Modelling time-varying relations in housing prices: a semiparametric panel approach

Abstract For the modelling of housing prices, housing-specific characteristics play a key role while external economic conditions are often omitted. We introduce a semi-parametric panel model with time-varying coefficients, incorporating housing-specific attributes, as well as both observed and unobserved external economic conditions. For a monthly data set spanning from 2006 to 2020 on 60 Dutch municipalities, our model accurately captures comovements in highly nonlinear housing price series, revealing significant variations in the relation over time. We find that housing characteristics and the external environment explain comparably high proportions of variation in the prices, which emphasizes that both factors should be included in empirical analyses.