Testing the predictive ability of house price bubbles for macroeconomic performance: A meta-analytic approach

Abstract

This paper tests for the predictive ability of bubbles in housing markets on several proxies of macroeconomic performance using a panel of eighteen advanced countries. We use robust inference methods to address the bias resulting from the unknown persistence of our house price bubble measure. Evidence of predictability is analyzed by using a meta-analytic p-value combination approach for an overall joint significance, a method that is rarely applied in the panel predictive regression framework. The advantages are that heterogeneous panels are accommodated, and one can make inference on the individual unit for which the null hypothesis of no predictability is rejected. Our findings reveal the following: First, house price bubbles consistently predict an increase in government expenditures, even in the presence of structural change, different testing horizons and sample periods, as well as the inclusion of credit bubbles as as an additional predictor. Second, we find greater evidence that house price bubbles enhance macroeconomic performance in the identified countries for which evidence of predictability exists.