Fitch recently completed a small study entitled “CDS Spreads and Default Risk: A Leading Indicator?” investigating whether default probabilities implied by credit default swap spreads serve as a good indicator for companies that eventually default.
Fitch went through a total of 27 credit events (including failure to pay, bankruptcy, restructuring) between 2008-2010. The summary of their findings is highlighted below:
- The performance of spreads as default predictors varies across sectors.
- CDS spreads did not appear to provide a leading signal of default risk for financial institutions. As of 12 months prior to the six credit events in the sector, the average one-year PD was 3.3%.
- For corporates, the predictiveness of CDS spreads was mixed. On average, spreads implied less than 10% cumulative PD for the 18 entities that incurred a credit event, or that fewer than one in 10 companies were expected to default over the following two-year period. However, at 12 months prior to the credit events, the average one-year CDS-implied PD had risen substantially to roughly 23%.
- Monoline spreads ramped up well in advance of the three credit events in this sector. As of a 24-month look back, average spreads implied a 20% chance of default over the ensuing two-year period, rising to an implied PD of roughly 60% one year prior.
Fitch also noted two important caveats in relying on CDS spreads to quantify default risk — the potential for “false positives” during periods of market stress, in which spreads overstate the ultimate realized default experience, and the potential for “false negatives,” in which spreads do not effectively signal default events down the road.
“False positives” are cases in which the default risk implied by CDS spreads overstates subsequent realized default experience. During the financial crisis, several US sectors (e.g. insurance, financial institutions, homebuilders, and real estate investment trusts) were subject to pronounced spread widening, but incurred few if any credit events. False positives can impose significant costs on market participants who rely on CDS spreads as default risk indicators. For example, risk managers hedging credit exposure might overpay for credit protection, and investors trimming their exposure could incur opportunity costs.
However, Fitch says that CDS can still be useful in finding outliers and relative values amongst companies.
Despite their limitations as quantitative estimators of default risk, CDS spreads are useful tools in differentiating relative credit quality across a population of entities. The potential utility of CDS spreads in identifying outliers is illustrated in the chart on page 3 [shown below], which compares average spreads prior to default on the 18 corporate entities that experienced credit events relative to average spreads on a Fitch-derived index of 50 high-yield names, which comprises a fixed cohort of large issuers across a mix of industrial sectors that were non-investment grade as of year-end 2010.
Two years prior to the occurrence of credit events, CDS spreads do not appear to distinguish between the credit risk of those corporates that ultimately defaulted versus their high-yield peers. However, as the point of default approaches, spreads on the defaulting names progressively widen relative to the broader high-yield universe. For example, six months prior to the credit events, spreads on the sample of defaulters were more than 2.5x the spreads on the index, indicating the prospective ability of credit markets to identify eventual defaulters over this horizon.
Note: Fitch calculated implied probabilities of default for one-year (PD) as equal to annualized CDS spread / loss severity (or alternatively 1 – recovery rate).
PD = CDS price / Loss severity
Using the example provided by Fitch, assuming a 60% loss severity (or 40% recovery rate), then an annual CDS spread of 120 bp would imply a one-year PD of 2.0%.
