GOING TOO VAR
Value-at-Risk is used widely in the US and Europe. But can CFOs rely on this statistical method in volatile Asian markets?
By Elizabeth Fry

Value-at-risk (VAR), a computer-driven technique long used by banks to assess liquidity and currency exposure, is now being touted for treasurers at Asian corporations. Widely used by multinationals such as Enron, the US natural gas supplier, and Siemens, the German industrial company, VAR has been slow to catch on in Asia's volatile market. Because VAR breaks down under extreme volatility, this skepticism could well be warranted.

VAR is a measurement process that translates the degree of risk in a portfolio (or in derivative transactions) into losses, expressed in dollar terms. It estimates the potential losses that occur when underlying markets fluctuate over a defined period of time for a given level of statistical confidence.

The technique is still used mostly by banks, which need to adjust their capital needs according to the degree of risk on the balance sheet on any given day. It was developed by a JP Morgan unit called RiskMetrics, a brain-trust of risk management bankers set up by Morgan in 1994 and spun off as a separate entity in 1998. RiskMetrics has long provided VAR models for banks, but last year it teamed up with Ford, Procter & Gamble and a few of the Big Five accounting firms to develop a version of VAR, called CorporateMetrics, for general corporate use.

RiskMetrics has since opened an office in Tokyo with the idea of spreading VAR to Asian corporations and increasing its use by banks. Proponents say that the arrival of VAR in Asia heralds the maturity of the derivatives market here. But like pharmaceuticals that come with a list of caveats on the label, VAR models come with lengthy descriptions of what it can't do. VAR generates statistical probabilities and its predictions are approximate. Its role is to give treasurers and risk managers a broad view of risk in a portfolio, and to show when risks are likely to push losses past a company's tolerance level. Even derivatives bankers, who have been using VAR to mark their instruments to market for years, caution against a complete reliance on the technique. "You shouldn't get too caught up in a single measure of risk," says Peter Colvin, senior vice president of derivatives at ABN Amro in Singapore.

The problem is that events tend to be more predictable on a short-term basis. If the market is volatile today, it is likely that the volatility will continue, at least for a short period. But long-term VAR forecasters face the conundrum of weather forecasters, who famously can't predict rain a week ahead. So many variables contribute to the puzzle, that a prediction eventually becomes guesswork.

Dramatic Predictions

Another, more serious short-coming hobbles the technique. VAR models are based on statistical methods that generate probabilities from data sets based on historical shifts in prices. But volatile changes in prices, like those that followed the currency devaluation in Thailand, "spike" probabilities to such a degree that traditional statistics simply ignores them. The result is that VAR cannot forecast dramatic changes in prices, such as predicting another currency collapse or liquidity crunch. According to Francis Heng, treasurer of Jardine Matheson, this makes the VAR model overly simplistic. By saying that the past will be repeated, it makes no allowance for extreme events - like the currency falls in Malaysia and Thailand. "So what good are these models?" he asks. Which is not to say that companies shouldn't measure risk, just that VAR is not a panacea for risk management. "It is a reporting tool and an imperfect one at that." More importantly, according to Heng, it does not help you hedge your risk."

The technique's defenders say that most price movements in markets are moderate, and the big swings happen so rarely that they can be excised from the calculation without destroying the model's accuracy. But this view has its critics. Benoit Mandelbrot, the inventor of non-linear geometry and a renowned researcher for IBM, has compared VAR to creating probabilities for sea waves, but discounting all waves higher than six feet.

His withering view has not swayed US regulators, who last year gave a resounding approval to VAR. The Securities and Exchange Commission issued several rules requiring public companies to disclose the value and level of risk in their derivatives portfolios. The regulator designated VAR as an acceptable tool for making calculations.

The Financial Accounting Standards Board also has proposed guidelines on disclosure of derivatives risk (FAS 133), and tipped its hat to VAR as one method for calculating portfolio risk exposure.

VAR models have won their widest acceptance at banks and investment houses. Indeed, the Bank for International Settlements, a global regulatory body established by central banks, has mandated that banks use VAR daily to help them comply with newly revised BIS capital requirements.

But treasurers that use VAR at large companies to measure currency exposure say that volatile Asian markets make its weaknesses stand out.

Teodoro K. Limcaoco, head of treasury for the Philippine conglomerate Ayala Corporation says: "VAR tends to be either too conservative or overstate the exposure." Still, he's intent on using VAR until something better comes along.

Elizabeth Fry is a contributor to CFO Asia based in Sydney.