The Effect of Union Status at First Childbirth on Union Stability: Evidence from Eastern and Western Germany

Abstract

It is often assumed that cohabitation is much less stable than marriage. If cohabitation becomes more common among parents, children may be increasingly exposed to separation. However, little is known about how the proportion of cohabiting parents relates to their separation behavior. Higher shares of childbearing within cohabitation might reduce the proportion of negatively selected couples among cohabiting parents, which could in turn improve their union stability. This study focuses on parents who were cohabiting when they had their first child. It compares their union stability within a context in which they represent the majority or the minority. The German case is well-suited to this research goal because non-marital childbearing is common in eastern Germany (60 %) but not in western Germany (27 %). The data came from the German Family Panel (pairfam), and include 1,844 married and cohabiting mothers born in 1971-1973 and 1981-1983. The empirical results suggest that the union stability of cohabiting mothers is positively related to their prevalence: survival curves showed that eastern German cohabiting mothers had a greater degree of union stability than their western German counterparts. This difference increased in the event-history model, which accounted for the particular composition of eastern German society, including the relatively low level of religious affiliation among the population. Controlling for unobserved heterogeneity did not change this result. In sum, these findings indicate that context plays an important role in the union stability of cohabiting parents.

Keywords: Cohabitation; German Family Panel; Marriage; Non-marital parenthood; Selectivity; Separation; Union stability.

Fig. 1

Results of Kaplan–Meier estimates (proportion of women who remained partnered with the child’s father 10 years after having their first child, by union form at the time they had their first child and region. Sources pairfam/DemoDiff (2008–2011), own estimates Weighted by sample design weight (including corrections for birth cohort and place of residence). Results of the Cox test (modified log-rank test) for equality of the survival curves of eastern and western German women: no statistically significant differences between cohabiting women (Pr > χ ² = 0.30); no statistically significant differences between married women (Pr > χ ² = 0.40)

Fig. 2

Results of an interaction of region and union form at the time the first child was born within the hazard Models 1b–5b, shown in relative risks with 95 % confidence intervals and significance levels. Sources pairfam/DemoDiff (2008–2011), own estimates. Significance levels ***p < 0.01; **p < 0.05; *p < 0.10. E.G. Eastern German women, W.G. Western German women, Cohab. cohabiting at first childbirth, Marr. married at first childbirth

Fig. 3

Robustness checks; results of the interaction of the region and the union form at the time the first child was born, shown in relative risks with 95 % confidence intervals and significance levels, the results of residual terms are shown in beta coefficients. Sources pairfam/DemoDiff (2008–2011), own estimates. Significance levels ***p < 0.01; **p < 0.05; *p < 0.10. E.G. Eastern German women, W.G. Western German women, Cohab. cohabiting at first childbirth, Marr. married at first childbirth. Notes All models based on model 5b. Robustness check #1: Model 5b without controlling for religious affiliation; Robustness check #2: Model 5b without controlling for union duration prior to childbirth; Robustness check #3: Model 5b, without coefficients that were insignificant; Robustness check #4: Model 5b, residual variance of hazard model fixed to 0.8; Robustness check #5: Model 5b, residual variance of hazard model fixed to 1.2