The importance of housing tenure on individuals’ behaviour is widely recognised and studied. Arguably, one aspect that is crucial, with wide-ranging implications for macroeconomic fluctuations and possibly for economic inequality, is the effect of housing on the portfolio composition of households, and more specifically how the fact of being a homeowner, as compared to a renter or a mortgagor, affects portfolio efficiency. We first explore this question theoretically, using a mean-variance framework adjusted for the presence of housing in the portfolio. We then turn to UK data from the Wealth and Assets Survey to confirm empirically the predictions of the model: the five waves of the survey allow to break down households' portfolios in detail, as well as to account for a wide range of covariates and for individual unobserved heterogeneity.