This paper investigates the persistence over time of living in a jobless household, aiming to
disentangle the roles of state dependence and unobserved heterogeneity. In addition, the
potential heterogeneity of state dependence is examined through estimation of interaction
terms with the lagged household joblessness variable. Finally, the robustness of results is
explored through the use of alternative definitions of household joblessness each based on
different variables available in our data. Using the two definitions that are most different, we
find substantial state dependence which is larger for women than for men under both
definitions. That is, being in a jobless household in the previous year increases the probability
of currently living in a jobless household by 7.7 to 17.2 percentage points for men and 12.7 to
25.1 percentage points for women. Although state dependence clearly is an important factor,
as are a number of observed characteristics, unobserved heterogeneity also plays an important
role for men and women: 32 to 40 per cent of the unexplained variance can be attributed to
unobserved heterogeneity for men, and for women this is 42 to 46 per cent. A few
characteristics (age, disability, student status, living outside of major cities, having a
university degree, presence of preschool children) seem to affect the level of state dependence
to some extent. However, aside from the age effect, which can increase state dependence by
up to 50 per cent for men aged 60 to 64, the level of state dependence seems fairly
homogenous amongst men and amongst women.