How to spot a house price bubble

House prices have become so high that first-time homebuyers often find it hard to get on the property ladder. In particular since the 1970s, house prices, adjusted for inflation, have started to accelerate. This is visible as a steeper increase in real house prices in the chart below. But what drives this acceleration and how can we spot a real estate bubble in the making?

Martin Droes and Alex Van de Minne have examined Dutch house prices since 1900 to identify the main drivers of house prices in that country. Their results are informative on two levels.

House prices in the US, the UK and Netherlands since 1900

Source: Jorda et al. (2019), MacroHistory Lab

First, they show that until the 1970s, house prices were mostly determined by the cost of materials and labour to build a house, the supply of houses and population growth. But since the 1970s the drivers have shifted. Between 1970 and 2012 about 70% of the house price increase can be explained by the increase in capital and the cost of capital, while an additional 20% is explained by population growth. In other words, people have increasingly bought houses on debt (i.e. using mortgages) rather than buying them outright. And as mortgage rates dropped, people could afford to pay more for a house.

Today’s housing affordability crisis is simply a reflection of low interest rates, but it also implies that building more houses is unlikely to alleviate the problem for long. People will simply buy more houses and bid up the price of these houses again. It seems the only way to reduce house prices is to increase the cost of mortgages or reduce population growth even more than is already the case. But that is pure conjecture on my behalf…

In any case, their research points to a second important lesson for all investors. They find that if people do not allow for drivers of prices to change over time but instead assume a stable equilibrium in the long run, significantly overestimate the appearance of a house price bubble. What the research on house prices found was that over time the influence of different drivers changes relative to other drivers. If one examines house prices using a very long time frame, one implicitly fixes the impact of each factor to the average over many decades. The result is that the model becomes less predictive of what is going on in house prices and overestimates the possibility that house prices have deviated too much from fundamentals. Using models that allow for the influence of different drivers to change over time gives a more accurate result and prevents investors from identifying bubbles where there are none.

And I think this is a lesson that all investors need to heed. Just think of all the people who argue that the long-term average of equity market price-earnings ratios is 16x to 17x. Hence, markets trading at PE-ratios of 20x or more necessarily have to be in a bubble or at least overvalued. The flaw in this argument is the implicit assumption that the long-term average PE-ratio is still relevant in a time when capital markets, interest rates and inflation are markedly different than 50 or more years ago. And by sticking with such long-term models and its implicit assumptions of stable drivers of value, one can come to very different conclusions and for example avoid equity markets even though they are not in a bubble at all.

I have written about this ‘long-term effect’ for interest rates here and shown how your assessment if interest rates are low or high can change dramatically based on how far back in time you look. The research of Droes and Van de Minne shows that the people who see bubbles in financial markets implicitly assume that long-term relationships between drivers of markets and market valuations are constant over time. But remember, there is no law of gravity in finance and these relationships change all the time, making the identification of bubbles harder and more flawed than one thinks.