Spatio-temporal Modelling of Irish House Prices

Abstract

Ireland, a prosperous nation in Western Europe, has experienced great instability in its property market in recent years. However, the nature, variance, and causative aspects of the recent increases in housing prices in the Republic of Ireland are poorly understood. Thus, this work uses statistical modelling techniques to define the temporal, geographical, and combination space-time patterns of housing prices in Ireland. Irish house price data have been collected for this study. First, several temporal models were used to analyse trends in Irish house prices, incorporating temporal components, and predictions. Among these models, the SARIMAX model was identified as the best fit, demonstrating the highest goodness of fit for Irish housing prices. The analysis revealed a decline in house prices from 2010 to 2013 but, followed by a steady increase, with some seasonal variations peaking in the third quarter. Second, a variety of geographically weighted regression (GWR) models with different distance kernel functions were applied to capture spatial effects and factors that influence house prices in Ireland. Diagnostic indicators showed that the GWR model with a bi-square kernel function and adaptive bandwidth was the most appropriate for fitting and analysing the spatial variability in house prices.The study demonstrated that the heterogeneity of spatial factors across regions or spatial units can be effectively captured by GWR models. Finally, a combined spatio-temporal modelling approach was investigated, simultaneously capturing temporal and spatial effects. Geographically and temporally weighted regression (GTWR) was used to account for both spatial and temporal correlations and dependencies. The study found that the most effective model for representing the fluctuations in the Irish housing market is the GTWR model, featuring an exponential kernel function and adaptive bandwidth. To evaluate the relative strengths of each approach, the study compared the performance of the temporal (SARIMAX), spatial (GWR), and spatio-temporal (GTWR) models. Although each model provided distinct insights, the comparison showed that the GTWR model performed better overall in capturing fluctuations in the Irish housing market, highlighting the significance of considering both spatial and temporal dimensions simultaneously. This research offers valuable insights for understanding the intricate variations in the Irish housing market. Specifically, it highlights the importance of considering both time trends and regional differences when analysing house prices. The findings could be useful for real estate professionals or anyone interested in understanding the Irish property market.