To see why, just ask: what sort of returns should we need on housing? Economists have an equation to answer this. It says that required returns should be the product of four things:
■ Our risk aversion. The more we hate risk, the higher are the returns we need on risky assets if we are to hold them.
■ The standard deviation of the asset's return. The greater this is, the higher must be returns.
■ The standard deviation of our consumer spending. If we are likely to fall upon hard times, we are less able to gamble with our wealth. We'll therefore need high returns to reward us for doing so.
■ The correlation between the asset's return and our spending. If an asset does badly in bad times, it is especially risky, and so we need high returns on it. By contrast, assets that do well in bad times are like insurance policies, and we'll accept low or even negative returns on these.
We can apply this equation to any asset. Let's do so for housing.
Since 1955, the annualised standard deviation of annual changes in the Nationwide's house price, after inflation, has been 9.1 percentage points. But just as individual shares are more volatile than the All-Share index, so individual house prices have been more volatile than the house price index. Surprisingly little work has been done to estimate just how much more volatile they are, but let's assume they are 2.5 times as volatile, giving us a standard deviation of 23 percentage points.
We know that the standard deviation of annual changes in aggregate real consumer spending since 1955 has been 2.6 percentage points, but of course individuals face more risk than this. How much more will vary from person to person - being less for retired folk than workers - but let's say its three times as much. This gives us a standard deviation of 7.8 percentage points.
In aggregate data, the correlation between house prices and consumer spending has been 0.67 for annual changes since 1955. Let's run with this.
This leaves our coefficient of risk aversion. Pinning this down is tricky - some economists believe impossible - but let's call it three, where one equals risk neutrality and higher numbers more risk aversion.
Putting these numbers together tells us that expected returns on housing should be: 3 x 0.23 x 0.078 x 0.67 = 0.036, or 3.6 per cent a year.
This compares well to the facts. Since 1955, Nationwide's house price index has risen by 3 per cent per year, with a standard error of 1.2 percentage points. By all means, quibble with my numbers - but I suspect that any reasonable tweaks to them wouldn't greatly overturn their consistency with the historic data.
Now, this leaves out a lot of things. I've ignored the costs of home ownership such as insurance and maintenance, but also the psychological benefits of putting down roots. I've ignored the fact that, for many, housing is a leveraged investment - although this increases risk as well as returns. I've ignored the fact that housing is risky not just because of its volatility but because of liquidity risk, too; in bad times, it's difficult to sell a house quickly. And I've glossed over the facts that risk aversion and income volatility vary from person to person, which means that housing is a better investment for some than others.
These omissions, though, probably don't overturn the key point - that from the point of view of many homeowners, housing isn't a fantastic investment. Yes, 3 per cent per year in real terms is a nice return. But it's only a reasonable reward for the risks of investing - which are not just the volatility of house prices themselves, but the danger that they'll do especially badly in bad times, when we are least able to cope with losses.
All this doesn't just have investment implications, but political ones, too. It's often said that politicians want to see house prices rise in order to satisfy older voters. But it might also be that we need decent rises in them merely to justify home ownership as an investment at all.
