Dynamic interactions between health and employment statuses : a nonparametric analysis

Despite numerous sociological results, there is few econometric evidence on the causal links between health condition and job status. It is important to investigate the stability of these causal links during one's professional life. Papers that treat causal links between health and job statuses, make the assumption that causal links are identical over time. This could lead to a weak assessment of the causal effects. In this paper, we use a non-parametric approach, the Kullback causality measure, to test for causal links among time periods as well as global causal links. Our approach is more robust than the ones available and allows the determination of the effects of individual characteristics on causal links. We find significant reciprocal causal links between health condition (regardless of disease severity) and job status. However, job status does not cause both illness with large disability index and illness with large risk of death. These findings confirm evidence from the literature. However, analyzing the dynamic of the evolution of causal links between job status health condition regardless of severity allows us to conclude that job status only causes health between the 11th and the 17th year of professional life while only at the same period, health condition does not cause job status.