Prices on the second-hand market for Swedish family houses: correlation, causation and determinants

The structural differences and the dynamics in prices on the second-hand market for family houses in large (Stockholm, Gothenburg and Malmo), medium-sized, small and industrial cities and sparsely populated areas are analysed in this paper. The basic house price data set used in the analysis consists of constant quality monthly price indices. The sample starts in January 1981 and ends in July 1997. The real price changes in house prices for all seven regions display a high degree of autocorrelation, and the correlograms reveal a mean reverting pattern. The Granger causality test indicates that the real price changes for the Stockholm area 'Granger cause' the price changes in the other areas. Thus, the real price change in the Stockholm area has a ripple effect on the six other areas. Both bivariate and multivariate Granger tests indicate information content in a number of macroeconomic variables versus the real price changes for the Stockholm area and the entire country. A simple VAR model was estimated with the price changes for family houses in the Stockholm area, a proxy for consumption growth and the change in the rate in the unemployment rate as endogenous variables and a number of exogenous macro variables. Experiments with impulse response functions show that a shock in the change in the rate of unemployment has a strong effect on real house prices and consumption.