Using German regional data for 1998, 2002 and 2006, this study examines the Oswald hypothesis, the conjecture that high levels of homeownership are linked to inferior outcomes in the labor market. Applying a set of control variables, three different econometric models are specified and estimated: a cross-sectional model, a pooled data model, and a model taking into account unobserved regional heterogeneity. Once unobserved regional effects are accounted for, the findings are consistent with Oswald's hypothesis. The economic significance of the relationship is at best marginal, however. © 2011 Springer-Verlag.