Working Paper: CEPR ID: DP4146
Authors: Thomas Piketty
Abstract: For given observable parental characteristics, children with divorced or separated parents tend to perform less well at school than children living with their two parents. This result has been used to argue that softening divorce legislation might be bad for children. This might, however, just reflect a selection effect: parents who decide to separate are presumably parents who fight with each other, etc., and it is unclear whether children growing up in a high-conflict, two-parent family are better off than children with separated parents. In this Paper, I develop two identification strategies suggesting that the selection hypothesis is indeed relevant. First, I look at the school performance of children a couple of years before their parents separate, and I show that they are doing as bad as children already living with only one of their parents. Next, I exploit the large increase in separation rates following the 1975 divorce law reform (as well as cross-regional variations in divorce rates) to show that the performance gap of single-parent children is a declining function of the separation rate, with an elasticity close to -1. Taken together, my results suggest that parental conflicts (rather than separation per se) are bad for children, and that the distribution of conflict intensity between couples has been fairly stable over time and was not significantly affected by the change in divorce law.
Keywords: divorce; educational performance
JEL Codes: I38
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Parental conflict (J12) | Children's school performance (I21) |
| Pre-separation conditions (Y20) | Children's school performance (I21) |
| Separation rate (J63) | Performance gap between single-parent and two-parent families (J12) |
| Single-parent family proportion (J12) | Children's school performance (I21) |