Effects of short- and long-term unemployment on health satisfaction: evidence from German data

Abstract

The purpose of this study is to analyse the effects of short- and long-term unemployment on health satisfaction. The data source used for the analysis is the German Socio-Economic Panel (GSOEP) which, given its longitudinal structure, allows one to better overcome the problem of endogeneity. Three different models are used in order to assess the effect of short and long-term unemployment and reemployment on health satisfaction. The results show that short-term unemployment has only a significant (and negative) effect for men, while for women short-term unemployment does not have a significant effect on health satisfaction. Being unemployed for a long period has a significant and negative effect for both men and women. Finally, it can be also concluded from the empirical analysis that reemployment has a significant and positive effect on health satisfaction for both unemployed men and women, independent of how long individuals have been unemployed.

Notes

¹ Source: EUROSTAT.

² Source: LABORSTA (Labour statistics database operated by the ILO Bureau of Statistics).

³ Source: Bundesagentur für Arbeit.

⁴ Married or not married.

⁵ In GSOEP the relevant family changes considered are: wedding/marriage, moving in with partner or spouse, divorce, separation from partner or spouse, death of partner or spouse, birth of a child, child moves out and other family changes.

⁶ The labour status variable has undergone some changes since the beginning of the GSOEP. The categories Non-working, Unemployed and Working have been maintained but other new categories have been included. In 1984, apart from the three main categories, the following were also included: non-working: age 65 and older, non-working: in education/training, non-working: military–civil service. These were included in the non-working category in the present study. Later, new categories were introduced, such as non-working: maternity leave, non-working: but sometimes secondary job, non-working: but regular secondary job. These have been also included in the non-working category. Finally, in the most recent waves (in 1999 and 2000) information was introduced about the last week before the interview: non-working: but working in the past seven days, working: but not working in the past seven days. These last categories have been also included in the non-working category, since there is no regular activity in the labour market.

⁷ For empirical evidence of the effect of health on labour status see Wang (1997) and Wilson (2001).

⁸ In this descriptive analysis, there is a certain bias since individuals with extreme responses (0 and 10) are included and they can only experience changes in one direction. In the sample, individuals reporting 0 satisfaction with health represent 1.24% of all observations. Of all the individuals who answered 0, 31% reported no changes in the following year, and the rest reported positive changes. Individuals reporting health satisfaction of 10 represent 9.23% of the sample. Of these individuals, 41.5% reported no change in the following year, and the rest reported negative changes.

⁹ One year approximately.

¹⁰ This table compares changes in health satisfaction between t − 1 and t (for individuals who remain employed and for individuals who are less than one year unemployed) with changes in health satisfaction between t − 2 and t (for individuals who are more than one year unemployed and less than 2 years) and with changes in health satisfaction between t − 3 and t (for individuals who are more than 2 years unemployed). These three different variables may not be perfectly comparable. However this comparison is useful for descriptive purposes. In the multivariate analysis it will be considered the same period for all individuals included in the analysis.

¹¹ However, the percentage of negative changes in health satisfaction remains unchanged. As a result there is a reduction in the percentage of men reporting no changes in health satisfaction.

¹² One does not consider other kinds of transition in the models. Other transitions, like leaving the labour market, are considered as missing values. If an individual loses his job between t − 1 and t but finds a job again before t, it is also considered as a missing value.

¹³ S is not constant across individuals. For some, S will be closer to t − 1 and for others, closer to t.

¹⁴ Since health satisfaction is an ordinal variable with 11 categories (from 0 to 10), health satisfaction differences can take 21 different values.

¹⁵ At t − 1 all individuals were employed. Main earners were those individuals whose partners were unemployed or out of the labour force at t − 1.