The Fiscal Costs of Marriage and Divorce

This morning’s inbox leads me to two observations:

1) There is some excellent research out there about marriage and divorce.
2) There is no shortage of ways for imaginative advocates to distort the findings of this research.

Let me begin with the first point: an intriguing paper by Elizabeth Ananat and Guy Michaels, forthcoming in the Journal of Human Resources.

A key problem with research linking divorce and income is that we can’t tell whether divorce causes lower income, or whether lower income (or its correlates) cause divorce. This observation leads these economists to analyze a set of divorces that are somewhat random.

Following an idea developed by Gordon Dahl and Enrico Moretti, they note that people have a greater tendency to divorce if their first-born was a girl. Reasoning that the gender of the child is random, they explore the consequences of this higher divorce risk on income.

A similar approach led Kelly Bedard and Olivier Deschenes to a rather striking conclusion: divorce may not be financially harmful to women. They find that divorce actually led women to live in households with greater income per person. (To be more precise, they argue that the marginal divorces caused by the gender of the first-born led women to live in higher income households.)

And Ananat and Michaels agree, finding similar effects. They then slice and dice this surprising finding, concluding that divorce may raise incomes on average, but it leads some women to lower incomes, and some to higher incomes. Those who gain tend to be in a (slight) minority, but they tend to gain more, which explains the rise in average income.

And now the second point.

A new report was released this morning by the Institute for American Values, the Institute for Marriage and Public Policy, the Georgia Family Council, and Families Northwest, using the Ananat-Michaels result to argue that family fragmentation costs the U.S. taxpayer $112 billion per year, or $1 trillion each decade.

How do these personal gains add up to a taxpayer loss? It is simple: in “advocacy science” one can pick and choose what to count. In this case they counted costs and ignored benefits.

The advocates suggest that the Ananat-Michaels results imply that if poor single women were to marry, around 60 percent of them would no longer be in poverty, leading to a decline in total poverty of about one-third. Thus, they attribute one-third of the cost of anti-poverty programs to the costs of “family fragmentation”. Citing research by Harry Holzer linking childhood poverty to crime, they also attribute one-third of the costs of poverty-related crime to divorce, and also attribute further tax costs to the intergenerational transmission of poverty. Add enough of this stuff up, and you hit over $100 billion per year.

What’s the trick here?

The Ananat-Michaels result is that divorce seems to help the finances of about as many women as it hurts, and those who gain, may gain more than those who lose. But this report counts up the costs to the taxpayer from the women who lose income, but refuses to count even a single dollar of the rise in taxes linked to those who gain income. Moreover, these winners are not only paying higher taxes — their kids are probably also committing less crimes, and they hope to transmit their higher economic status to their kids, who in turn will also pay more taxes. Moreover, the link between divorce and crime is not so obvious — as dissolving violent marriages reduces domestic violence.

Amazingly, the advocates put together “fiscal” costs of divorce without even understanding the tax code. The U.S. tax system is structured so that when poor single mothers marry men with higher incomes, in most cases, the total tax paid by husband and wife would fall. Yet this isn’t counted.

Those poor single women aren’t robbing us of tax revenue, they are actually paying more than if they were married! (Yes, the tax code does include a marriage penalty for some couples who are both high earners, but for most couples, the U.S. gives you a tax break for getting married.)

And finally, the advocates fundamentally miss what marriage is about.

Many of the gains from marriage that they count are gains mainly from forcing poor single women to live with others, thereby realizing economies of scale. If there is a fiscal case to be made for encouraging such behavior, the same fiscal case suggests we should encourage them to live with just about anyone — a same-sex lover, a polygamous family, or even with good friends. Yet for some reason, the advocates seem reluctant to extend their argument to its natural conclusion.

Perhaps the fact that many women are willing to face the prospect of poverty to get out of their marriages tells us that, beyond fiscal implications, there are other, more important, costs and benefits of marriage and divorce that also need to be counted.

I must admit that I find the Ananat-Michaels result surprising and interesting. And there remains a lot to be learned about the effects of divorce on the income of divorcees, or on its fiscal impact more generally.

But the first law of advocacy science coincides with a well-known economic principle: any cost-benefit analysis that only looks at one side of the ledger will always come to a reliable conclusion.

[Thanks to Betsey Stevenson for her insights on this post.]

Leave A Comment

Comments are moderated and generally will be posted if they are on-topic and not abusive.



View All Comments »
  1. Matt says:

    “Reasoning that the gender of the child is random, they explore the consequences of this higher divorce risk on income.”

    That may be a flawed assumption:

    Thumb up 0 Thumb down 0
  2. Austin says:

    How about the cost that is incurred from divorced when these men start over in other relationships and build new families? Robbing Peter to pay Paul doesn’t seem very efficient to me. But if you’re rich, this is a non-issue.

    Thumb up 0 Thumb down 0
  3. xyb says:


    Nicely said. Thank you for calling out all this craziness.

    Marriage and divorce are _not_ economic issues!

    Thumb up 0 Thumb down 0
  4. Mike says:

    Wow. I mean, wow. Picked up by MSNBC:

    Thumb up 0 Thumb down 0
  5. scott cunningham says:

    Matt – Isn’t the Annat-Michaels paper looking at births? Your article is probably more focused on survival. It is true that there is both a slightly higher probability of a male birth and a lower probability of male survival. But it would therefore make sense that you’d see child sex ratios look most like the sex ratios at birth among wealthier people, since those people can afford better neonatal and postnatal healthcare. I think the Annat-Michaels identification isn’t subject to this problem so long as they’re focused on the event of birth. So the dataset would need to have data on the gender of births that both lived and died.

    On the other hand, maybe you’re on to something because I’m not sure of the male fetus’s survival versus the female fetus’s survival in utero. If the male fetus has a lower survival rate gestationally, then I think your point is a good one. If the male has a lower gestational survival rate, then even sex ratios at birth will be correlated with income, and thus endogeneous to marriage types. Interesting observation.

    Thumb up 0 Thumb down 0
  6. scott cunningham says:

    xyz – I don’t think Wolfers was at all saying that marriage and divorce are not economic issues. What do you think economics is anyway? Look up Wolfers’ CV and you’ll see a load of papers on marriage and divorce. Economists have been studying marriage and divorce at least as far back as the early 1970s, formally, and probably even sooner.

    Thumb up 0 Thumb down 0
  7. Matt says:

    scott cunningham wrote:
    “Matt – Isn’t the Annat-Michaels paper looking at births? Your article is probably more focused on survival.”

    No, the article is talking about births, not survival. This phenomenon has been observed in a variety of animals as well.

    From the article:

    “We find that married, better educated and younger (advantageous from a biological perspective) mothers are more likely to bear sons.

    “For instance, relative to a mother with some college education, a mother without a high school degree was approximately 0.6% less likely to bear a boy.”

    Thumb up 0 Thumb down 0
  8. Mike says:

    They have economic aspects and implications. Get off your emotional, knee-jerk reaction wagon and think like a scientist. Yeah, divorce sucks for those involved in it. But don’t let that make you incapable of studying it scientifically.

    Thumb up 0 Thumb down 0