The links between property and the economy – evidence from the British and German markets

Abstract

This study supplies empirical evidence on the dynamic interactions between the property markets in Germany and the United Kingdom and their country‐specific macroeconomic environment. Using a VECM framework, the findings contribute to improving the evaluation of the properties’ behaviour by considering a wide range of macroeconomic risk factors. On a long‐term basis, we find remarkable similarities between both examined real estate markets with respect to significance, signs and magnitude of coefficients, despite essential differences in terms of market structure, conditions and performance. This suggests that the fundamental role of property markets in an economy dominates the country‐specific characteristics in the long run. However, the distinctive features of the national property markets, including differences with respect to the financial systems, are primarily relevant during the short‐term adjustment process back to the long‐term equilibrium.

Keywords:

• property markets
• macroeconomics
• cointegration
• vector error correction model

Notes

1. With a turnover of approximately 577.6 million, the British market was the largest European market in 2005. Germany was fourth after France and Italy with 136.9 million; see ZERP (2007).

2. For a discussion see Bond and Hwang (2007).

3. For further details see Maurer, Reiner and Sebastian (2004).

4. We use the time‐series released by the Deutsche Bundesbank for the German model, while the UK counterpart covers unemployed people over the age of 16.

5. All macroeconomic time‐series are taken from the Datastream database.

6. The industrial production indices for both countries and the consumer price index of Germany are available as seasonally adjusted time‐series. The UK consumer price index has been seasonally adjusted using the additive variant of the X12 procedure.

7. The test decisions are based on the critical values of MacKinnon (1991, 1996). The number of lags is determined in the framework of the ADF test with the aid of the Akaike information criterion (AIC), and the PP test is based on Newey and West (1994) bandwidth using Bartlett kernel.

8. Prior to this decision, it was necessary to conduct further analyses in order to preclude the possibility, that other reasons, such as, for instance, high values of correlation among the selected variables, were responsible for the significant deviations from the null hypothesis of the White test.

9. The estimated models are free of possible hazards caused by autocorrelation occurring within the residuals, too, although this is not explicitly mentioned in Table 3.

10. Both VEC models are additionally tested for stationarity by the Dickey–Fuller (DF) test using the critical values according to Banerjee, Dolado, Galbraith and Hendry (1993).

11. As the base rate variable displays significant coefficients within the short‐term matrix Σ ΓiΔYt‐i of the VECM equation (see Equation (3)), omitting the base rate variable does not improve the validity of the results in the case of the UK, even though we cannot detect a significant contribution within the cointegrating relationship π = α β′.