While the measurements related to market risks of bank portfolios were in much advanced shape by the late eighties, majority of developments in credit risk measurements were realised only during the last decade. Bank’s credit losses primarily originate from a large portfolio of assets like loans, guarantees, letters of credit, derivatives etc extended to various obligors or clients.
The core objective for any credit portfolio modelling is to simulate the probability distribution of overall credit losses over a given time horizon.
Subsequently, expected losses are identified at given percentiles from the modeled distribution to evaluate economic capital requirements. This is similar to the VaR (value at risk) concept for market risk.
Any credit model analysis primarily revolves around individual credit default profile (PD: probability of default), obligation related exposures (EAD: exposure at default) and recovery rates (RR: recovery rate). However, one cannot ignore the various levels of secondary interactions which may occur within the credit portfolio. Significant secondary interactions, aka correlations, can occur either between two or more obligors’ credit default profiles or between obligor credit profile and its obligations or between obligor credit profile and the recovery rate of its obligations. All of these interactions can significantly alter credit loss distribution which, in turn, has a pronounced impact on the economic capital of an organisation. Although banks have often utilised these correlations to achieve certain level of diversification benefits, accurate measurement of the same continues to be massive challenge from a technical perspective. Simulating real life like simulations for a credit portfolio has been a dream for every modeller and indeed, correlation stands out as a missing link for most. Some of those significant correlation aspects are described here.
1. Correlation between obligors’ credit default profiles (PD):
This is one of the most easily observable events during any widespread scenario of credit deterioration. A set of obligors could be related by being under a common legal group. Let’s say, two obligors ‘A’ and ‘B’ are engaged in unrelated businesses, but they have a common promoter. In the event of a serious default by obligor ‘A’, one can expect likely credit deterioration for company ‘B’. Multiple obligors can also be related contractually on business terms. Let’s take a supplier ‘A’ and a large consumer ‘B’ with both of them being obligors to the same bank. Most of the supplies of obligor ‘A’ are consumed by ‘B’. In the event of a credit default by ‘B’, one would expect ‘A’ to undergo a likely credit related stress, thereby adding to the deterioration in the bank’s portfolio. The above correlations between obligors can be further extended between related industry groups or sectors, countries and even to larger geographical regions including economic unions (like Eurozone). As an example, when the foreign fund flows dry up or reverse suddenly in emerging markets, almost the entire set of obligors based in the region gets affected immediately. We have seen many of these emerging market events in a “risk off” situation. Of course, each of these events has a different level of impact over credit portfolios and therefore, needs to be analysed separately. In fact, one also needs to be aware that correlations may not be static over time. They would keep varying in different stages of credit cycle for a country or region. Under a benign macro condition with low interest rates, adequate credit growth and controlled level of inflation, credit correlations may be remarkably lower relative to an adverse economic situation involving a credit crunch. Correlations generally turn higher (worsens) resulting into simultaneous defaults across related obligors under an economic downturn. Careful analysis is required for credit loss distributions for longer horizons which may involve turn of credit cycles or transitioning of credit default profiles.
2. Correlation between obligor default profiles (PD) and obligation (EAD):
Correlations between the credit rating of the obligor and its exposure obligations produce a higher EAD within portfolio models. As an example, an unfunded letter of credit may actually be converted into a funded exposure due to credit deterioration of the borrower’s financial health. Similarly when an obligor faces financial challenges, it tends to draw down most of the undrawn bank facilities in the expectation of a near term funding crunch. In case of derivative facilities, such correlations show up in form of wrong way risks. Let’s say, an importer executes a swap under which it assumes a foreign currency obligation. Mark to market of the swap (and consequently EAD) is likely to increase at the same time when the cost of inputs (imported) rise for the obligor and thereby impacting the financial health of the obligor.
3. Correlation between obligor default profiles (PD) and recovery rates (RR):
When counterparties are unable to fulfil its obligations to a bank, either its assets or the specific securities tied to its obligations are liquidated for payment purposes. The valuations of obligor’s assets are expected to be impacted by the obligor as well as underlying credit cycle. As an example, if a bank’s senior lending facility is secured by bonds from a related or connected organisation, the recovery rate is expected to be low in the event of a credit default. On the other hand, if the security is a piece of land, the valuation may be dependent on the overall credit environment. While banks may choose to ignore any explicitly correlated collateral during credit modelling, any correlation existing between obligor credit profile and unrelated collateral may be a hard topic to tackle.
In the modern times, most of the management decisions in a bank centre around risk adjusted return on an economic capital basis. Such assessments turn more insightful when viewed in the overall credit portfolio perspective, where marginal risk and return impacts are analysed on a very realistic basis. In the near future, such abilities may turn out to be a key differentiating factor between the organisation’s success and failure.
— The writer is a Dubai-based risk manager working for a global bank. Views expressed are his own.
