This paper attempts to explain the repeated empirical finding that homeowners have shorter unemployment durations than tenants, even though Oswald's hypothesis predicts longer unemploy- ment durations for homeowners. The search models that have been proposed to motivate Oswald's thesis have difficulties in providing an explanation for the reverse of the Oswald effect. The model proposed in this paper is close to the ones proposed earlier, in that it also studies search behaviour, but contains a richer set of effects of homeownership on search behaviour. In our model, homeowners may have a higher intensity of job search (and hence shorter unemployment durations) when their housing expenses areöall other things being equalöhigher than those of tenants. Some studies have indeed found that the shorter unemployment durations occur especially among highly leveraged homeowners. We show that, in the Netherlands, many homeowners have higher housing costs than otherwise comparable tenants. The model developed in this paper is therefore able to explain the existing evidence of shorter unemployment durations for Dutch homeowners. © 2009 Pion Ltd and its Licensors.