"Ripple" effects in South African house prices

Abstract

This paper analyses the ‘ripple’ effect of house prices in large-, medium- and smallsized houses of five major metropolitan areas of South Africa—namely, Cape Town, Durban Unicity, Greater Johannesburg, Port Elizabeth/Uitenhage and Pretoria— based on available quarterly data covering the period of 1966:Q1 to 2010:Q1. Following the extant literature, the issue is contextualised as a unit root problem, with one expecting the ratios of metropolitan house price to national house price to exhibit stationarity to an underlying trend value, if there is diffusion in house prices. Using Bayesian and non-linear unit root tests, besides the standard linear tests of stationarity with and without structural break, overwhelming support is found for the existence of robust ripple effects. Also factor analysis conducted suggested that ripple effects originate in Cape Town for the large housing segment and in Durban for the medium- and small-sized houses.