The housing wealth-to-income ratio has been increasing in most developed economies since the 1950s. In this paper we provide a novel theory to explain this long-term pattern. We show analytically that house prices grow in the long term if i) the housing sector is more land-intensive than the non-housing sector, or ii) technological progress in the construction sector is weaker than in the non-housing sector. The model is calibrated separately to the US, UK, France, and Germany. Quantitatively, both channels are equally relevant for rising house prices. We then study the dynamics in the housing wealth-to-income ratio by computing transitions. The model replicates on average 89 percent of the observed increase in the housing wealth-to-income ratio. The key for replicating the data is the differentiation between residential land as a non-reproducible factor and residential structure as a reproducible factor.