BY CONTINUING TO USE THIS SITE, YOU ARE AGREEING TO OUR USE OF COOKIES. REVIEW OUR PRIVACY & COOKIE NOTICE
X
HOME > OUR THINKING > Credit Analysis > RESEARCH

U.S. LossStats® Model

After experiencing the worst global recession since the 1930s1 that led to one of the longest quantitative easing cycles in U.S. history, the U.S. economy is giving more and more signals of recovery, and the Federal Reserve has finally started a progressive interest rate hike. This could have profound implications on the default rates of bank loans and corporate bonds which have been at historical lows for several years, partly due to the accommodative monetary policy of the U.S. Federal Reserve.

At the end of 2016's third-quarter, the total amount of outstanding U.S. Corporate bond debt was in excess of $8.5 trillion, and growing.2 Arguably, a surge in interest rates could exacerbate the default risk of bonds and loans as issuers find less opportunity to refinance at low rates, leading to higher default rates of non-financial corporate companies in both the investment grade, and to a greater extent, the non-investment grade space. This confirms the importance of assessing credit risk, looking at “both sides of the coin”: default risk and loss given default.

To assess default risk, analysts can rely upon the credit ratings of an established rating agency or the outputs of a statistical model, either internally developed or distributed by a third party provider: S&P Global Market Intelligence redistributes the credit ratings of S&P Global Ratings and offers multiple quantitative tools, based on company fundamentals or market signals that estimate the credit risk of a rated or unrated obligor.3

To estimate loss given default (LGD) or recovery given default (RGD),4 financial practitioners traditionally revert to non-robust methods, including:

  • A fixed value for loss given default (around 45% for senior unsecured claims and 75% for subordinated claims5); on one hand, this penalizes lending areas such as trade finance and project finance where default rates are high, but many factors mitigating risk of losses are in place6; on the other hand, it does not reflect the actual distribution of recoveries that features a pronounced bi-modality, with recoveries concentrating around 0% and 100%.
  • A “look-up tables” of averages;7 this approach is well-suited for small and granular retail exposures, but often not robust for exposures to large-revenue corporate companies, due to the paucity and inconsistency of data collected around low-default asset classes, the peculiarities of legal systems in different countries, the lengthy process (up to several years) before final resolution and the multitude of recovery calculation methods.

S&P Global Market Intelligence’s U.S. LossStats Model represents the latest addition to the Credit Analytics model suite. This statistical model estimates the distribution of loss given default of bonds and loans issued by corporations, taking into account industry- and instrument-specific characteristics, and leveraging an extensive recovery database,8 thus making the estimates robust. This tool is a useful component for risk analysts looking for new business opportunities beyond low default asset classes, or for investment managers seeking to diversify their exposures portfolio; it also helps regulated financial entities calculate more realistic credit risk capital requirements, potentially saving a significant amount of capital, or simply benchmarking their calculations and performing scenario analysis.

This paper summarizes LossStats Model coverage and features, the analytic framework employed, and the overall model performance. We conclude with a case study that highlights the advantage of assessing the recovery of a corporate exposure with several metrics.

READ THE FULL WHITEPAPER


1 "World Economic Outlook - April 2009: Crisis and Recovery". Box 1.1 (page 11-14). IMF. 24 April 2009 (as of November 28, 2016).

2 US Bond Market Issuance and Outstanding (xls) - annual, quarterly, or monthly issuance to February 2017 (issuance) and from 1980 to 2016 Q3 (outstanding), updated 03/07/17, from www.sifma.org.

3 For example, the reader can refer to the white papers on PD Model Fundamentals, PD Model Market Signals, and CreditModelTM 2.6, available here.

4 Recovery Given Default (RGD) is simply defined as 1 – LGD.

5 Section III.B, § 23 – 30, of Basel Committee on Banking Supervision (2001), “The Internal Ratings Based Approach”. Available at http://www.bis.org/publ/bcbsca05.pdf.

6 Usually, a combination of collateral, letters of credit, third-party guarantees and insurance.

7 This approach is commonly employed across banks with an Advanced IRB approach.

8 The LossStats database is available via S&P Global Market Intelligence’s CreditProTM