Price responsiveness of housing investments across major European economies

This paper contributes to the cross-country empirical literature on residential investments in the six largest economies of the EU, Germany, France, Italy, Spain, the Netherlands and the United Kingdom. The six economies account for over 80% of EU GDP and they are thought to have rather different housing investment reactions, in particular, for what regards the elasticity with respect to house and land prices. Despite the potentially key role of housing investments for the rest of the economy, there is a relative lack of empirical work in this field (DiPasquale, 1999). The focus is on the short-to-medium term dynamics of residential investment that are often obscured in the existing empirical literature in Europe and that may be important to the build up of house price bubbles and business cycles.

 An empirical study of shorter term dynamics presupposes a reasonable characterisation of the long term residential investment. We propose an error correction approach based on a parsimonious set of long-term fundamental variables without imposing a common error structure across markets. The set of variables included in the long-run equation comprise real private residential investments (RHI), real gross domestic product (GDP), real long term interest rate (INT) and a proxy for land prices (LAND) while we also consider other additional variables such as costs of construction (COST) and house prices (RPP). The dataset has been constructed trying to cover the longest possible time horizon. We have collected time series staring from 1970 at a quarterly frequency for all countries and variables. We employ a common modelling structure based on an error correction approach and country specific models. Co-integration techniques are applied in this study following Engel-Granger. A cointegrating regression based on dynamic OLS (DOLS), as suggested in Stock and Watson (1993), is estimated in the first stage. We provide an extensive set of cointegration tests based on Phillips (1987), Phillips and Ouliaris (1990) and Mackinnon (1990, 2010). Additional tests are also based on Pedroni (1999, 2004) and Johansen (1991) tests. The second step of the empirical analysis involves the estimation of the short-run dynamics where the error correction component is embedded into a dynamic model estimated using OLS and TLSL and panel methods.

 First, cointegration among the parsimoniously specified set of fundamental variables is detected in all countries employing a wide set of tests. Second, cross country differences are found in the short- and long-run responsiveness of private residential investments to real prices and to other relevant drivers. Germany has the highest elasticity to a proxy for asset values. Spain, Italy and the Netherlands show the lowest responses to asset price both in long- and short-run. Additionally a simulation exercise has been conducted. Spain and Italy are the less responsive countries to common small permanent shocks to real asset values.