When firms implement capital models in line with supervisory standards, a range of interpretative and implementation choices inevitably arise. These choices reflect differences in modeling approaches, data availability, system architecture and risk management practices, and can lead to variation in model design and outcomes across the industry. The degree of flexibility may be relatively limited or more substantial, depending on the specific capital framework – such as credit risk, market risk or counterparty credit risk – and on whether the model is based on a standardized or advanced approach.

Internal models for counterparty credit risk (CCR) are an advanced regulatory approach for measuring CCR exposures. The resulting exposure measures are used for both risk management purposes and for the calculation of exposure at default (EAD) in default risk capital. These models are typically based on Monte Carlo simulation techniques, requiring firms to make a series of choices relating to inputs, models and techniques when developing and maintaining their internal models.

All these modeling choices have varying levels of complexity with respect to approach, data requirements and associated costs and benefits. Firms typically assess the soundness of these choices through internal model validation practices – for example, by referencing previous model versions, comparing outputs to front-office pricing models or analyzing sensitivities.

Benchmarking to peers complements these internal assessments by extending model validation to external comparison across firms’ internal models using hypothetical portfolios. In this paper, benchmarking refers to structured, like-for-like comparison of results and modeling choices across institutions, providing an external reference point that cannot be obtained through firm-specific analysis alone.

A robust peer benchmarking framework provides benefits to both firms and supervisors by introducing transparency and comparability of internal model outcomes across institutions. By answering cross-sectional questions, benchmarking supports supervisory dialogue, promotes convergence towards supervisory expectations and strengthens confidence in the credibility and consistency of internal models.

This paper puts forward a benchmarking approach that can be applied to the internal model for CCR. This is an industry approach developed with engagement from the UK Prudential Regulation Authority (PRA) as part of a wider regulatory initiative. The paper sets out what benchmarking is, the questions it answers and how this complements model validation practices. It then puts forward the criteria for a robust capital models benchmarking framework, using ISDA’s experience in the development of its award-winning standardized approach benchmarking solution, which includes ISDA Analytics. Finally, it illustrates how this approach has been leveraged to benchmark internal models for CCR.

Click on the PDF to read the paper in full.