Join our community of smart investors
OPINION

Predicting house prices

Predicting house prices
June 10, 2015
Predicting house prices

Since 1986, just three indicators have, taken together, been powerful predictors of the subsequent three-year change in inflation-adjusted house prices. These are:

- Affordability, as measured by the Nationwide's data on mortgage payments as a percentage of first-time buyers' disposable income.

- The dividend yield on the All-Share index. The idea here is simple. House prices are slow to respond to changing economic conditions whereas share prices respond more quickly. In 2008, for example, share prices fell quickly during the financial crisis whereas house prices fell slowly and gradually. A high dividend yield thus predicts bad economic times and hence falling house prices.

- Momentum. If house prices rise in one three-year period, they tend to continue rising; the market is prone to bandwagon effects.

These three factors, taken together, can explain three-quarters of the variation in subsequent three-yearly real changes in house prices. They successfully predicted the house price falls of the early 1990s and 2007-10.

Right now, these indicators point to house prices rising only very slightly - by just over 2 per cent in real terms by December 2017, which implies a nominal increase of less than 10 per cent. This implies that housing will probably give us a slightly lower total return than equities over the next two or three years.

The reason for this is simple: affordability is a problem. Because prices are so high, mortgage payments are an above-average fraction of first-time buyers' incomes even though mortgage rates are low - although affordability is much better on this measure that it was in 1990 or 2007.

I stress that I'm talking pure predictability here: we're using information now to predict prices in three years' time. We don't need to make any assumptions about future economic conditions, other than that past relationships will continue to hold. However, we can say that if - as is expected - mortgage rates rise next year then the affordability problem will get worse, and threaten to force house prices down.

You might think this model is too simple to be plausible. (I think such an objection is mistaken - but that's another story.) However, there are some other variables that have predicted house prices, but which become statistically insignificant when added to my three factors. These, though, are sending mixed messages. They are:

- The ratio of consumer spending to house prices. When this is high, house prices subsequently rise. As it is now below its 30-year average, it points to only small rises in prices.

- The ratio of household bank deposits to house prices. High cash holdings predict rising prices. This ratio is above its 30-year average, pointing to rising prices in the next three years.

- The gilt yield curve. If this is inverted - in the sense that 10-year yields are below three-month interest rates - house prices tend to fall. This is because inversions lead to recessions. As the curve is now upward-sloping, however, it predicts rising prices.

Of course, predictability is imperfect; the margin of error in my model implies that there's around a one-in-four chance of house prices rising 10 per cent or more in real terms between now and the end of 2017 even of past relationships continue to hold. Nevertheless, the message seems clear. Unless you think affordability no longer matters, we should not expect big returns on housing from current levels.