To answer this, of course, we must first define defensives. There are many possibilities. For the purposes of this inquiry, I define them as an equal-weighted basket of six sectors: utilities, food producers, pharmaceuticals, tobacco, beverages and food retailers. These have been the most defensive FTSE sectors (defined by monthly betas) since December 2000.
This basket has done well. Since December 2000, it has risen by 165 per cent while the All-Share index has gained just 26 per cent.
To identify the circumstances in which is has done especially well or badly, I looked at annual changes in this basket relative to the All-Share index and asked: with what factors are these changes correlated? Surprisingly, just four things explain (in the statistical sense) three-quarters of the variation in defensives' annual relative returns.
One of these is the performance of the All-Share index itself. When the market rises, defensives tend to underperform. This only confirms what we know - that defensives have a low beta.
A second factor is the Alternative Investment Market (Aim). Defensives underperform when this does well. This captures investor sentiment. When sentiment improves, investors switch out of safer stocks and into riskier ones such as those on Aim.
Our third factor is index-linked gilt yields. Defensives do well when these rise. But the effect is small. A one percentage point rise in 10-year index-linked yields (which is a massive move) is associated with my measure of defensives outperforming by only three percentage points.
This, I suspect, is because a rise in index-linked yields has an ambiguous effect. On the one hand, yields rise when investors are more confident about taking risk and so sell safer assets. These are circumstances in which defensives should underperform. But, on the other hand, higher long-term interest rates mean that future cash flows should be discounted more heavily. This is bad news for growth stocks, which means that defensives should outperform. On balance and on average the latter effect is the more powerful one.
Our final factor is commodity prices. Here, there are mixed effects. On the one hand, defensives do relatively well when oil and gold rise. This is consistent with them being relatively safe havens against the uncertainty that often drives up gold prices and which results from a rise in oil. But, on the other hand, defensives do badly when other commodities rise: there's a negative link between defensives and the S&P/GSCI. This is probably because industrial commodities are a cyclical indicator. When they rise, it's often because the global economy is doing well, and this benefits cyclical stocks. But if cyclical stocks are doing well, defensives underperform.
All this is pretty intuitive. But there are two surprises here.
One is that many things aren't statistically significantly correlated with defensives. Inflation, industrial production, exchange rates and nominal interest rates don't seem to matter for them. There's a lot, therefore, that holders of defensive stocks needn't worry about.
Secondly, even if we control for the things that do matter, defensives do well. My measure of them has beaten the All-Share index by 5.9 per cent per year, controlling for everything else. This confirms that defensives have generally been underpriced. This is consistent with the possibility that some fund managers underweight them for fear they'll underperform their peers in a bull market.
Now, I wouldn't get hung up about the precise numbers here: precision can be the enemy of truth, and I've used only one of countless possible measures of defensives. Nevertheless, this seems to confirm the case for defensives. They generally do well, and the main risk to their relative performance is when the market shoots up. For retail investors, however, this is a circumstance in which underperformance can be tolerated.
