Housing Yields

We build a granular dataset of residential property yields using rental and sale listings from a major German real estate platform. Equipped with more than 1.5 million property-level rent-to-price ratios, we document a novel heterogeneity puzzle. About one-third of dispersion in yields can be explained neither by an extensive array of property-specific observable features, nor by accounting for any possible below-zip code-level time-varying factor through a rich fixed effects structure. Unexplained yield heterogeneity is sizable and economically significant. Whereas property yields predict returns and rent growth rates, we show that their time-series variation largely originates at a highly local level. Our evidence may point to the importance of heterogeneity in investors' beliefs and preferences, as opposed to a battery of alternative explanations for which we directly test.