UK Housing Bubbles and Contagion: Identification under Cross-Sectional Dependence and Spatial Heterogeneity

Recent trends in econometrics have led to a proliferation of statistical tools that lend themselves to data-driven evaluations of the housing market (Phillips et al., 2011, 2015; Bailey et al., 2016). This thesis evaluates the suitability of existing statistical procedures in the identification of explosive unit root processes, alongside the analysis of price diffusion in a time and cross-sectionally dependent system. Having date-stamped three regional housing bubbles consistent with the historical narrative, the predictive ability of macroeconomic and financial variables in bubble formation is estimated. Implementing the multi-step procedure proposed by Bailey et al. (2016), we estimate a heterogeneous dynamic spatial autoregressive model of the English housing market. The spatial parameters yield unexpected results in contradiction to prevailing economic theories of spatial dependence that remain unaddressed in the literature. To this end, we derive a unifying framework that captures the endemically heterogeneous characteristics of house price spillovers with joint treatment of common factors without loss of generality. The spatio-temporal autoregressive model with factors presents a parsimonious representation of house price diffusion with directional analysis of spillovers and identification of dominant units in the network. The derived spatial and system-wide diffusion multipliers provide meaningful insights into how a perturbation in neighbourhood house price inflation impacts a given district over a specified time horizon. The results reflect the London-centric ripple effect as a dominant factor while core-periphery spillovers dominate in urbanised areas. This thesis contributes a salient evaluation of housing dynamics and network effects consistent with theories of rational bubbles, new economic geography and spatial dependence.