Do pro-marriage policies work? The August issue of Demography includes an analysis of that question, and it’s fascinating reading in part because the three Austrian economists who wrote it defined “working” more broadly than I think American authors would have. In Austria, where the total fertility rate currently stands at 1.43 children per woman, part of successful marriage promotion is the production of healthy children. A shortage of healthy children creates much difficulty with supporting the elderly. Those of us in the U.S. understand this—even people who think that Social Security got into trouble mainly because of increased life expectancy know that the size of the workforce is a huge component of the worker-to-retiree ratio, which decreased from over 40 in 1945 to less than 3 today.

But the Austrians might understand these dynamics better, for their lower fertility means that their elderly dependency ratio is even higher than ours. Austria has been contending with slow labor force growth for decades due to below-replacement levels of fertility since the early 1970s. The early 1970s are also when the Austrian government switched from promoting marriage through tax deductions for household furnishings for newlyweds to a straightforward cash incentive: 7,500 Austrian schillings given to every citizen entering marriage for the first time. That amounts to about $5680 per couple (in 2010 dollars) if neither partner had been married before. In other words, the government increased its marriage promotion efforts when the fertility rate dipped below replacement.

With this backdrop, the scope of Wolfgang Frimmel, Martin Halla, and Rudolf Winter-Ebmer’s inquiry into whether marriage promotion works make perfect sense. Americans might stop after analyzing whether marriage promotion increased the number of successful (i.e., enduring) marriages, but these three Austrian economists went on to analyze whether it promoted marriages that produced children and whether the children were healthy at birth. Good questions. But just how does one go about answering them?

Their work includes much technical detail, but the key methodological point is really quite simple. In late August of 1987, the Austrian Minister of Finance quite unexpectedly announced the suspension of the marriage subsidy that had been in place since 1972: couples marrying before the end of 1987 would receive it, but there would be no subsidy thereafter. Not surprisingly, the marriage rate spiked in late 1987. The couples who married during that time are an unknown combination of those who would have married then anyway, those who rushed their planned weddings to capture the subsidy before it went away, and those who would not have married in the near future were it not for the policy change. But the authors make this unknown combination somewhat known by using the marriage dearth in early 1988 to estimate the number of accelerated marriages, and then attributing the rest of the excess marriages in late 1987 to “marginal marriages,” i.e., those that would not have occurred without the policy change.

It takes expected or actual relationship stability to promote investing in children. Do marginal marriages have that?

Just what are marginal marriages? The authors define them as matches that became acceptable only because of the increased cost of a longer search. “Consequently, marginal marriages should be of lower match quality than average marriages, whose match quality would be sufficient even without state intervention.”

They then analyze the stability of marginal marriages not just because “the benefits of marriage require a certain level of marital stability to materialize,” but also because “expected or actual stability is a prerequisite for marital investment.” Read: it takes expected or actual stability to promote investing in children—and do marginal marriages have that?

The answer is mixed. They find that marginal marriages are—surprisingly—just as stable as average marriages, but they don’t produce as well as average marriages: they resulted in 17% fewer children (among those marrying for the first time), and those children were more likely to have a low birthweight (less than 5.5 lbs), which is associated with a host of health complications. Anyone concerned about the production of a sizable and healthy labor force or simply concerned for child welfare can easily see that getting couples married and keeping them married does not make marginal marriages equal to average marriages. So it would be very easy to conclude my comments here in much the same way as I did when describing Christina Gibson-Davis’s work, emphasizing that marriage promotion in the U.S. has an uphill battle to wage.

Instead, I offer a critique that might allow us to be a little more optimistic. Frimmel and his colleagues compared the fertility and child health outcomes from marginal marriages to average marriages and found them wanting—but is that the right comparison? In Austria, more than half of reproductive-aged women are unmarried, but these women account for only 39% of all births. Back-of-the-envelope calculations confirm that the fertility of single women is more than 17% below that of married women; therefore, despite the fertility “deficit” of marginal marriages, they still result in higher fertility than if the marriages had not occurred. This would be a problem if those additional children were more likely to become children of divorce, but they are not.

With respect to infant birthweight, Frimmel’s other recent work is instructive: women marrying during pregnancy in Austria had heavier babies than those who remained single. Pregnant brides can be expected to have lower match quality than other brides (in Frimmel’s terminology, pregnancy increases the cost of a longer search), but marriage still increased birthweight. So even though those in marginal marriages are more likely to have low-birthweight children than those in average marriages, there is reason to believe the odds of healthy birthweight are improved by marginal marriage.

The outcomes in marriages promoted by state policy do not have to match those in other marriages to represent an improvement.