A survival analysis approach to the duration of union membership in Italy

Abstract

Thanks to direct access to union databases, this article applies survival analysis to a sample of 47 637 Italian workers trying to explain the determinants of the duration of union membership. The results show that union membership duration is a positive, though declining, function of age. Furthermore, women, flexible workers, foreign ones and those working in cities tend to show less attachment to union membership than other workers. The estimated median duration is about 6 years. Positive hazard duration dependence is also found. Unobserved heterogeneity is detected but it does not significantly affect model estimates.

Acknowledgements

The author would like to thank Alfiero Boschiero and CGIL Veneto for the assistance in the data collection process. The usual disclaimer applies.

Notes

¹I also introduced time dummies depending on the year of quitting to capture possible business cycle effects. However, they were not significantly different to zero, possibly due to the fact that in recent years the two counties under study did not experience marked variations in unemployment rates.

²Vaona (2006) shows that the Weibull model fits the data better than the log-logistic one. The results for the log-logistic model are shown in Vaona (2006) and they are very similar to those contained in the present article. So results are even robust to model misspecification.