Is it possible to detect housing bubbles?


Answering this question is intrinsically tied to the question of the fundamental value of the asset in consideration. Bubbles are formed if the price of an asset grows faster than its fundamental value or, described in another way, if trade in products or assets with inflated values is conducted. The crux of the fundamental value, however, is that it is fundamentally unobservable. Analysts use several methods to reach a reliable fundamental value but different methods might produce different results. A bubble can only be truly detected if it bursts and Alan Greenspan, the former Fed chairman, even claimed that it is essentially impossible to detect or prevent or address asset bubbles in advance or in real-time. On the other hand, there is a semi-widespread belief that we should and can identify and deal with asset bubbles which should be especially true for large ones. For example, the housing bubble in the United States was identified by several of the most renowned analysts including Robert Shiller and William White chief economist of the Bank of International Settlements as early as 2005.
Housing markets are the type of market that is susceptible to systematic mispricing. This market is quite thin as, for example, most market participants have little experience, transactions are made only infrequently, and asymmetric or incomplete information between buyers and sellers regarding demand and prices is acute. It is possible to separate the methods used to gauge the fundamental value of housing into two broad approaches. The econometric approach uses a reduced-form price equation which is estimated based on some underlying notion of determinants of supply and demand. Potential determinants that are regularly used (all or some) in house price regressions are demography, cost (interest rates) and availability of credit, real income, supply of housing, fiscal policy, and technology. Interestingly, econometric studies on the US and UK housing market that use data sets covering several decades came up with the assumption that the single most important determinant of real house prices is real income and that real house prices have risen broadly in line with income. Therefore, the median real income and house prices provide a gross overview of price-to-income regularly used as a first approximation to justified house prices.
The alternative, more finance-based, approach can be characterised by an underlying notion of arbitrage where the return to investing in housing relative to some other asset are evaluated or the costs and benefits of renting relative to buying are compared. One standard and very simple metric used in this context is the ratio of rental income to house prices. Usually, almost all of the movement in price-rent ratio is accounted for by two factors – the proxy for future growth in rents and the proxy for future returns. A drawback of many finance based approaches is that underlying demand and supply factors enter only indirectly into the model.
The price-to-income and the price-to-rent ratio are useful measures as they give a general idea about fair house prices but they are somewhat flawed by omitting other important factors. Nevertheless, they are popular and often used indicators as, first, the necessary data is usually readily available and, second, the ratios are intuitively understood. Both measures would have also been very useful tools to detect the bubble on the US housing market – a fact which can be deduced by looking at graph 1 and 2. Graph 1 shows the price to rent ratio, whereby it is based on the data of the Case-Shiller National Home Price Index and the national Owners' Equivalent Rent from the BLS. It is easy to see that the price-to-rent ratio left in the year 2000 the boundaries it kept since 1982 and continued to increase with a rapid, unprecedented growth. It is important to note that the originator of the graph has normalised it to 1 in 1982 and hereby indirectly assumes that this is the long-run equilibrium for the price-to-rent ratio. This however must not be so. Graph 2 shows the price-to-income ratio and is based on the Case-Shiller index, and the Census Bureau’s median income Historical Income Tables - Households. The conclusion that can be drawn from the graph is similar to the previous one. Of course, the increase of the graphs far over “normal” levels is not a sufficient condition, allowing necessarily for the conclusion that a bubble is building up, as changes in determinants other than rents and income might move the ratios to new, higher long-run equilibriums. Yet, the “abnormal” movement of the graph should at least induce the question whether the change in other determinants can really justify these increases – especially, as long-time econometric studies for the US and UK housing market have found real incomes to be the single most important determinant for the change in house prices.
In dynamic and volatile markets – regularly found in emerging markets or new-frontier countries – the price-to-rent and price-to-income ratio might be less valuable bubble indicators than in the US or UK as other determinants may be of much more importance for the movements in house prices. Still, we share the opinion with several other analysts that it is possible to detect an asset bubble (or better said: the probability of a bubble) to a certain degree and that governments should address it.

Question Corner

  • What triggered the US subprime crisis? Can you compare it to past crises such as the 80s crisis? [Answer]
  • What might be the best housing finance system? [Answer]
  • Is it possible to detect housing bubbles? [Answer]

Further Questions

Graph (Click to zoom)

print sitecontact
Home Site DirectoryImprintTermsPrivacy