When Ignorance is Not Bliss … Just Risky

By Emily Tooker, Student of Master of Business (Applied Finance)

In the popular 1983 comedic film, Trading Places, commodities broker, Louis Winthorpe III, and rags-to-riches profiteer, Billy Ray Valentine, bet their life savings on the frozen orange juice market to accomplish a lucrative trade in a single master stroke.

At the end of the film, we are treated to Winthorpe and Valentine basking on a tropical island, and the token yacht decorating an expanse of blue ocean in the background.

Real or imagined accounts of overnight monetary success, such as that depicted in Trading Places, have given us the false impression that risk will give its due reward. Likewise, basic finance theory tells us that there is a risk-return trade off – to obtain a higher return, you must be willing to take on more risk – and this theory has driven a risk-seeking culture in finance.

Nowhere has the disregard for formal risk management been more apparent than in the unsophisticated application and interpretation of a simple risk measurement model, Value at Risk (VaR). VaR was never intended to accurately predict all possible risk and yet, despite this limitation, risk managers and traders have relied almost exclusively on VaR to estimate total portfolio risk.

Think of it this way. In the hands of a master sculptor, a hammer and chisel can recreate Michelangelo’s David; in the hands of an apprentice, you’ll more likely get his outwardly-challenged little brother. The master sculptor wields the same hammer but has superior aim, understands just how hard to strike and the right amount of leverage required. Like a hammer, VaR’s accuracy depends on the ability of its user.

Next year, JP Morgan will celebrate the 20 year anniversary of RiskMetrics, the free online service that launched VaR into financial risk management. Regulatory commentary throughout the 1990s established VaR as a credible model for risk measurement. For example, both the 1993 Group of 30 report on derivatives practice, and the 1995 Basel Amendment to market risk capital requirements recommended the use of VaR for risk assessment.

As it happens, the popularisation of VaR coincided with exponential growth in the opaque and largely unregulated over-the-counter derivatives market. Unsurprisingly, the increase in the derivatives market led to increased risk exposures. Britain’s oldest merchant bank, Barings, suffered staggering losses through speculative trading in futures and options (Barings famously collapsed in 1995) and Metallgesellschaft lost $1.3 billion in 1993 by speculating in oil futures (Metallgesellschaft is now a part of GEA Group Aktiengesellschaft).

Clearly, risk measures were needed.

At the time the Global Financial Crisis, Deutsche Bank, Bank of America, and Morgan Stanley were among the major financial institutions employing VaR to quantify their total risk exposure. The Basel Committee on Banking Supervision made VaR the mandatory measure to determine credit, market and operational risk in the 1996 Amendment and re-affirmed this in the 2004 Basel Accords agreement. The result was that VaR was interpreted as the only relevant measure of risk exposure rather than as one input into a varied and diverse risk management program. It was the only hammer in the toolbox.

This was compounded by the fact that VaR is a simple model, just as a hammer is a simple tool. It predicts that, say, 95% of the time you may lose this much. VaR’s simplistic assumptions, such as a normal distribution of returns and historical data to predict the future, make VaR easy to use. Yet these assumptions mean that VaR can give a bad approximation when returns are not normal (GFC anyone?). Unknown future events cannot be captured in historical data if they have not happened before, and the worst-case scenario is the biggest unknown of all. Even in the best hands, VaR is limited by its simplistic assumptions.

Since VaR is a purely numerical measure, it does not capture the subjective risk in certain securities. This means that traders can hide the true level of risk of their trades while staying within the parameters of VaR. Mortgage-backed securities, which carry the inherent dangers of mortgage default and devaluation of the underlying asset, are an infamous example.

Relying on only the VaR measure, which could not factor in these inherent risks, financial institutions from 2004 onwards dangerously increased their leverage by way of mortgage-backed securities. It has been well-documented how mortgage-holders defaulted and the housing price bubble burst in America resulting in the GFC.

The GFC was an extraordinary event – unpredictable and catastrophic – exactly the kind of event that VaR cannot predict. Swiss giant UBS recorded 29 VaR exceptions in 2007 and 50 VaR exceptions for 2008. At the 99 percent confidence interval, there should only have been 2.5 exceptions. Countless other financial institutions, including JP Morgan and Bank of America, recorded VaR exceptions well above the anticipated value. But did VaR fail?

Warren Buffet once said that “Risk comes from not knowing what you’re doing.” Risk managers relied on the VaR measure to predict risk without taking into account its limitations. False confidence in VaR led to financial institutions taking on unseemly leverage in the form of mortgage-backed securities. And managers were too ready to rely solely on VaR since they could take on this leverage, thus increasing their profits, while “managing” risk. This was the nail in the coffin.

In fact, we are too ready to blame VaR for a universal human failing – greed.

The financial world is becoming more integrated and competitive. Financial innovations, such as credit derivatives, will continue to evolve and so, consequently, will financial institution’s risk exposures. Emerging markets, such as in Asia and the South Americas, combined with increasing international operations mean that we are living in an uncertain world. And our fortunes depend on the fortunes of others more than ever.

We can never aspire to accurately measuring risk. But we should still attempt to contain risk where we can. To ensure some degree of protection against these exposures, we need to combine a variety of risk measures with subjective appraisal of market, credit and operational risks.

There are many instruments in the risk management toolbox; VaR is but one. Other measures to consider include expected shortfall (Conditional Value at Risk) and extreme value theory (EVT). Although suffering from some of VaR’s limitations, expected shortfall better captures the tails of the normal distribution and correctly reflects the benefits of diversification. EVT, while time-consuming to compute, can be used to estimate VaR when the confidence level is high (say 99.9%).

Given the number of tools in the toolbox, we must not only question risk manager’s ignorance of the limitations of VaR, but also their ignorance in the application of a single measure of risk when other useful risk measures are available. At the end of the day, we must also recognise that wilful ignorance of the limitations of risk measurement models is a symptom of a deeper rooted problem – the prevailing attitude in finance towards risk.

Tags: , , , ,

Comments are closed.

Privacy | Copyright matters | Accessibility
Contact us | Feedback | Disclaimer
Opinions expressed in this blog are those of the individual contributors only.
QUT Home | Blog Home