Wednesday 16 December 2009

On dollar, risk models, and market stability

Buy dollar and sell stocks. This piece of advice has been handed out by investment advisers recently. They refer to the correlation between dollar and the stock markets, according to which a stronger dollar automatically should lead to weaker stock markets. I don't think so. The dollar strengthening is likely to be the result of a decreased willingness to carry leverage in a currency already cheap on almost any indicator. The stock markets have not dropped correspondingly because investors increasingly believe that the economic recovery eventually will lead to improved top line growth in most companies. Two entirely different stories.

This belief in a stable connection between dollar and the stock market touches upon one of my main criticisms against the financial institutions, the use of risk models.

Even if this is a subject for an academic discussion rather than a blog post I will try and explain. Virtually every model used to control risk or "optimise return" builds on a number of assumptions regarding the statistical properties of the markets. Unfortunately, two central assumptions have repeatedly been proved wrong. One is that the daily investment returns are log-normally distributed. The other is that correlations between asset classes are stable over time.

If we leave aside the issue of the distribution of investment returns and concentrate on the stability of the correlations, we can take as a point of departure the USD/EUR exchange rate and the stock markets. This observed correlation has been rather stable over the past two years. Episodes of dollar weakening have coincided nicely with periods of increasing stock markets.

In other words, if there are reasons to buy dollar then investors should also sell stocks. Interestingly enough, the causality is not explained. No explanations are given WHY a stronger dollar should lead to weaker stock markets. There is certainly no economic theory supporting this story, because if the a stronger USD were to drive stock markets down, it should only be the markets in the USD based economies as they lose competitiveness. European stock markets should by the same logic fact strengthen.

Lacking a theoretical foundation, one could argue that the carry trades, borrowing dollars to buy stocks has been driving the correlation. That could work for the past 9 months, but certainly not for the time between September 08 and March 09. During this period the explanation should have been that investors were shorting the stock markets in order to buy dollars??? Surely not.

Most likely, the observed correlation between dollar and the stock markets is just spurious – as it has happened many times before. There is no single explanation, just a series of events that have led to this observed connection between the two asset classes.

Risk models are always backward looking, since they are based on historical data – what else? They are constantly recalibrated in order to have the most recent developments integrated. Since the connection between dollar and stocks has been visible for two years, the models will now have picked it up and use it as an element in calculating portfolio risk.

But what happens if for some reason the correlation changes? For the users of the risk models, it means that the risk calculation is wrong for the time it takes for the models to detect the changes. Unfortunately, risk models are complicated and require a lot of daily observations in order to pick up a new trend. In this period the portfolio risk is likely to be much higher than stated by the models. Because a stable correlation between two asset classes can be used to hedge risk.

For the financial institutions that build their advice on the same connection between dollar and stocks something much simpler happens: they simply give their clients wrong advice for the period it takes to realise that something has changed.

The conclusion of all this not only that risk models are inherently misleading, and that they probably underestimate risk. It is also that the financial sector has a tendency to present ill-founded short term observations as undisputable long-term facts. Either way, the investors lose because the habits and the techniques in the financial markets are not tuned towards the detection of changes.

No comments: