Could this be genetic? you ask. People who have impulse-control problems might be more likely to divorce and pass those traits on to their kids. Partially, sure. But two evidence points argue against genetic determinism. First, similar, although less severe, patterns show up in the case of kids who lose one parent, which is mostly not going to be due to homicide. And second, if this is genetic, how come it has changed over time? Have we all gotten genetically less able to stay out of jail or sustain a long-term marriage?
We know that children of single-parent households have worse outcomes than children of two-parent households. To simplify, let us say that there are favorable family patterns and unfavorable family patterns.
First question: how much of this is causal?
It could be that an inability to do well on the marshmallow test causes you to be less likely to raise children in a favorable family pattern and also more likely to pass on to your children genes that cause them to be unable to do well on the marshmallow test. That is how genetics could account for the relationship between family patterns and child outcomes.
Megan asks, what has changed over time? It could be two things. First, nowadays it may be that you have to be much better at the marshmallow test to sustain a favorable family pattern. Second, it may that we have gone through two or three generations of increasingly assortive mating.
Until 1965, a man who was in the top third on the marshmallow test might very well have been married to a woman in the bottom third, and conversely. For one thing, the top third and the bottom third were not that far apart. For another, the signals of being able to do well on the marshmallow test were not as clear (college education was too rare to be a reliable signal, particularly among women). Finally, men and women cared more about separate respective roles (breadwinner and homemaker) than about common abilities in the marshmallow test.
But in the 1960s that began to change. So you get one generation of assortive mating, and for the children of these marriages the difference between the top third and the bottom third on the marshmallow test starts to widen. Then they grow up, engage in assortive mating, have children, and difference widens once more. And so on.
But suppose we assume that there is a strong causal relationship between bad family patterns and bad outcomes. That leads to our
Second question: what can policy makers do to improve family patterns?
If anti-poverty programs are the solution, then why has the problem been getting worse? The Center on Budget and Policy Priorities (pointer from Mark Thoma) will tell you that anti-poverty programs are working to keep people out of poverty. So why are we not seeing more family stability? (Ross Douthat makes related points. Pointer from Tyler Cowen.)
Of course, there is a hypothesis, going back to Moynihan’s analysis, that anti-poverty programs are the problem, rather than the solution, because on the margin they reduce incentives to marry. I am skeptical about that, but as you know I am all for replacing current means-tested programs with a universal benefit that has a low implicit marginal tax rate. The idea is to reduce the adverse incentives that presently exist.
Megan, like Charles Murray, would like to see elites proclaim the benefits of good family patterns. I am skeptical of that, also.
My guess is that family patterns are not amenable to public policy interventions.