To know this, we must first know what long-term returns on commodities should be. In theory, they should be the product of four things:
■ Volatility. The riskier the asset, the higher the returns on it must be to compensate us for holding it.
■ Our risk aversion. The more we hate risk, the more we need high returns to induce us to take it.
■ Background risk. If we face lots of risk in our everyday lives - in the form of the risk of lower incomes, job loss or the loss of a business - we are less able to take risk in other assets. We therefore need higher returns on those assets.
■ The correlation between background risk and the asset's risk. If an asset does badly when our other assets do badly then it is especially risky and so must offer us high returns. If, however, an asset does well when our other holdings do badly then it is a form of insurance - and we are prepared to pay for insurance.
And here's the thing. By one measure, the correlation between commodity prices and background risk is negative. A reasonable measure of background risk is consumer spending. This is an indicator not just of our wealth and incomes, but also of our expectations. A fall in spending is a sign that bad times are either here or coming, and a rise is a sign of better times, either actual or expected.
Since 1997 - when the ONS's current data begin - the correlation between annual changes in retail sales volumes and in the S&P/GSCI index of commodity prices (in real sterling terms) has been minus 0.1. This tells us that, commodities have moved in the opposite direction to background risk. For example, commodities have sometimes done well when retail sales have been weak - for example in 2007-08 or 2010 - which means they have insured us against bad times on our other assets. Equally, commodities have sometimes fallen when retail sales have been strong - for example in the late 90s and just recently. This means commodity price risk has materialised at times when we are best able to bear risk.
In this important sense, commodities, despite being volatile in their own right, are in fact a safe investment. And safe investments should offer low returns.
Let's put some numbers on this. Since 1990, the standard deviation of the S&P/GSCI in real sterling terms has been 21.2 percentage points. The volatility of aggregate retail sales volumes has been 2.3 percentage points - though of course background risk is much higher for individuals than it is for the whole economy. The correlation between the two has been minus 0.1. And a reasonable coefficient of risk aversion would be around three. Multiplying these four numbers together gives us a justified annual return of minus 0.15 per cent. But in fact, the S&P/GSCI has risen by 0.6 per cent per year in real sterling terms since January 1990.
Now, I wouldn't set much store by these precise numbers - not least because commodity prices are so noisy that we can't be at all sure what their true returns are. Think of the above as a type of Fermi estimate - rough numbers used to get a feel for the problem. (Such estimates are, in my opinion, much under-used in the social sciences.) The key point is that as long as the correlation between commodities and background risk is low or negative, then commodities should deliver low or negative long-term average returns because they are a way of diversifying the other risks we face.
In one sense, this might not be the case: to the extent that commodity prices rise in global economic upturns and fall in slumps, then they are positively correlated with our other risks and so must offer high returns because they are genuinely risky.
In two other senses, though, they can be negatively correlated. One is that any supply disruption which raises commodity prices (such as a war) would tend to squeeze real incomes. That's bad for our shares and jobs.
To see the second, think of commodity prices as depending upon a race between diminishing returns and technical progress. Diminishing returns tends to push prices up; for example, as demand for oil increases producers have to drill in expensive deep seas rather than cheap desert. But technical progress, such as fracking, pushes prices down by increasing effective supply.
Now, imagine a world in which technical progress is weak and so diminishing returns win the race. Commodity prices would rise. But this would also be a world in which we suffer not only a squeeze on real incomes but also pessimism about the future. Our other assets - shares and our jobs - would therefore do badly. Good returns on commodities would therefore insure us against losses on other assets. By the same token, if technical progress wins and commodity prices fall, gains on our other assets would offset commodities' losses.
It might seem paradoxical given their huge volatility, but in this sense commodities are a safe asset because they insure us against some bad times. They are not wholly safe, given that a global recession would depress both them and our other assets. But they are protection against some types of risk.
The case for investing in commodities, therefore, is completely different from what is often thought. We should regard them not as a speculative asset offering the possibility of high returns but rather as a form of insurance. And insurance policies don't offer high average returns.
