Blog article
See all stories »

Frankie does VaR

On perusing an email from Lepus some weeks ago I was struck by the title of one piece of their research "VaR, what is it good for!"

Having been involved in working with several banks on market risk implementations a decade ago, I remember the mathematical simplicity and purity of VaR models had at the time, the impact that the original RiskMetrics publication had, and so on.. I don't recall too many comments back then regarding the problems of VaR models being so deficient (with some exceptions such as capturing non-linearity etc), but the Lepus headline brought it home in terms of where some of the challenges may now be.

We can debate until tails get long on whether the challenge is model, data or governance. Whether risk models are art imitating science, or whether they give a false comfort of risk insight based on poorly communicated assumptions.

Either way if we believe risk management needs to re-establish its credentials to re-capture the heady days of a decade back, can we look for the perceived universal saviour that VaR may have been at that time, or do we expect a tougher, more fragmented and less standardised approach to risk best practice going forward..


Comments: (2)

Elizabeth Lumley
Elizabeth Lumley - Girl, Disrupted - Crayford 02 December, 2008, 11:06Be the first to give this comment the thumbs up 0 likes

As the author of said Lepus peice, I thought I would lend a comment. I do tend to prefer catchy headlines--much better than 'Use of VaR at desk level'--which had been the working title.

The humble VaR has come under intense scrutiny of late. Early last year the Bank of England issued a warning about the use of VaR. The UK central bank argued that when volatility is low in the markets, VaR tools typically offer a benign view of risk taking. This causes banks to take on more risk, which reduces market volatility. However, the Bank argued in April last year that if the market were to ever turn volatile, this dynamic could quickly unravel. The Bank estimated that a typical bank’s VaR might theoretically double, with the same assets, if volatility increased.

In an opinion piece in the Wall Street Journal this April, Bruce Wasserstein, chairman and CEO of Lazard, offered a vicious attack on VaR models. “The protective pyrotechnic mechanisms such as Value at Risk, which were supposed to predict the risk in a portfolio of assets, are duds.”

Speaking also to the Wall Street Journal this May, Malcolm Knight, the chief executive of the Bank of International Settlements does not lay all the blame on the VaR models themselves. He lays some of the blame for the current crisis in risk management on the quality of the information and its transmission along the securitisation chain.

Despite all this, of course, VaR is still widely used in the financial markets. As the financial markets have seen over the past few months, market turmoil can highlight the limitations of sophisticated measures of market risk.

VaR will continue to be used as highlighted by the banks interviewed for the research. VaR has not been discredited for us in stable market conditions, despite several market theorists who claim that the use of VaR can exacerbate market volatility.

Almost all the banks interviewed relied on historical simulation to calculate VaR. It could be argued that all risk management is based on the calculations of past numbers and market conditions.

However, great market turmoil tends to come from unexpected events that could not have been predicted by historical data modelling. This is where VaR falls short.



A Finextra member
A Finextra member 16 December, 2008, 03:53Be the first to give this comment the thumbs up 0 likes

The Financial Stability Report (FSR) of the European Central Bank (ECB) [December 2008], published last evening, I think may answer your concern; a brief quotation from the special report section goes to the nub;-

"Risk management should not rely on a single risk methodology, taking into account the limitations of models and risk measurement techniques such as the value at risk (VaR). In an environment like the financial markets, in which extreme price movements can be frequent, seemingly robust risk management tools can prove to be inadequate.

With this in mind, firms should expand the range of risk metrics, moving beyond VaR as the dominant risk measure to include a range of stress tests, scenario analyses and other measures that could be useful in revealing portfolio risk characteristics, again taking into account the possible model drawbacks of these alternative risk measures."

I do hope this may be of use, J A Morrison

Now hiring