Abstract
In this paper we apply dynamic model averaging (DMA) to forecast Portuguese and Spanish house prices. DMA is a useful method for forecasting because it inherently allows for uncertainty in both the combination of predictors (model uncertainty), as well as in the marginal effect of each predictor (parameter uncertainty). In doing so we are able to track which predictors are relevant over the forecast period. Besides fundamental macroeconomic determinants to house prices dynamics we also include as predictors business and consumer confidence and financial markets volatility. We find that different predictors have varying inclusion probabilities for both Portugal and Spain. In Portugal, most predictors appear to have some value when it comes to forecasting changes in house prices, including volatility and consumer confidence. Furthermore, each predictor’s importance appears to increase over time. For Spain, most economic predictors appear to be useful for forecasting, and there appears to be less variation in each predictor’s importance over time. However, volatility measures appear to be less important in Spain than in Portugal for predicting house prices. (JEL: C22, C53, R31)
