The maximum loss on loan portfolios with multiple products

Abstract

In our previous study of one type of credit product, when we used a single external, so-called macro explanatory variable, and concluded that by developing estimators that also reflect management capabilities, we are likely to get closer to determining the true probabilities of default. In this paper we assume that, after careful analysis, we have a correlation describing default rates by loan product as a function of macro conditions.  However, a macro state may have different effects on default rates, and hence on the resulting losses, for different types of credit products. The task therefore arises to analyse the possible macro states in order to simultaneously identify the critical macro states affecting the whole portfolio. In particular, it is important to prepare for the worst outcome, as a bankruptcy can be avoided in this way. The complexity of the model formulation of the problem favours the use of simulation techniques, but at the cost of making it more difficult to understand the causes of extreme losses due to the known drawbacks of simulation approaches. This paper provides insights into the methodology for these three factors: the identification of the set of relevant macro factors at stake, the identification of the relationship between default rates and macro states, and the identification of the macro states that cause maximum loan portfolio losses.