Portfolio secrets

This is childishly obvious. But many investors ignore it when they are building their portfolios.

There’s a golden rule here: what we should buy depends on what we already own.

Some assets merely duplicate others, and so are redundant – equivalent to having two car insurance policies. Conversely, If you’re exposed to the risk of big losses you might well need assets that do well in such an event. Just as you need insurance against a car crash, you might need insurance against a financial crash. 

This might seem obvious. But many people ignore it. In a study of 20,000 German investors Daniel Dorn and Gur Huberman, two US-based economists, show that a “substantial number” are prone to what they call “narrow framing”. They evaluate stocks one at a time in isolation and do not ask whether a particular stock or fund adds to the risk of their portfolio or reduces it.

Similarly, in laboratory experiments with undergraduates, Erik Eyster at the London School of Economics (LSE) and Georg Weizsacker of Humboldt University in Berlin have found that people “neglect correlation in asset returns”. When correlations are positive they diversify too much and take on more risk than they think. And when correlations are negative they diversify too little and so take on more risk than they need to.

In fact, things might be even worse than this. Yann Cornil at the University of British Columbia and Yakov Bart at Northeastern University in Boston have found that some investors misunderstand correlations completely. They perceive negative correlations as adding to their risks, and high correlations as reducing them. They do so, they say, because they regard assets that move in the opposite directions to ones they own as being unfamiliar, and so they shy away from them.

What’s more, investors’ perceptions of how to spread risk are distorted by the options available to them. The Nobel laureate Richard Thaler and his colleague Shlomo Benartzi have shown how. They studied the pension investments of airline pilots and teachers, and found that they tended to divide their money equally across several funds. In itself, this isn’t disastrous. Except that those investors who were offered many equity funds and one bond fund invested plenty in equities, while those offered lots of bond funds and one equity fund bought more bonds. Diversification decisions, then, are determined not so much by correlations as by irrelevant information – the menu of assets on offer.

The evidence, then, is clear. Many people ignore or misunderstand correlations and so make bad asset allocation decisions. Some end up with more risk than they bargained for, and others with less.

For example, viewed in isolation many equity unit trusts might look attractive. But most of them are highly correlated with each other simply because almost any basket of stocks tends to rise and fall as the global stock market rises and falls. Investors who judge funds in isolation can therefore easily end up holding a portfolio of funds that moves as a tracker fund does, except that they pay thousands of pounds extra in management charges.

We can, however, avoid these errors. We should start from a basic fact – that what matters is your portfolio as a whole. Assets are only good or bad insofar as they affect the risk and return of our whole portfolio.

Car insurance is a good asset if we own a car and a lousy one if we don’t. Exactly the same is true for shares and funds. Whether you should buy them depends on a simple question: what does this asset add to my portfolio?

There are only three possibilities here. It might add insurance, by reducing your portfolio risk. Or it might add risk, by increasing it. And, of course, it might add returns while leaving portfolio risk unchanged.

Let’s take an example – foreign currency. Viewed in isolation, this is of dubious value. It might even lose us money, because sterling might well rise in coming months.

However, on the two occasions in the past 30 years when we’ve seen a significant fall in house prices, sterling has also fallen, meaning that holding foreign currency has paid off. Euros and US dollars, then, have provided insurance against falling house prices. Just as car insurance loses you money in good times – if you don’t have a prang – but pays out in bad, so too can foreign exchange. If you’re worried about a fall in property prices, therefore, foreign currency might be useful insurance.

Here’s another example. For much of the 1990s and 2000s, Japanese workers suffered a double hit: stagnant real wages as the economy faltered and falling share prices. Investors who held non-Japanese stocks avoided the worst of this, as profits on those shares offset falling real wages. This tells us that investors who are exposed to the domestic economy – because they need to earn a living or own a cyclical domestically-oriented business – should hold overseas shares to spread their risks.

Or take another example. If your money is tied up in illiquid assets such as property or your business, you might well need to hold lots of cash. And illiquid assets such as private equity are especially unattractive for you as they compound a risk you already carry – that of liquidity drying up.

The converse of all this is also true. If your other assets already give you insurance you can afford to take risks that others can’t. If you have a secure income – say because you have a final salary pension – you can afford to take on the risk of recession better than can someone owning a small business. Housebuilding stocks, then, are more attractive for you than they are for them.

Let’s apply this more concretely. To do so, we must start with a particular portfolio. Of course, everybody’s is different but for the sake of illustration let’s consider a balanced(ish) portfolio comprising 50 per cent in global equities (measured by MSCI’s world index), 20 per cent each in cash and gilts, and 10 per cent in gold. Such a portfolio would have given us decent and quite stable returns. It only lost significant sums in 2001-03 and in 2008-09.

We can now ask of any asset: does it add to the risks of this portfolio, or does it reduce them?

We have a simple measure to tell us this: beta. If an asset has a beta with respect to this portfolio of more than one it adds to its volatility. In effect, it gears up the portfolio. A beta of less than one, on the other hand, reduces risk.

Table 1 shows these betas for some main assets. They show much what you’d expect – that equities add to portfolio risk while bonds reduce it.

Only one asset in this table, however, has a negative beta; it tends to move in the opposite direction to the portfolio. This is the euro. It has been especially effective at diversifying the risks of a balanced portfolio.

One way to reinforce this point is to change the way we describe assets. Forget their names. They should instead be described by numbers – by their volatilities, correlations with other assets and alphas (that is, their returns controlling for risk). If two assets – shares, funds, whatever – move in the same way, then they are the same asset for practical purposes whatever their names are.

Table 2 gives an idea here. It shows correlations between major assets since 1991. A correlation of one means assets always rise and fall together; a correlation of zero implies that one is as likely to fall as rise if the other falls; and negative correlations mean one has a better than 50:50 chance of rising if the other falls. The point here is that major stock markets move together. That means that many equity assets largely duplicate each other except to the extent that one might offer higher returns.

Note also that no asset offers perfect insurance against another, in the sense of having a correlation of minus one.

None of this is to say that you need to do the maths or even that you need precise numbers. The key thing is that you ask the question of any share or fund: what does this add to my portfolio? You can do this informally, by asking: in what possible states of the world would my portfolio do badly, and how would this asset perform then? Just remember that one significant possibility is a general fall in world equities, which tends to drag down most shares.

So far, we’ve considered what adds risk to a balanced portfolio and what doesn’t. But what adds returns? Again, a simple measure helps us: alpha. It shows an asset’s returns controlling for its sensitivity to our balanced portfolio. A positive alpha adds value, in the sense of giving us extra risk without adding to our portfolio’s volatility (this isn’t to say it’s a perfect measure, because a high alpha might be only a reward for taking on some risk other than extra volatility, but this is another story).

Taking the period since 1990 as a whole, major equity markets have had negative alphas: their returns haven’t justified their extra risk. This, though, might not be the case in future. It might instead be merely an artefact of the fact that bonds have done extraordinarily well since the 1990s and so made equities look bad.

If we look at UK equities from this perspective, however, we see an astonishing pattern. Since 1990 there has been a massive negative correlation between sectors’ betas with respect to a balanced portfolio and their alphas. Across 25 major FTSE sectors, the correlation coefficient has been 0.84, which is astoundingly high by the standards of financial data.

Sectors with high betas – those that add to portfolio risk – have had negative alphas; this is true for miners, financials and tech stocks. Sectors with low betas on the other hand have had high alphas. These are defensive stocks such as tobacco and food producers.

For investors with balanced portfolios, it has not paid to take on risk. Quite the opposite. Defensive stocks have done better than they should.

Why? One possibility is that investors have underrated the virtues of dull stocks with sources of monopoly power such as big brand names (think of Diageo or Unilever) or high capital requirements such as utilities. Another possibility is that investors who are bullish of the market but are prevented from borrowing to buy shares generally seek leverage instead by buying high-beta stocks. This causes them to be overpriced and defensives to be underpriced.

This doesn’t mean defensive stocks are great risk-reducers: their betas are higher than the betas on (say) gold, cash or gilts. But it does mean that they have added value to balanced portfolios.

Nor, of course, does it follow that future alphas will resemble past ones.

Instead, the point is simply that investors should not consider assets in isolation. We must instead ask: what does this add to my existing portfolio? And if the answer is only more stock market risk, then history warns us to be very sceptical of it.