Understanding rental prices in the UK: a comparative application of spatial modelling approaches

Abstract

Purpose – Accounting for locational effects in determining price is of fundamental importance. The demise of the mainstream property market has culminated in increasing appetite and investment activity within the private rental sector. The primary purpose of this paper aims to analyse the local variation and spatial heterogeneity in residential rental prices in a large urban market in the UK using various geo-statistical approaches. Design/methodology/approach – Applying achieved price data derived from a leading internet-based rental agency for Belfast Northern Ireland is analysed in a number of spatially based modelling frameworks encompassing more traditional approaches such as hedonic regressive models to more complex spatial filtering methods to estimate rental values as a function of the properties implicit characteristics and spatial measures. Findings – The principal findings show the efficacy of the geographically weighted regression (GWR) technique as it provides increased accuracy in predicting marginal price estimates relative to other spatial techniques. The results reveal complex spatial non-stationarity across the Belfast metropole emphasizing the premise of location in determining and understanding rental market performance. A key finding emanating from the research is that the high level of segmentation across localised pockets of the Belfast market, as a consequence of socio-political conflict and ethno-religious territoriality segregation, requires further analytical insight and model specification in order to understand the exogenous spatial and societal effects/implications for rental value. Originality/value – This study is one of only a few investigations of spatial residential rent price variation applying the GWR methodology, spatial filtering and other spatial techniques within the confines of a UK housing market. In the context of residential rent prices, the research highlights that a soft segmentation modelling approaches are essential for understanding rental gradients in a polarised ethnocratic city.

A mass-market appraisal of the English housing rental market using a diverse range of modelling techniques

“…higher limits for larger, more desirable, and expensive properties. Other studies arbitrarily trim the top and bottom 1% [15] or trim those that are more than 3 standard deviations beyond the mean [30].…”

“…Some studies recognise that the rental value can be positively skewed and for modelling purposes apply a logarithmic or square root transformation prior to analysis [16, 22–26]. However, others prefer to retain the original scale of the rental price variable [27–30].…”

“…This volume of data presents some challenges and opportunities. Recognising that such data are not always validated, prior to analysis, these data are sometime cleaned, for example by ignoring rental prices outside some range [27], trimming the top and bottom 1% of data [15] or removing statistical outliers [30].…”

“…the Coefficient of Dispersion (COD) or Price-Related Differential (PRD) [31]. If obtained from an administrative or commercial source, the volume of data also permits the estimation and prediction to be carried out on distinct sets of data [28, 30].…”

“…There is considerable debate about which, if any, approach is best. Some argue that the inclusion of spatially varying attributes in the MLR can stand instead of an explicit spatial structure in the analysis (see Bourassa, Cantoni [35] in the context of house sales and the findings of McCord, Davis [30] in the Belfast rental market). In most cases however SEM and SAR models are only found to provide modest improvements in model diagnostics.…”