A continuous spatio-temporal model for house prices in the USA

We revisit the studies on the evolution of house prices in the USA using a spatio-temporal model estimated using a Bayesian method. This method introduces a new specification of an error correction model with random effects measured continuously in space. This model allows observing the deviations from the co-integration relationship in each analyzed location and a clearer interpretation of the house price dynamics between 1975 and 2011 for 381 metropolitan areas in the USA. The results indicate the presence of a housing price cycle, consistent with the patterns observed in the analyzed period.