CREDIT SPREADS BASED ON REAL-WORLD PROBABILITIES
The premiums, mentioned earlier, against the risk-neutral default probabilities and resulting expected credit losses, are defined by using arbitrage free credit spreads. These credit spreads are applied as discounting factors as explained in Chapter 6.
However, unexpected credit losses are defined based on real-world expectations of the default probabilities and credit ratings within a certain time horizon. Thus, both default probabilities and credit ratings driving credit spreads are risk factors which are applicable in risk management analysis. In such analysis credit spreads are stressed to a certain degree. In fact, we need to consider both spreads and ratings and observe their historic performance and/or applying scenarios based on what-if or a stochastic process analysis.7The stress of credit spreads is linked to the stressed counterparties' ratings. It has been observed, however, that during credit crises, credit markets increase (adjusted) credit spreads even before the ratings have changed. Moreover, when sharp downgrading is applied, markets may tend to influence credit spreads more. Thus, any association of credit spreads with the credit ratings should be well defined and tuned within the credit spread modelling process. Of course, the adjustments of credit real-world probabilities will change the degree of expected credit losses and thus will impact the adjustments of discounting risk-free credit spreads.
As we will explain in the next paragraphs, credit spreads are also dependent on market uncertainties. This implies that the identification of credit spread fluctuation is based on future market volatilities and correlations. A good methodology for identifying credit spread fluctuations is to use the historical volatilities and/or simulating stochastic scenarios of market risk factors linked to counterparties' ratings. Analysis of fluctuations is also applied to identify the haircut on credit spreads.
Volatilities and stress conditions can also be observed from the markets such as from the premiums of single-name CDSs, bond prices and traded spreads of swaps. In addition, if no observations can be made, proxies or other mapping methods can also be used; these are defined based on some descriptive characteristics of the counterparty, e.g., region, industry, but also based on the counterparty credit rating and seniority class of the financial contract. Notably, under stress conditions markets tend to adjust spreads in a more dynamic manner than the actual real-world default probabilities and credit ratings.
Concentration and system risks also play a significant role in credit spread analysis. In fact, it is important to identify any systemic consequence to the market when a major concentrated counterparty defaults. This is also discussed in Chapter 11 (Systemic and Concentration Risks).
Finally, news and rumors in the market related to the counterparty or the sector(s) may also be considered and may impact the size of credit spreads. Such qualitative elements are very difficult to model and evaluate. Speculation is mainly based on qualitative, soft-risk parameters that influence the spreads. Deterministic stress on credit spreads may be applied to reflect such cases.
In marketplace lending, most of the interest rate that borrowers pay consists of credit spreads containing profit and counterparty risk. This makes the analysis rather complicated with low transparency because the individual spreads are all lumped together.
7.6