Friday, August 9, 2013

When historical correlations fail to hold up

This summer had investors across asset classes and geographies puzzled by the "wrong" correlation between treasury yields and risk assets (such as corporate credit and equities). Historically, it was argued, that the correlation has been strongly positive. That is, when the treasury yields rise (or fall), the prices of risky assets rise (or fall). I would argue that the dismay with the rationality of this year's correlation was misguided in more ways than one. Firstly, when we talk of historical correlations, we tend to think of a continuous time series between, let's say, 40 years ago (assuming that to be a ballpark time at which markets, at least US markets, decisively matured) to today. While it is true that this correlation shows itself to be markedly positive, a strong positive correlation needn't necessarily allude to an economic thumb-rule. And in this case it doesn't. It is important to see that there have been pockets of time in the last 40 years when the correlation was negative, but since they weren't the norm, they don't affect your long term coefficient of correlation in any big way. But just because they weren't the norm does not mean they were periods of random noise. It might do us good service to identify those specific time-periods and then analyze the difference between those time periods and the rest of history. From a study such as the one I just outlined, my far from exhaustive observations suggest an element of causality as a big differentiator in the correlation behavior. That is, it is important to consider what's causing the rates to rise - good economic outlook (in which case the positive correlation does indeed make complete sense and holds up almost always) or something else (lack of confidence in the government (as in 2008), or Bernanke sneaking away his Santa Claus hand (this summer)).

Traditional wisdom for the positive correlation goes like this: The rise of treasury yields (or in other words the falling of treasury bond prices) reflects a transfer of the world's funds from riskless to risky assets and vice versa. Naturally, as a result of this transfer of funds into risky assets such as high-yield credit and equities, they rise. All very well, except this and that and that. We saw this in 2008 at the time of the financial crisis that the treasuries fell (yields rose) and equities fell too. Any positive correlation between yields and equities was thrown out of the window. Now lets try to see why the traditional wisdom is far from a holistic perspective. Most of all, it's because it relies too much on correlation and correlation alone, while only superficially digging into causality. At its core this conventional wisdom makes one grand assumption, that money either flows from risky assets (HY, Equities) to riskless assets (treasuries, gold) or the other way. Such a transfer always gives us a conveniently positive correlation between treasury yields and risky assets, and we rejoice. What it does not take into account is that in times of unprecedented uncertainty (such as Bernanke leaving the markets alone, in the present case) people don't necessarily merely shift their money from one asset class to the other - they may wait for the uncertain period to pass, holding cash, and taking off again only when the sky gets clearer. In such times, it is not uncommon to see both sovereign bonds and equities falling, as we saw this summer, and have before. This is also the reason why the 'aberrations' or breaks in the utopian positive correlation between yields and equities are much more common in the rising rates environment (associated with things going wrong) than in the falling rates environment.

No comments:

Post a Comment