Evaluating Alternative Methods of Forecasting House Prices: A Post-Crisis Reassessment

This paper compares the performance of different forecasting models of California house prices. Multivariate, theory-driven models are able to outperform a theoretical time series models across a battery of forecast comparison measures. Error correction models were best able to predict the turning point in the housing market, whereas univariate models were not. Similarly, even after the turning point occurred, error correction models were still able to outperform univariate models based on MSFE, bias, and forecast encompassing statistics and tests. These results highlight the importance of incorporating theoretical economic relationships into empirical forecasting models.