July 29, 2004
Analyzing the Marriage Gap
ARRIED men make more money than single men. A lot more: labor economists estimate that even when you control for age, education and other demographic effects, the "marriage wage premium" is 10 percent to 50 percent.
The question is why.
There are two broad classes of explanation. One view holds that marriage causes men to receive higher wages. The other view is that higher wages are simply correlated with, but not caused by, marital status.
There are a variety of reasons that marriage might cause higher wages.
It might be that employers prefer married men to single men because they are more productive. After all, they have spouses who share responsibility for household chores and provide other sorts of support and assistance. Single men just have that empty apartment.
On the other hand, it could be that employers have an irrational prejudice for married men. Employers might view them as more productive, more reliable and more committed, whether or not these things are true.
Whether employer preference for married men is accurate is irrelevant. If marriage has a causal effect, single men who choose to marry will tend to receive higher wages.
The other class of theories holds that being married is simply correlated with higher wages. Just as with the causal theories, the correlation can work through productivity effects or through prejudice.
Perhaps women are attracted to stable, hardworking men. Employers also find such men attractive, so they want to hire and promote them. If true, we would see married men having higher wages. But it isn't that marriage caused those higher wages - rather, the same things that caused marriage caused the higher wages.
Such a correlation could also arise from irrelevant characteristics. Maybe women prefer good-looking guys, and employers also like handsome men, even when appearance has nothing to do with job productivity.
In this case we would also observe that the same men who are more likely to be married are also more likely to be employed and have higher wages.
In the correlation explanations of the marriage premium, the same factors that caused the men to marry caused them to get higher wages - but there is no direct causal link between marriage and high wages.
To drive this point home, suppose Hamlet is considering proposing to Ophelia, but is consumed with doubts. "To be married or not to be married, that is the question."
If the causal theories are correct, then if Hamlet proposes and Ophelia accepts, his future wages would be higher on average than if he stayed single. If the correlation theories are correct, then choosing to marry would have no effect on Hamlet's future wages, and he may as well remain single.
Recently two economists, Kate Antonovics and Robert Town of the University of California, San Diego and the University of Minnesota, respectively, have come up with a clever way to decide between the causal and the correlation theories. Their paper, "Are all the good men married? Uncovering the sources of the marital wage premium," was published in the May 2004 issue of The American Economic Review.
Their approach is based on looking at monozygotic, or identical, twins. The authors argue that twins have the same genetic endowment and (usually) the same upbringing. Since twins have the same underlying physical and mental capabilities, they should have similar productivity. Even if employers are biased toward certain irrelevant characteristics, monozygotic twins should be affected by such biases equally.
Hence differences in wages between married and unmarried twins should control for most of the effects that might cause a spurious marriage-wage correlation. If a married twin has a higher wage than his single brother, the difference is probably caused by marriage, not just correlated with it.
The economists drew on a unique dataset, the Minnesota Twins Registry, which tracked most twins born in Minnesota between 1936 and 1955. In the mid-1990's, the Registry sent the twins a questionnaire asking about their socioeconomic status.
Using this data, the researchers were able to construct a sample of 136 pairs of monozygotic twins, of whom 85 percent were married. In 23 percent of the cases, one twin was married, the other wasn't.
They extracted data from the survey on the hourly wages, weeks worked a year, age and educational attainment of the men in their sample and compared these with figures for all American males. The results implied that their sample was reasonably representative of the nation's population.
Consistent with other studies, they found a significant marriage premium: controlling for education, age and other variables, the married men in their sample earned about 19 percent more than unmarried men.
They then examined just the wage differences between twins, while still controlling for education. They found that married twins had 26 percent higher wages than their unmarried siblings. Hence, even among very similar men, those who are married earn substantially more.
The authors found essentially the same results if they factored in divorced or widowed status, or added variables like a spouse's work experience, number of children and wage history.
This result suggests that marriage really does have a causal impact on wages. Of course, it is not conclusive. After all, maybe the married twin really is different in some way from his brother, and that difference is important to both potential spouses and employers. Still, it is suggestive evidence.
So, here's the advice Hamlet would get from an economist: put your doubts aside and marry Ophelia. Stop moping around and go get a job. You may not be any happier, but at least you'll make more money.
Hal R. Varian is a professor of business, economics and information management at the University of California, Berkeley.