The housing investment sector is crucial to the economy, yet there is no commonly agreed theoretical framework to model and forecast it. This paper proposes a model averaging approach to forecast housing investment, estimating many VECMs using a wide set of short and long-run determinants. The study provides guidance to forecasters and modellers in four ways: a pseudo-out-of-sample forecast exercise shows that their model averaging approach significantly beats a battery of benchmark models, they establish the most promising long-run determinants of housing investment, find substantial cross-country diversity as concerns the significance of mortgage interest rates as a short-run determinant, and suggest that the restriction of Tobin´s Q and income to one can substantially help in housing investment equations.
The paper suggests several practical policy implications. Country heterogeneity in the drivers of housing investment makes a strong call for country-specific policy measures, and common policy measures at the European level should be expected to result in different country impacts. Credit and housing affordability matter for housing investment in the long run, and policy measures that support housing affordability are expected to encourage housing investment. Improvements in forecasting and modelling housing investment might be helpful in mitigating booms and busts in housing investment, useful for regulators and macro prudential policymakers. Better modelling housing investment might also be useful for investors and construction builders as it might avoid unnecessary losses and bankruptcies. Finally, better forecasting and modelling housing investment might be helpful for avoiding possible crowding-out effects with other business investment and between private and public investment, especially in the current global context characterized by a green transition and monetary policy tightening.