Why measuring market risk is a path rife with dangers

Is it possible we underestimate market risk? Consider the range of possible outcomes from the recent repealing of the rate-cap law for instance (out of the myriad that could also be chosen): Those involving its possible impact on company profitability, default rates and pricing.

How confident are investors that the outcomes will not cause the market to implode? Are investors correctly measuring their tolerance for risk on possible future reactive policy decisions? Is our reliance on “historical-based” risk measures unfounded when the future is full of uncertainties?

To be clear, market risk is not volatility (we use it since it’s easily quantifiable). Only schooled guesses are our best estimates of future risk.

For that reason, in today’s article, we look at the underlying weaknesses of the main risk measures and appreciate their “risk” attached.

Let’s start with the Value at Risk (VaR) – measures the minimum loss in either currency units or as a percentage of portfolio value that would be expected to be incurred as a certain percentage of the time over a certain period of time given assumed market conditions – as a key measure.

Besides being highly subjective, it’s known to underestimate the frequency of extreme events and often fails to account for lack of liquidity. As a result, it’s blamed for oversimplifying the picture of risk.

Standard deviation – measures how much an investment’s returns differ from its average return – on the other hand assumes a normal distribution pattern or the bell curve. What’s problematic is that extreme stock returns occur much more frequently than expected. In other words, in “real life”, asset-class return distributions have fat-tails: there’s a lot of stuff at the ends we don’t know.

Sharpe Ratio – which shows how much additional return we earn by taking additional risk – has lately attracted serious criticism over whether it’s actually useful in comparing different portfolios.

Research now shows that the ratio assigning the same score to horribly different portfolios. In other words, it’s measuring risk inaccurately.

Beta, on the other hand, is another useless metric. Past beta is not a good predictor of future beta for stocks. In fact, research shows that betas eventually revert back to the mean over time.

This means that higher betas will tend to fall back toward 1 and lower betas eventually rise toward one. Regression to the mean — that good periods are more likely than not to be followed by poor ones, and vice versa — is not a sure thing as some would want us to believe.

Just remember the Japanese stock market is trading for barely 50 percent of where it stood 30 years ago.

To close, investors need to understand that risk management is more art than science. They need to use human judgment.

Theory is only theory. Perhaps, the greater risk may lie in relying too much on magic numbers. Yes, numbers don’t lie but the past might never repeat itself.