Identifying the determinants of housing prices in the City of Johannesburg for the year 2012: a spatial analysis approach

Abstract

An attempt has been made to identify the determinants of housing prices in the City of Johannesburg, Gauteng Province, South Africa. A dataset of 2012 housing prices at sub-place level and variables were categorised into housing demand and housing supply. The regressions; Ordinary Least Squares (OLS), spatial lag and error models and geographical detector technique were employed. The results obtained from Moran’s I test suggested that there is spatial dependence present in the data. From the properties that were captured in 2012, the maps indicated that the full title ownership type is the most favoured one in the municipality. The hot spot analyses suggested a significant clustering of housing prices around Sandton and Houghton area in the region B and the low-low spatial cluster was observed in region D, Soweto area and region G the Lenasia area. From spatial regression model, the results showed negative effects exerted by unemployment, number of sales, and renting whereas income, mortgages loans, city services, transfer duty and property ages are positively and significantly associated with the housing prices. The study analysed the direction and strength of association between housing prices and their potential determinants. The influence of the factors on housing prices were ranked by power of determinant (PD) value whereby transfer duty value was 0.7658 and household income was 0.6499. The geographical detector technique results revealed that transfer duty and household income exerted strong influence in relation to housing prices followed by renting and unemployment while city services shown to have weak positive influence with a value of 0.0911.