Value-at-Risk benchmarking during a crisis


Value-at-Risk benchmarking during a crisis

Implications of the COVID-19 turmoil

Executive Summary

The COVID-19 crisis has highlighted the limits of our standard frameworks for estimating financial risks in an unprecedented way. Value at Risk (“VaR”) models rely on historical data to estimate future risks, which fails—by nature—to capture the unpredictable character of large scale events impacting the overall economy, but they are expected to adapt quickly as such events unfold. Finding the right balance between reactivity and stability of VaR models is part of the objectives of modern model risk management. Model risk is usually mitigated by a validation framework involving back-testing; a technique aimed at assessing statistically the ability of a VaR model to accurately forecast risks, completed by analysis of VaR forecast overshootings. When too many overshootings occur, is back-testing still useful? How would another VaR model have performed in the same period? Relying on a range of models with diverging assumptions and a benchmarking approach can provide key insights.

PDF - 635kb

Performance magazine issue 33, September 2020

Performance is a triannual digest, dedicated to investment management professionals, which brings you the latest articles, news and market developments from Deloitte’s professionals and clients.

PDF - 8.78mb
Did you find this useful?