Friday, February 12, 2016

New Evidence Against a Causal Marriage Wage Premium

“It is a truth universally acknowledged, that a single man in possession of a good fortune must be in want of a wife.” Or is a man with a wife more likely to earn a good fortune? Many social scientists have observed a “marriage premium” for men, where their wages increase with marriage. Some policymakers have tried to encourage marriage in order to increase social welfare. But is the relationship between marriage and higher pay causal? In this week’s WAPPP seminar, Professor Alexandra Killewald, John L. Loeb Associate Professor of the Social Sciences at Harvard University, discussed her work casting doubt on the causal marriage wage premium.

The most common explanation for the marriage premium is specialization. Once married, the theory goes, couples are able to divide up the work that they previously had to do alone. If one partner is particularly good at laundry but terrible at doing the dishes, they can do laundry while their partner does the dishes. This division of labor is an efficient use of each partner’s talents and cuts down on the overall time spent doing housework. In Gary Becker’s original theory, men were thought to have a comparative advantage for work outside the home, while women specialized in domestic work. Adjusting for these specializations meant that married men would spend more time on paid labor, which would increase their wages.

However, there are many other possible explanations for the marriage premium. Professor Killewald focused on two: the reverse causation explanation and the transition to adulthood explanation. It could be that causation between marriage and higher wages actually works in reverse. Men anticipating marriage may seek out better-paying jobs to enhance their financial stability before they get married. Alternatively, it could be that most adults get married in a period of great personal growth in many areas, including work. Wage increases and marriage may be two entirely separate phenomena that tend to happen around the same time, which leads us to infer causation.

Using data from the NLSY79, a representative sample of individuals who have been surveyed from 1979 through the present day, Professor Killewald set out to empirically adjudicate between these competing explanations. She presented four tests for our three proposed explanations. However, she noted, it was unlikely that we would discover one perfect alternative explanation for the observed marriage premium. Instead, this work is meant to question the presumed causal relationship that marriage leads to higher wages.  

Test 1: How far in advance of marriage do wages increase?

If the causal marriage wage premium existed, we would expect a sharp spike in wages immediately after marriage. On average, men’s wages increased starting five years before marriage, continued to increase for three years after marriage, and then flattened out. This pattern seems to rule out specialization, as we don’t see a shift in wages at the point of marriage. (Similarly, the timeline for increasing wages doesn’t fit with discrimination: if employers were inclined to pay married men more than unmarried men, it is unlikely that we’d see that effect beginning five years before marriage!) Instead, this pattern fits much better with a reverse causation argument—that men seek out high-paying jobs in anticipation of marriage—or a transition to adulthood explanation, as a young man’s wages increase from the early stages of his career, around the same time he chooses to marry.

Test 2: What happens around divorce?

If marriage causally affected wages, we would expect to see a decline in wages immediately after divorce. Instead, wages begin to decrease about 1-2 years prior to divorce and continue to decrease thereafter. This evidence is most consistent with a reverse causation argument: anticipating divorce may mean that financial stability isn’t as important, so men are more willing to take on riskier jobs with lower pay. Alternatively, a decrease in financial stability may be a key factor leading to divorce.

Test 3: Does age of marriage matter?

Based on a causal argument, we would expect wages to increase with marriage regardless of timing. By contrast, a reverse causal or transition to adulthood explanation would predict that wages increase the most for those who get married young. Splitting the data into three groups, we find large wage increases for men who get married at 22 or under and from 23-26. However, men who marry at age 27+ show very little change in wages. At minimum, this suggests that anticipation of marriage doesn’t help men who get married late, and lends some support to the transition to adulthood hypothesis.

Test 4: What about shotgun weddings?

The reverse causal explanation rests on men anticipating marriage and seeking out higher-paying jobs ahead of time. But what about marriages that are less likely to be anticipated? Professor Killewald examines marriages in the dataset where the couple’s first child was born within seven months of the wedding. In this data, we still see a wage increase in the years leading up to marriage even when the marriage was not anticipated. This is consistent with the transition to adulthood hypothesis rather than either of the causal theories.

Based on these four tests, the evidence provides support for the reverse causal explanation and particularly for the transition to adulthood hypothesis. None of these tests validated the causal specialization explanation: being and staying married is not directly related to higher wages for men. This study should make us cautious about using marriage as a policy lever.

No comments:

Post a Comment