Is VaR on its deathbed?
Time and again, the Value at Risk (VaR) model has been threatened but it somehow survived the financial turmoil owing to the market's resilience and inertia to change. However, VaR now seems to be on its deathbed with the Basel Committee all set to replace it.
VaR proponents had always backed it because of its simple and objective approach, but it failed miserably during the financial crises owing to the underlying weakness in the basic model itself as well as its ignorance to black swans, which exposes banks and financial institutions to catastrophic countercyclical scenarios. Events like the black Friday or the recent US$ 2 billion loss reported by JP Morgan has exposed the vulnerabilities in VaR and its ignorance to the flat tail risk.
VaR as a model does not venture beyond the 1% threshold. Ironically the widespread use of VaR as a risk control metric prompted traders to drag the eminent risk to below the 1% VaR threshold. While this ensured that they were complying with the trading book rules, it eventually exposed the system to catastrophic and extreme losses which VaR was never intended to address. VaR leaves a lot of room for capital arbitrage as it ignores the patterns and severity of losses at both extremes of the tails.
Is weakness in the model to be blamed for the catastrophic losses or is it the urge of the financial system to lookout for every possible arbitrage opportunity? The crisis arising out of the vulnerabilities in VaR could have been subdued had the industry taken cognizance of its widely known pitfalls and used stress testing and scenario analysis in conjunction with VaR.
Nonetheless, the Basel committee is all set to mend the trading book rules and has proposed scrapping the "Value at Risk" model with the "Expected Shortfall" approach, which unlike VaR is a coherent risk measure. The expected shortfall being a convex function is also sub-additive in nature. It does not penalize diversification; thereby overcoming a severe deficiency in VaR. Though expected shortfall is theoretically superior to VaR and addresses the tail risk (at least as a conditional expectation) in a better way, the operational details can turn out to be quite challenging entailing a painful system overhaul. Outliers and black swans reside in the tail of the distribution and the expected shortfall intends to address this world of ambiguity which is a disturbing fact for many risk professionals.
The risk fraternity may back a particular model but the measuring scale named "VaR" has certainly lived its life and calls for replacement with a new scale called "Expected Shortfall". All the more, this new landscape presents ample business process change opportunities for the IT industry.
In my next blog post, I will dive deeper into some of the opportunities that exist for the IT industry. In the meantime, I look forward to hearing your views on VaR vis-à-vis Expected Shortfall.



