How causal is separation? Lessons learnt from endogenous switching regression models for single mothers' economic strain in Germany

Single mothers often experience precarious financial conditions. However, it is not fully understood to what extent separation is the cause of these conditions versus being their consequence. Estimating an endogenous switching regression model based on a sample of 626 separated and 5,525 non-separated mothers drawn from the German Socio-Economic Panel (SOEP) 1984-2018, we disentangle the roles of causation and selection for separated mothers' individual earnings as a measure of economic well-being. Our results indicate that separated mothers increase their working hours and sometimes adjust industry in anticipation of the separation event and afterwards. Adjusting for these processes that can be considered caused by the upcoming event, the positive selection into separation turns negative, while the non-separated are clearly positively selected. Thus, comparing average women with mean characteristics, the actually (non-)separated earn lower (higher) wages than women who are randomly assigned to a (non-)separation scenario. Additionally, the separated are more negatively selected into employment. Robustness checks largely confirm our results against changes in sample composition, eliminated group differences in period distribution, and model specification. Thus, our data support the notion that both chronic strain and crisis-caused factors diminish single mothers' economic well-being. Unobserved traits associated with lower labor market investments and productivity explain part of separated mothers' economic strain after separation.