Print Post
  • New research on the causal effects of the timing of wage changes and marriage offer nothing definitive. Tweet This
  • Marriage is a social institution with a set of normative expectations that can transform behavior. Tweet This

Academics have devised a number of different ways to try to figure out whether any of married men’s earning advantage—the “marriage wage premium,” as it is often called—actually comes from being married. The latest work addressing this question concludes that “Marriage is useless (for wages).” At least, that is how Alexandra Killewald and Ian Lundberg title the summary of their work for an academic news magazine, but their conclusions in their peer-reviewed article are not as dramatic. In that article, they write:

Although we cannot provide definitive evidence that marriage does not benefit men’s wages, our analyses…are entirely consistent with an alternative explanation: men marry during a transition to adulthood when their wages are already improving, and they tend to divorce after their wages are already declining.

Previous work has also found that among men who were going to marry, the marriage wage premium kicked in before the wedding. At first blush, a pre-marital “ramp-up” in men’s earnings seems like pretty good evidence that marriage doesn’t cause the marital wage premium, but, because relationships do not develop instantaneously, anticipating marriage could improve men’s work behavior ( see, for example, Cheng 2016). Thus, the jury was still out on whether this pre-marital ramp-up in men’s earnings contributed anything to understanding whether marriage helps men become better earners.

Killewald and Lundberg dissected the ramp-up in two new ways. First, they compared it among men marrying at various ages. Among men first marrying after age 26, the ramp-up was absent. In other words, it was a characteristic only of men marrying at younger ages. They conclude that the most parsimonious explanation is that the transition to adulthood is a common cause of both marriage and wage growth. If anticipation of marriage caused wage growth, the ramp-up should occur regardless of age at marriage. But among those who had already experienced the transition to adulthood, there was no above-average growth around the time of marriage.

Note, however, that they did not compare those marrying after age 26 with those who were unmarried at the same ages and conclude that there was no marital wage premium. An absence of above-average wage growth is not the same as the absence of a marital wage premium—their analysis was all about the timing of wage growth. Killewald and Lundberg show that individual wage trajectories did not improve significantly around the time of marriage for those marrying later. (The jargon version of this reads something like: controlling for individual fixed effects, the slope on either side of the knot in the spline function was indistinguishable). This means that a marriage-minded young man who plans his education and job choices around making a strong contribution to the financial security of his family “doesn’t count” among those affected by marriage if he marries at age 29, because much of his wage growth substantially predated his marriage (29 was the median age at first marriage for men in the United States in 2016). While I am certainly not claiming that all those who marry at age 29 have been carefully planning to do so since their early 20s, I am arguing that it is a little odd to conclude that marriage plans don’t change wage-related behavior just because they don’t do so in three years right before marriage. This is especially odd when discussing a society that advocates a capstone model for marriage.

Killewald and Lundberg make a similar point in their nuanced conclusions, stating: “Although we find no individual-level effect of marriage on men’s wages, we are unable to evaluate the effect of marriage culture [italics mine] on all men’s wages.” They go on to explain that a nearly universal expectation of marriage (96% in their data) could increase the productivity of all men. It seems to me that even where 96% of men expect to marry at some point, the strength of that expectation and the amount of dedication to that expectation could affect individual wage trajectories—and not just in the years surrounding marriage. Again, the authors seem to agree, noting:

Of course, the absence of statistically significant group differences does not rule out the possibility that subgroups may have true differences in wage patterns or that subgroups of men not identified by our analyses may experience a causal marriage premium.

The second way that Killewald and Lundberg dissected the ramp-up is to chart the course of earnings among men who experienced shotgun marriages separately from those who were more likely to have anticipated their weddings further in advance. The idea here is that if the ramp-up in pre-marital earnings is due to the anticipation of marriage and its attendant responsibilities, it would be pronounced among those who prepared for marriage and muted (or absent) among those whose marriages were less likely to have been planned. Instead, they find similar premarital wage gains for both groups and argue that this is not consistent with the idea that anticipating marriage changes men’s work behavior for the better. Indeed, I find it more consistent with the idea that good financial trajectories predict marriage when a child is on the way, but that men with poorer earnings growth are viewed as “less marriageable” (by themselves, by their partners, and by normative standards). This would mean that wage growth facilitates marriage, rather than that marriage causes wage growth.

But note, again, that the focus of this new research is on what we can learn about causal effects from the timing of wage changes. The most reasonable answer seems to be “nothing definitive” rather than “marriage is useless (for wages).” Let me back up that assertion by turning to their evidence on the timing of wage changes in relation to the timing of divorce. They argue that if marriage itself conferred men’s marital wage premium, wages should decline following a divorce. Instead, they find that wage declines predate divorce (in the same way that wage ramp-ups predate marriage). Pre-divorce wage declines are consistent with an interpretation that work problems destabilize marriage (rather than marriage stabilizing work). In Killewald’s other work, she has argued persuasively that men’s unemployment is a threat to stable marriage.

Nonetheless, divorce does not happen instantaneously. In a failing marriage, men may feel less motivation to fulfill expectations of a husband as a provider because they anticipate stepping out of the husband role, so the causal question remains unanswered. Further, married men who experience earnings set-backs but never divorce are, by definition, not part of the divorce analysis. They may stay married because their earnings recover, or their earnings may recover because they stay married. Thus, Killewald and Lundberg do not estimate the benefits of staying married, rather they only show that some of the marriage wage premium dissipates even before divorce among those who divorce.

Their article is titled “New evidence against a causal marriage wage premium,” and indeed, the evidence about the timing of greater-than-expected wage changes can be explained more parsimoniously than some of the alternative explanations I’ve offered here. But it seems to me that to focus on the timing of wage growth rather than the marriage wage premium itself moves the analysis too much to the individual level—effectively removing the influences of marriage culture and long-term expectations and plans. As Steven Nock has noted, marriage is not simply an agreement that two people enter: It is also a social institution with a set of normative expectations that can transform behavior. Those changes don’t have to happen right around the time of marriage or divorce to be causal.

Laurie DeRose is a Research Assistant Professor with the Maryland Population Research Center at the University of Maryland at College Park, where she has served since its inception in 2001. She is also Director of Research for the World Family Map project.

Editor’s Note: The views and opinions expressed in this article are those of the author and do not necessarily reflect the official policy or views of the Institute for Family Studies.