One of my favorite teaching exercises in my Introductory Income Taxation class is to go through the New York Times wedding announcements from late December and early January, and to have students identify couples that got married on the “wrong” side of the New Year divide. Some single-earner couples that presumably receive a marriage bonus—e.g., an engineer marrying a grad student—nonetheless wed in early January, though they likely would have been better off accelerating their nuptials to December. Other couples that probably face a marriage penalty—e.g., two associates at the same law firm—wed in late December, though they likely would have been better off waiting a few more weeks. I ask my students, “What were these couples thinking?” The answer that I get is: “They were probably thinking about love and not about tax.”
For those of us who love thinking about tax, this is all somewhat of a puzzle. Do people really make such momentous life decisions without considering the tax implications? Ed Fox seeks to answer that question in a creative new paper. His main finding is that, at least historically, tax incentives have had a quantitatively and statistically significant effect on marriage patterns. While anecdotes from the Times might lead us to believe that couples make marriage choices without considering the tax consequences, Fox’s study suggests that tax incentives do indeed affect the decisions of some couples to say “I do.”
Fox seizes upon a natural experiment that the Supreme Court and Congress unwittingly created at mid-century. From 1913 until 1948, federal tax law required individuals to file their own returns regardless of whether they were married or single. This gave rise to the question of how couples in community property states should allocate income between spouses. The Supreme Court answered that question in the 1930 case Poe v. Seaborn, which held that married couples in community property states could split their income 50/50. That holding resulted in a significant marriage bonus for single-earner couples in community property states, but no such bonus in common law states.
To understand the differential effects of Seaborn, take a look at the 1930 tax brackets. Rates were steeply progressive, with a 1.5% rate on the first $4,000 of income, rising to a 25% rate on income over $100,000. If Mr. Seaborn earned $100,000 and Mrs. Seaborn earned $0 in a community property state, they could split their income and each pay tax on $50,000—for a final bill of $5,260 each ($10,520 total). If they were married and lived in a common law state, or if they were unmarried, then Mr. Seaborn would be taxed on $100,000 and would end up paying $16,440. Consequently, high-earning Mr. Seaborn had a strong tax incentive to find a lower-earning bride in a community property state, but no such incentive in a common law state.
All that changed with the Revenue Act of 1948, which allowed income splitting for couples in common law and community property states alike. Thus the marriage bonus that previously existed only in community property states now extended to couples in common law states. If taxes affect marriage decisions, then we might expect to see a post-1948 uptick in common law states’ marriage rates.
The 1948 change is an especially good natural experiment because it comes with two built-in control groups. First, the rule regarding income splitting changed only in common law states, so we have no reason to expect any effect in community property states. And second, the benefits of income splitting are largest for couples with one high-income earner (which at mid-century usually meant a male). So if taxes affect marriage, we would expect to see a more pronounced rise in marriage rates among high-income males in common law states than among high-income males in community property states or low-income males in common law states. (Of course, the only marriages at mid-century were opposite-sex, so male marriage rates and female marriage rates were not far apart.)
Fox draws data from the 1960 Census, which asked a sample of the population what year they first married. He finds an 0.52 percentage point increase in the marriage rate of high-income, never-before-married men in common law states following the 1948 law change. Put differently, Fox’s findings suggest that high-income, never-married males in common law states entered their first marriages approximately 3 months earlier (on average) than they would have in the absence of a marriage bonus—not quite a race to the altar, but not an insubstantial effect either.
What does this historical evidence tell us about the present day? Fox writes that his estimate from 1948 “can reasonably be viewed as a floor on the current effect” of taxes on marriage rates among single-earner middle-income couples. This is so for two reasons. First, cohabitation outside of marriage is much more common today than it was at mid-century. Back then, the choice for most couples was between living separately and getting married. Now, the choice for most couples is between living together unwed and getting married. In this respect, marriage was a more dramatic step 70 years ago than today. Second, marriage bonuses today are even larger than 70 years ago—at least for single-earner middle-income couples. Fox notes that marriage bonuses for single-earner couples making $50,000 to $80,000 are at least twice as large today as circa 1948. If a smaller tax incentive led couples to take the more dramatic shift from living separately to getting married 70 years ago, then a larger tax incentive is even likelier to lead couples to make the less dramatic shift from unmarried cohabitation to marriage now.
There are, to be sure, limits to the inferences that we can draw about the present based on mid-century marriage patterns. As L.P. Hartley writes in The Go-Between, “The past is a foreign country: they do things differently there.” Two differences are especially relevant: (1) the female labor force participation rate has increased massively; and (2) the gender wage gap has narrowed significantly (though not entirely). As a result, the share of couples with a single earner has declined, and the wage gap between working husbands and working wives has declined as well. Recall that the marriage bonus is largest for couples with a single earner or for spouses with disparate incomes. As the percentage of couples with those characteristics shrinks, the portion of the population that can claim a large marriage bonus does too.
None of this contradicts Fox’s claim that the marriage bonus is larger today for single-earner middle-income couples than it was at mid-century. But it is also true that single-earner married couples are much rarer today than they were decades ago. How this all affects marriage decisions is difficult to know—alas, Congress and the courts have not given us the sort of natural experiment in recent years that Fox so skillfully leverages here. For the time being, then, Fox’s study—though focused on an earlier era—may offer the best available evidence on the relationship between federal tax law and the timing of marriage. Better, even, than Sunday Styles.