Linear and nonlinear causality in the UK housing market: a regional approach

Abstract

We examine the causal relationship between 12 UK regional house prices. Our data span from 1983:Q1 to 2012:Q4. The causal linkages both for the first differences and the levels are examined via Granger causality. The former allows us to examine short-run predictability while the latter the long-run. We relax the assumption of linearity and examine nonlinear causality both for the levels and the first differences. We find that long-run causality among the regions is mainly linear while in the short-run is nonlinear. London’s effect on the other regions is found to be mainly nonlinear in the short-run.