Abstract
We show that a proper assessment of the linkages between real estate markets and the economy requires state of the art modelling techniques, which treat economic variables endogenously and allow for a number of long-run relationships. We therefore use a long-run structural modelling approach, which incorporates equilibrium relationships that are predicted by economic theory, in an otherwise unrestricted vector autoregressive model. The application of this approach to Swiss multifamily residential data shows that four long-run equilibrium relations exist among inflation, long- and short-term interest rates, real M2, real GDP, real construction expenditures, real market rents and capitalisation rates. Disturbances to the equilibria last for approximately five years before they completely vanish. The analysis of the short-run dynamics additionally suggests that the linkages between real estate and economic variables are bi-directional. Our findings should provide for a better understanding of the linkages and feedback mechanisms between a developed economy and its real estate markets and thereby help in the identification and quantification of both market interventions by policy-makers and risks borne by investors.
Notes
1. Since 2000, end-of-year sales for clothes and footwear have been included which introduces seasonality in the index, creating a need to smooth the series. The post-2000 series is smoothed by taking the 12-month average of the clothing and footwear sub-index which is added to the CPI without clothing and footwear, using the appropriate weighting.
2. See Garratt et al. (Citation2006, pp. 67–85) for details and Assenmacher-Wesche and Pesaran (Citation2009) for an application to the Swiss economy.
3. Ling and Naranjo (Citation1997) and Liow (Citation2004) apply these pricing models to commercial real estate in the US and Singapore, respectively. They find that the return above the risk-free rate is time varying and that economic variables such as the per capita consumption, GDP growth, interest rates, the term structure of interest rates, and unexpected inflation help to explain the time-varying behaviour.
4. Exact identification of Equation (Equation5(5) ) would require 16 restrictions. The -matrix contains 32 parameters and the intercept and trend vectors together another eight. Out of these 40 parameters 30 are restricted, hence 14 restrictions are over-identifying.
5. For example, persistent profiles of all four cointegrating relations tend toward zero reasonably fast and the effects of shocks on the cointegrating relations eventually vanish.
6. The eight plots that visually compare for each of the error correction equations the actual and fitted values also confirm this observation. These plots are available upon request.
7. Available online at www.joneslanglasalle.com/GRETI/en-gb/Documents/GRETI/docs/TransparencyIndex_2012.pdf.