Child Well-being

From RAND, a new working paper finds that divorce might not so bad on kids after all.

Social scientists and commentators disagree on how much of the association between parental divorce and child well-being is causal. This paper reexamines the claim that parental divorce is detrimental to children’s emotional well-being, measured in terms of behavior problems. The author analyzed panel data from the 1986-2002 waves of the National Longitudinal Survey of Youth, 1979, and found that parental divorce is associated with a higher level of behavior problems in children. However, after controlling for unobserved factors that are either constant over time or change at a constant rate over time, the effect of parents’ divorce substantially declines and its influence on their children’s emotional well-being is not statistically significant.

Via the Sloan Work and Family Research Network Blog.

On Thanksgiving, USA Today ran a story with the headline "Neediest Kids Live in Rich States." The story, based on a Kids Count Report to be released this week, opens with this lead:

Low-income children who fare the worst in health care, education and family structure live in some of the nation's wealthiest states, including Massachusetts, Rhode Island, New York, New Jersey, Maryland and Delaware, a study to be released next week reveals.

....

States where low-income kids fared best are clustered in the Great Plains and Rocky Mountain regions. The top five: Utah, North Dakota, Idaho, Wyoming and South Dakota.

The frame and lead used by the reporter implies that there is a strong relationship between overall well-being of low-income children and the share of children in low-income households by state, and that the relationship is a counter-intuitive one, i.e., low-income children do worse in states where there are relatively few of them.

Based on the information actually provided in the story, however, there isn't much evidence of an actual relationship, positive or negative, between these two factors. Using the state-by-state data in the story, I put together the graph below. Each point charts an individual state's rank on the low-income child well-being index and its share of low-income children. If low-income children did better in states where there were lots of them, the points would cluster around a downward sloping line. But what the chart shows looks like a fairly random distribution of dots.

What's missing from the story (and possibly the study it's based on) is any consideration that inequality, rather than absolute income levels, is what's driving state-level differences in child well-being. As I blogged about recently, a new study by Kate Pickett and Richard Wilkinson finds that inequality does a better job of explaining differences in child well-being than absolute income level:

Income inequality at the state level was significantly correlated with rates of teenage births, juvenile homicides, infant mortality, low birth weight, child overweight, mental health problems, and high school dropouts as well as with worse educational scores. States with higher average incomes had significantly fewer teenage births and fewer children dropping out of high school, but they did no better than poorer states on the other six measures of child wellbeing.

The study the USA story story is based on hasn't been released yet, so perhaps it makes this connection. But my guess is that it doesn't. The importance of relative differences in income has yet to be taken seriously by most anti-poverty researchers and advocates in the United States.

In a recent paper, economist Shirley Liu of the University of Miami investigates the association found by various researchers between parental divorce and children's educational attainments is a causal one, and finds no evidence that divorce per se is the factor that causes reduced educational attainment:

.... Using annual data on a representative sample of children and their families since childbirth until the completion of their schooling, we model child's schooling attainments and parents' marital dissolution as joint decision processes, where unobservables triggering divorce and affecting child development are assumed to be correlated. After accounting for both observed and unobserved characteristics, we find no evidence that the lower educational attainments is attributable to parents' divorce per se, but rather due to unobserved factors that are correlated with divorce. This result is robust to a more general definition of relationship dissolution, by treating spousal separation and divorce both as family "disruptions".

While further research is into the mechanisms through which parents' divorce affects child attainments is needed, our findings provide some evidence that divorce in itself does not negatively affect children's educational outcomes. The reasons for dissolving a marriage tend to be family-specific and usually observed only by the parties involved. Assuming that parents are altruistic and internalize potential gains (and costs) of dissolving their marriage on their children when deciding to divorce, on the margin there may be benefits to divorce (at least for some). If our goal is to help children from divorced families to become self-reliant adults, policy-makers need to be mindful that divorce in itself may be a consequence, rather than a culprit, of the same underlying mechanism that causes the inferior child outcomes. Simply restricting access to divorce without consideration of individual needs and circumstances can be irresponsible; it may also risk jeopardizing the welfare of many families and their children.

In a related paper published last year, Liu and Frank Heiland found that among children born to cohabiting biological parents, those children whose parents marry within a year after childbirth do not display significantly better outcomes than children of parents who continue to cohabit.

Since marriage promotion popped up on the policy agenda in the early years of the Bush administration, there's been a somewhat inexplicable consensus inside the beltway that, certeris paribus, marriage has positive effects on child well-being. Leaping past the question of causation, the only question that has seemed open for debate in policy circles is what kind of "marriage promotion" programs would increase marriage rates and child well-being. (As an example of the Washington Consensus on marriage, see this article). It's time to rethink that consensus. Based on the evidence we have to date, it's way too early to reach any consensus for policy purposes that the marriage/well-being link is a causal one.