Should the government promote marriages among the disadvantaged?

Perhaps the most socially divisive question in post welfare reform America has been whether the federal government should encourage unmarried parents to wed. The Bush administration plowed hundreds of millions of dollars into its Healthy Marriage Initiative, a program that the Obama administration has actually continued to fund (albeit with much less fanfare).

In a thought-provoking post, our friend Bill Gardner reviews some recent statements from social policy conservatives calling for a “re-moralizing” of social norms, and end to the permissive attitude toward single parents. The primary motivation is to improve child wellbeing, since children of single mothers do fare worse than children from two-parent families. Bill quotes Ron Haskins of Brookings, who points to evidence that, “Children from single-parent families have, on average, more developmental problems, including lower educational achievement, than children of married parents. This perpetuates poverty and lack of mobility into the next generation.”

Unfortunately, much of the evidence showing a positive effect of marriage on child outcomes is plagued by severe selection bias. All else equal, parents that are more likely to marry are also likely to have other qualities – better mental health, stronger social support, better labor market prospects – in ways that are not easily controlled. In a 2010 paper, Keith Finlay and David Neumark took an ingenious approach to the selection bias by using state-level incarceration rates as an instrumental variable for the availability of eligible male partners.

Although the set-up for this paper is relatively intuitive, some of the details deserve careful scrutiny, and I recommend you check it out for yourself. The main explanatory variable is the presence of an unmarried versus a married mother, and the outcome is high school dropout rates among teens ages 15 to 17. The logic is that state-level incarceration booms tend to remove prime-age male partners from the marriage market, and reduces the “quality” of partners by simultaneously reducing skills and increasing stigma, thus changing marriage rates.

During the study period of 1970 to 2000, incarceration rates skyrocketed. This was likely driven by many factors. The more such factors reflect exogenous changes in political shifts and unintended consequences of sentencing laws, the more plausibly the instrument meets the all-important exclusion restriction (i.e. incarceration rates should affect child outcomes only through its correlation with marriage rates). Arguably, however, incarceration rates may also respond to social problems that themselves either impact both marriage rates and child outcomes (such as the crack epidemic, the labor market, or a general rise in lawlessness). Although Finlay and Neumark control for these factors directly in IV regressions using annual state-level indicators, their residual effect is difficult to completely purge from regression models.

Interestingly, the data reveal that the time period from 1970 to 2000 is a period of both rapidly increasing single motherhood rates among blacks and Hispanics (an almost eight fold increase among black and tripling among Hispanics), but also very substantial declines in the high school dropout rates. The incarceration rates also skyrocketed for black and Hispanic males during this period in many states.

The key table from the paper is shown in two pieces below. In OLS regressions for high school dropout probability the association between having a single mother and being a dropout is positive for black and Hispanics, even with detailed controls (the OLS models). Once the incarceration rates (first the current rate, next the ten-year lagged rate) is used in IV regressions, the sign on the marital status coefficient actually reverses. That is, once we isolate variation in marriage rates that is driven by incarceration rates, we find that married children fare somewhat worse. In instrumental variables speak, this is the local average treatment effect (LATE), and it provides an estimate only for those populations that are likely to be impacted by the instrument. However, since the focus is on low-income minority children and their mothers, this is arguably a highly relevant sub-population.

Should we believe it?

This is the kind of study that needs to be approached with a heaping dose of caution. First, the use of incarceration rates as an instrument for marriage rates raises some pretty knotty questions about the causal relationships between men going to prison, fertility decisions (which they do try to control for), partner formation, and child development that are difficult to unknot. Children growing up in high incarceration environments are likely to experience stress and trauma that will impact their schooling outcomes – and the effects will likely spill-over to children with fathers in the home. Second, the (very promising) secular rise in school completion during this period raises questions about a large scale social transformation in attitudes toward school, and also might make us wonder how different the “marginal” high school dropout was in 1970 than in 2000.

The largest question is what relevance this might hold for policy. If policies naively try to increase marriage rates without changing the resources and parenting abilities, we might expect to see some unintended consequences (like the lingering presence of an abusive father or high levels of family stress). If, however, the intervention is to both encourage marriage and increase parental skills and couple skills, then this study cannot shed light on that question. It may be true that married parents with low-education contend with an array of stresses that undermine their chances of success, but that their outcomes could be dramatically improved with programs to increase their assets and to connect them with social supports.

To conclude, this study should cause marriage advocates to rethink the putative evidence showing that marriage is better that comes from cross-sectional regressions, but it does not by itself undermine the case for programs to build relationship skills and improve the financial resources of low-income couples considering marriage.

About Brendan Saloner

I am a postdoctoral fellow at the University of Pennsylvania in the Robert Wood Johnson Health and Society Scholars Program. I completed a PhD in health policy at Harvard in 2012. My current research focuses on children's health, public programs, racial/ethnic disparities, and mental health. I am also interested in justice and health care.
This entry was posted in Articles and tagged , , , , , . Bookmark the permalink.

3 Responses to Should the government promote marriages among the disadvantaged?

  1. Ben Baumberg says:

    Ah, this is what we need more of – instrumental variables models, with the assumptions turned into plain English, and subjected to critical scrutiny. Personally speaking, any economist that can’t talk about their IV models in this way shouldn’t be allowed to do them at all. (I say this as a non-economist currently trying IV models for the first time!)

  2. Thanks Ben. As IV papers go, this was a relatively clear one, but I have to say that sometimes I find myself needing to diagram these paper to understand which ways the causal arrows need to go.

  3. Pingback: Homemade colon cleanse products » Blog Archive » Full Yogic Breath With Retention ~ Pranayama The 3 Day Body And Colon Detox

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.