “Banking” or “FIRE”, where FIRE stands for Finance, Insurance, Real Estate.
The findings on discriminatory power confirm previous results obtained by Löffler (2007), who combined Moody’s rating information with EDFs (Expected Default Frequencies) from Moody’s KMV.
A default event also counts as a dropout, but in the data, there is no case in which a firm defaulted without its default risk surpassing the critical level that is employed in the analysis.
References
Acharya VV, Pedersen LH, Philippon T, Richardson M (2017) Measuring systemic risk. Rev Financ Stud 30:2–47
Allen F, Carletti E (2013) What is systemic risk? J Money Credit Bank 45:121–127
Altman EI, Rijken HA (2004) How rating agencies achieve rating stability. J Bank Financ 28:2679–2714
Baghai R, Becker B, Pitschner S (2018) The private use of credit ratings: evidence from investment mandates. Working Paper
Becker B, Ivashina V (2015) Reaching for yield in the bond market. J Financ 70:1863–1902
Bongaerts D, Cremers KM, Goetzmann WN (2012) Tiebreaker: certification and multiple credit ratings. J Financ 67:113–152
Campbell JY, Hilscher J, Szilagyi J (2008) In search of distress risk. J Financ 63:2899–2939
Cantor R (2001) Moody’s investors service response to the consultative paper issued by the Basel committee on Bank supervision “a new capital adequacy framework”. J Bank Financ 25:171–185
Dilly M, Mählmann T (2015) Is there a “boom bias” in agency ratings? Rev Financ 20:979–1011
Driessen J (2005) Is default event risk priced in corporate bonds? Rev Financ Stud 18:165–195
Duan JC, Sun J, Wang T (2012) Multiperiod corporate default prediction—a forward intensity approach. J Econ 170:191–209
Dwyer GP, Tkac P (2009) The financial crisis of 2008 in fixed-income markets. J Int Money Financ 28(8):1293–1316
Gersbach H, Rochet JC (2017) Capital regulation and credit fluctuations. J Monet Econ 90:113–124
Güntay L, Kupiec P (2014) Taking the risk out of systemic risk measurement. Working Paper
Gupta V, Parwani K (2009) Moody’s senior ratings algorithm & estimated senior ratings. Special Comment, Moody’s Investors Service
Hanley JA, McNeil BJ (1982) The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 143:29–36
Hilscher J, Wilson M (2017) Credit ratings and credit risk: is one measure enough? Manag Sci 63:3414–3437
Iannotta G, Pennacchi G, Santos J (2019) Ratings-based regulation and systematic risk incentives. Rev Financ Stud 32:1374–1415
Kealhofer S (2003) Quantifying credit risk I: default prediction. Financ Anal J 59:30–44
Löffler G (2004) Ratings versus market-based measures of default risk in portfolio governance. J Bank Financ 28:2715–2746
Löffler G (2007) The complementary nature of ratings and market-based measures of default risk. J Fixed Income 17:38–47
Malherbe F (2015) Optimal capital requirements over the business and financial cycles. CEPR Discussion Paper 10387
Merrill CB, Nadauld TD, Strahan PE (2019) Final demand for structured finance securities. Manag Sci 65:390–412
Merton RC (1974) On the pricing of corporate debt: the risk structure of interest rates. J Financ 29:449–470
Newey WK, West KD (1987) A simple, positive semi-definite, heteroskedasticity and autocorrelationconsistent covariance matrix. Econometrica 55:703–708
Repullo R (2013) Cyclical adjustment of capital requirements: a simple framework. J Financ Intermed 22:608–626
RMI Staff (2016) NUS-RMI credit research initiative technical report version: 2016 update 1. Global Credit Review 6:49–132
Acknowledgements
I am very grateful for the hospitality that I enjoyed during a research visit at the Risk Management Institute, National University of Singapore. The paper has greatly benefited from comments by an anonymous referee, the editor and from participants at the 2018 conference of the Financial Engineering and Banking Society and the 2018 International Risk Management Conference.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Löffler, G. The Systemic Risk Implications of Using Credit Ratings Versus Quantitative Measures to Limit Bond Portfolio Risk. J Financ Serv Res 58, 39–57 (2020). https://doi.org/10.1007/s10693-019-00321-9
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10693-019-00321-9