Estimating Spillover Effects in the English Housing Market: A Heterogeneous Spatial Autoregressive Panel Approach

Abstract

This paper estimates a spatio-temporal model of house price changes in England, allowing for heterogeneity in spatial effects between districts and the separation of the spatial dimension from the impact of common factors. We model house price growth from January 1995 to August 2016 across 325 local authority districts. The renewed interest in the analysis of cross-sectional dependence has led to a range of novel techniques in modelling cross-dependent systems. Despite the considerable spatial literature concerning house prices, studies often fail to assess the level of cross-sectional dependency. Models incorrectly specified for spatial connections where factor dependence persists can engender spurious outcomes pertaining to the strength and prevalence of cross-sectional dependence. We endeavour to fill this gap in the literature by applying the novel methods developed by Aquaro et al. (2015) and Bailey et al. (2016) to the English housing market. In reference to the ripple effect hypothesis, our results indicate a more nuanced spatial impact than the theory implies, where areas with higher levels of economic activity find stronger spillover effects that are not unique to London but also districts surrounding cities such as Manchester and Nottingham. Furthermore, the variation in spatial dependence across districts demonstrate how delineating the housing market by regions fails to sufficiently allow for spatial heterogeneity.