An Empirical Investigation into the Presence of Chaotic Behaviour in UK Property Markets

Citation

Lizieri, C. (1998), "An Empirical Investigation into the Presence of Chaotic Behaviour in UK Property Markets", Journal of Property Valuation and Investment, Vol. 16 No. 4, pp. 422-423. https://doi.org/10.1108/jpvi.1998.16.4.422.1

For a few years, the complex mathematical concepts known collectively as chaos theory captured the public imagination. Popular texts such as Gleick (1988) or Stewart (1990) made the basic ideas accessible to general readers, while computer‐generated fractal art created a visual counterpoint. However, the principles, and, more importantly, the implications, of chaotic systems are subtle and difficult. Simplistic versions: “a butterfly flapping its wings in China alters the weather in Reading” led to a rash of ill‐judged pseudo‐scientific articles. The property market was not immune from this nonsense as authors (no names, no references!) claimed that chaos theory implied that real estate performance was unforecastable and not amenable to quantitative analysis. This conflates chaotic systems with more popular concepts of chaos. A chaotic system, by its nature, is not random, it is strictly deterministic. However, in the presence of shocks, its behaviour may be highly volatile.

Graeme Newell and George Matysiak present the results of an empirical investigation of chaotic behaviour in UK property markets, a project funded, encouragingly, by the Royal Institution of Chartered Surveyors’ Education Trust. Their paper is a useful antidote to populist nonsense: a careful, quantitative analysis of property company share prices using a battery of statistical and mathematical tools. To begin to test for chaotic or non‐linear structures, it is necessary to use property company data since there are insufficient data in the direct market. The hunt for chaos consumes large amounts of information.

Newell and Matysiak’s results confirm many studies of financial market performance series and are consistent with the handful of published real estate studies. They find little conclusive evidence for chaotic structures. However, there were stronger indications of non‐linear behaviour even though the precise stochastic non‐linear form did not emerge. This does suggest that existing forecasting models may need to be modified or reviewed to account for the presence of non‐linearity. This is likely to be a fertile area for research in the future.

The authors note that the use of property company data presents a problem since, although “in the long term, commercial property share price performance should reflect the performance of the underlying asset”, in the short term property shares behave more like the stock market than the property market. This presents a problem in analysis. If the underlying property market were chaotic, but equity prices were not (I cannot think of an a priori reason why this might be so), then the correlation of property company shares with other stocks may mask the chaotic structure. Unfortunately, we are unlikely to be able to test this in the foreseeable future. One avenue might be to attempt to decompose property share prices to isolate any unique real estate factor and test this for chaotic and non‐linear structure.

Without criticising the empirical research underlying Newell and Matysiak’s report, I should state some reservations. These essentially relate to presentation. The style is terse, dry to the point of aridness. There are more pages of tables than of text. As a result there is insufficient explanation of the techniques used, of the findings or, critically, of the implications of the results. To take an example at random (!) we are told (p. 5) that Lyapunov exponents will be calculated: but not what these measure nor why they are used. Page 7 tells us that Table 3 shows estimates for “the embedding dimensions of 3, 6 and 9”, whatever those are, and that this provides only tentative support for chaos. Table 3 itself shows 144 coefficients but provides no information on what these mean or which are significant. This is the general tone of the report: to a reasonably quantitative lay reader it is, frankly, unhelpful.

At the heart of this criticism lies the purpose of the report and its intended target audience. If it is aimed at a readership already familiar with these techniques, then the terse approach may be justified, the paper is a technical note. I suspect, however, that the intention is rather different: to provide a rigorous counterpoint to superficial assertions of chaos. If this is so, then the attempt falls short. The paper needs to have a balance tilted much more to explanation and discussion, with the results supporting the debate not dominating the report. If the level of analysis and research in property markets is to improve, it is incumbent on researchers to make their findings accessible to those in practice willing to listen and learn. This does not mean lowering standards but improving communication. Where results are presented in an opaque fashion, they can have little, if any impact. The debate is not advanced and the gap between academic research and professional practice becomes still more of a chasm. There is much to commend in this research ‐ but as a RICS report, the message needs to dominate the method.