Hypothetical yield curve scenarios for credit stress testing

Bhavin Desai and Kausick Saha

Abstract

Central counterparties (CCPs) use several extreme but plausible scenarios for assessing their credit exposures on account of member defaults under stressed market conditions. These stress testing results help in sizing the resources (also known as default waterfall resources) that could be required for mutualizing losses if low-probability, high-intensity events were to materialize. Stress testing also aids CCPs in assessing the adequacy of their prefunded default waterfall resources on a regular basis. In this paper, we discuss a set of hypothetical yield curve shift scenarios generated by applying extreme value distributions and a shaping procedure. These statistically derived hypothetical stress scenarios could be susceptible to model risk, leading to under- or overestimation of a CCP’s resource requirements. This paper highlights the merits of the model and its associated risks as well as the methods used for minimizing these model risks. The methodology has been developed by the Clearing Corporation of India Limited for stress testing its exposures in the government securities market. While modeling extreme movements of interest rate curves is not a new theme, we have attempted to present a procedure for deriving stress scenarios that can be applied to portfolios containing long and short positions in debt market securities of varying residual maturities.