Houses and/or jobs: ownership and the labour market in Belgian districts

In a number of papers A.J. Oswald (1996, 1997) argues that high rates of home ownership may imply inferior labour market outcomes. This paper tests the Oswald hypothesis in a panel of 42 Belgian districts since the 1970s. The use of data going back to 1970 allows us to embed the Oswald hypothesis in a broader model including other key determinants of employment like labour costs and productivity, the skill level of the population, and demography. Considering that ownership may be endogenous to (shocks in) employment, we use IV estimation methods. Overall, we find evidence in favour of the Oswald hypothesis. We observe that a 1 percentage point rise in the rate of home ownership in a district implies a statistically significant fall in the employment rate by about 0.3 percentage points. Our results underscore the importance of including other determinants of employment, of controlling for unobserved fixed regional and time effects, and of appropriately dealing with endogeneity. Disregarding these issues, as is often done in the macro labour literature, may imply very different estimation results. Additional estimation reveals that the size of the Oswald effect falls in the fraction of high skilled in a district.