Rereading Tim Harford’s chapter three, this question runs through my mind. Numerous studies correlate divorce with subpar outcomes for the children, though Justin Wolfers once told me he was not convinced these studies had proved a causal relationship. Perhaps divorce-prone families have other dysfunctionalities which correlate with the kids having later problems in life and that the divorce is not causing those problems.
My wondering is more fundamental. Does a marginal (expected) increase in divorce increase or decrease the number of children who end up being born? On one hand the prospect of divorce may cause some people to limit the number of children they have. On the other hand, there is a surplus of women on the marriage market. Divorce, followed by male remarriage or at least siring, tends to increase the total number of children. I suspect this latter effect predominates. If divorce is unexpected, this latter effect almost certainly predominates.
If divorce causes more children to be brought into the world, it is hard for me to believe that divorce is bad for "the children" overall. It’s better to be born, at least for most kids. You might argue that "children existing now" have a special moral privilege over "children in the abstract." Sometimes, yes (we don’t value human life at replacement cost), but if we are asking "will divorce be good for children thirty years from now" currently they are all "children in the abstract."
I believe this defense of divorce is consistent with Tim’s overall take. Of course a person who fears overpopulation might see this as additional reason to oppose divorce or make it more costly.