Although both marriage penalties and marriage bonuses exist at all income levels under the federal income tax, the system is tilted toward penalties for lower-income couples, toward bonuses for middle-income couples, and back toward penalties for upper-income couples. This Article begins by explaining how the tax rules produce these differing treatments of marriage at different points in the income distribution. It then argues that the increase in recent decades in the social acceptability and prevalence of cohabitation makes tax marriage effects a more serious concern—in terms of both behavioral effects and fairness—than in earlier decades. After demonstrating that Congress has never offered any justification for the differing tax treatments of marriage at different income levels, and that no plausible defense exists for the current distribution of penalties and bonuses, the Article offers several policy recommendations. The most basic and most important recommendation is simply that, whatever Congress does in this area, it should make conscious decisions about the appropriate distributions of penalties and bonuses at various income levels, instead of following its current practice of stumbling into a set of poorly understood and almost-impossible-to-defend effects.
Introduction: Marriage Penalties, Marriage Bonuses, or Both
It is the classic trilemma of the income tax treatment of marriage. As a matter of simple arithmetic, it is impossible for a tax system to feature simultaneously (1) progressive marginal tax rates, (2) joint filing by married couples (in the service of producing equal taxes on equal-income married couples), and (3) marriage neutrality (that is, no tax marriage penalties or bonuses). If the legislature insists on the first two features, as Congress has for many decades, then the third desideratum is unachievable. Instead of marriage neutrality—under which any two people would pay the same combined tax whether unmarried or married—there will be marriage penalties, marriage bonuses, or both. Although a legislature committed to the first two goals must violate marriage neutrality, the legislature has a great deal of freedom to determine whether, and to what extent, to skew the neutrality violations toward either marriage penalties or marriage bonuses.
The choices facing the legislature can be illustrated with a simple two-bracket tax rate system, featuring an exemption level (in effect, a zero bracket) and a single positive rate applying to all income above the exemption. Suppose the legislature begins by setting the rate structure for single taxpayers: a $50,000 exemption, and a flat tax of 20% on income above the exemption. Given the decision to tax a married couple on their combined income, without regard to how the income is distributed between the spouses, the only remaining question for the legislature is the exemption amount for married couples.
There are two polar choices, and any number of intermediate options. At one extreme, the legislature could choose to set the exemption level for married couples at the same amount—$50,000—applicable to single taxpayers. This choice would result in marriage neutrality for one-earner couples, substantial marriage penalties for two-earner couples, and marriage bonuses for no one. If two people had individual incomes of $120,000 and zero, their combined tax liabilities as singles would be $14,000 (a 20% tax on $70,000 of above-exemption income on the one, and no tax on the other), and their tax as a married couple would also be $14,000 (again, a 20% tax on $70,000). This is tax marriage neutrality for one-earner couples. If two people each had individual incomes of $60,000, they would pay a combined tax of $4,000 as singles (a 20% tax on $10,000 for each), but they would owe $14,000 tax as a married couple with a combined income of $120,000. The result is a $10,000 marriage penalty on the equal-income two-earner couple. Married couples with income splits between 100-zero and 50-50 would also be subject to marriage penalties, although their penalties would not be as large as that of the 50-50 couple. For example, a couple with individual incomes of $90,000 and $30,000 would pay a combined tax of $8,000 as singles (imposed on the $40,000 above-exemption income of the higher earner), compared with a joint return liability of $14,000—resulting in a marriage penalty of $6,000.
At the other extreme, the legislature might decide to set the exemption level for married couples at $100,000—twice the exemption level for single taxpayers. In that case there would be marriage neutrality for equal-income two-earner couples, and marriage bonuses for everyone else. The couple with individual incomes of $60,000 and $60,000, with a combined tax bill of $4,000 as singles, would also owe $4,000 (on $20,000 of above-exemption income) as a married couple filing a joint return. The one-earner couple with incomes of $120,000 and zero, with a $14,000 combined tax bill as singles, would owe only $4,000 tax as a married couple; their marriage bonus would be $10,000. The couple with incomes of $90,000 and $30,000, with an $8,000 combined tax bill as singles, would also owe only $4,000 as a married couple; their marriage bonus would be $4,000.
Between these two extremes, the legislature could set the joint return exemption level at more than the singles exemption but less than twice the singles exemption. Instead of producing either penalties without bonuses, or bonuses without penalties, any intermediate approach will produce both bonuses and penalties. Suppose, for example, that the legislature decides to split the difference by setting the joint return exemption at $75,000—that is, 1.5 times the singles exemptions. In that case there will be both marriage bonuses for one-earner couples and other couples with very unequal divisions of income, and marriage penalties for two-earner couples with more nearly equal divisions of income. Although having both penalties and bonuses might seem like the worst of both worlds, this intermediate approach produces smaller penalties than the only-penalties approach and smaller bonuses than the only-bonuses approach. The one-earner couple with $120,000 income enjoys a marriage bonus of $5,000 ($9,000 tax married versus $14,000 combined tax unmarried). This is $5,000 less than their marriage bonus under the bonuses-only approach. The equal-income two-earner couple (each with $60,000 income) faces a marriage penalty of $5,000 ($9,000 tax married versus $4,000 combined tax as singles). This is $5,000 less than their marriage penalty under the penalties-only approach.
In any intermediate system, there is a breakeven division of income (which may be different, however, at different combined income levels). With the intermediate system imagined here, the breakeven division of $120,000 would be $95,000 and $25,000 (roughly 79% and 21% of combined income). A couple with those incomes would pay the same combined tax—$9,000—whether married or unmarried. As illustrated by the examples in the preceding paragraph, a couple with $120,000 combined income and a more unequal income division would enjoy a marriage bonus, while a couple with the same combined income and a less unequal division of income would suffer a marriage penalty.
These examples are offered in support of two simple points. First, if we are to have progressive marginal rates and joint filing by married couples, then we are forced to accept that the tax system will not be marriage neutral. Second, even within the constraints of progressive rates and joint filing, a legislature has choices as to the nature of the marriage non-neutralities. Basically, the options are (1) big marriage penalties and no marriage bonuses, (2) big marriage bonuses and no marriage penalties, and (3) both penalties and bonuses, but smaller penalties than under the first option and smaller bonuses than under the second option. As a matter of first impression, it might seem that the fairest approach (assuming that reconsideration of the commitments to progressive rates and joint filing is not an option) would be the splitting of the difference embodied in option three. If marriage penalties and bonuses are an unfortunate result of the marriage tax trilemma, rather than the embodiment of a legislative decision to penalize or subsidize marriage, then a system with little or no net marriage penalty or bonus—because total penalties and bonuses in the system are roughly equal and offsetting—may be the best that can be done. This approach could be thought of as a rebuttable presumption, under which the choice should be to split the difference unless there is a persuasive reason to skew the system toward penalties or bonuses.
Broadly speaking, the current federal income tax does split the difference, in the sense that it features both substantial marriage penalties and substantial marriage bonuses. What is odd, however, is that the system splits the difference quite differently at different income levels. A simple way of measuring how the system splits the difference is to compare, at various income levels, the maximum marriage bonus (for a one-earner couple) and the maximum marriage penalty (for an equal-income two-earner couple). If the two maximums are the same, the system splits the difference evenly. If the maximum bonus is larger than the maximum penalty, the system is skewed toward bonuses. And, of course, if the maximum penalty is larger than the maximum bonus, the system is skewed toward penalties.
Table 1 sets forth the maximum marriage penalties and bonuses at selected combined income levels, under the federal income tax rules in effect for 2014. As shown in Table 1, at the very low combined-income level of $20,000, the system splits the difference almost perfectly; the maximum bonus and the maximum penalty differ by only $4. The picture is very different, however, at the moderately low income level of $40,000. When there is one “qualifying child” (for purposes of the child tax credit, the earned income credit, and the dependency exemption), the maximum penalty is more than twice the maximum bonus. And when there are two qualifying children at the $40,000 income level, the maximum penalty is 2.7 times the maximum bonus. (The maximum penalty of $4,271 is also impressively large in percentage-of-income terms, at more than 10% of the combined pre-tax income of the couple.) Remarkably, this skewing toward penalties is utterly reversed in the middle-income ranges. At $90,000 combined income, the maximum bonus is $5,334 and the maximum penalty is zero. There is a re-reversal in the upper-income ranges. At $1,000,000 of combined income the maximum penalty (of more than $32,000) is almost three times the maximum bonus. In sum, the system is strongly skewed toward penalties for the working class, is maximally skewed toward bonuses for the middle class, and is again strongly skewed toward penalties for the wealthy.
Table 1: Maximum Marriage Bonuses and Penalties at Selected Combined Income Levels Under 2014 Federal Income Tax Rules
|Combined Income||Maximum Bonus||Maximum Penalty|
|$20,000 (with one qualifying child)||$347||$351|
|$40,000 (with one qualifying child)||$1,265||$2,579|
|$40,000 (with two qualifying children)||$1,581||$4,271|
|$90,000 (no children)||$5,284||zero|
|$1,000,000 (no children)||$12,749||$32,660|
The different skewings at different income levels are not necessarily irrational. Couples with $40,000 combined income are differently situated from those with $90,000 income, who in turn are very differently situated from those with $1,000,000 income, so it is not inconceivable that policy considerations could favor marriage penalties for the first and third groups, while favoring marriage bonuses for the middle group. On the other hand, Congress has never offered an explanation for those different treatments (it is not clear that Congress even realizes what it has done in this area), and certainly no explanation comes readily to mind.
This Article takes a critical look at the differing distributions of marriage penalties and bonuses at different combined income levels—skewed toward penalties at the bottom and top, and toward bonuses in the middle—under the federal income tax. Part I explains how the dramatic increase in recent decades in the prevalence and social acceptability of cohabitation has made tax marriage penalties and bonuses a more serious concern than ever before, in terms of both fairness and incentive effects (especially with respect to the choice between cohabitation and marriage).
Part II describes in some detail the sources and distribution of marriage penalties and bonuses under current law—including explanations of how the tax laws produce the results set forth in Table 1. It notes that, although discussions of marriage penalties and bonuses have traditionally focused on the basic tax rate structure of § 1 of the Internal Revenue Code, other Code provisions also produce marriage effects. At lower income levels, the earned income credit (“EIC”) tends to be a more important source of marriage effects than the basic rate structure. At higher income levels, the alternative minimum tax (“AMT”) is a major source of marriage effects (mostly penalties) for couples within its domain. Several provisions recently added by the Affordable Care Act—most notably the premium tax credit and the 3.8% tax on investment income of upper-income taxpayers—also produce significant marriage effects (again, mostly penalties) for the couples to whom they apply. Part II also explains that the magnitude of marriage tax effects (as contrasted with the skewing of the effects toward penalties or bonuses) is a function of the extent of the marginal tax rate progressivity built into § 1 and other provisions with rate-like effects. The less progressive the system—in terms of marginal rate structure—the smaller the penalties or bonuses resulting from the joint taxation of married couples.
Part III considers whether there is any plausible policy justification for mostly penalizing marriage for working class couples, while overwhelmingly rewarding marriage for the middle class and mostly penalizing marriage for the upper class. The task requires some imagination, given that Congress has never attempted to justify these disparate treatments. After considering some reasons why legislative tax policy toward marriage might vary by income level, Part III concludes that the particular differences embodied in the current income tax structure are difficult or impossible to defend.
Part IV concludes with four tentative policy recommendations. First: At all income levels, Congress should pay attention to marriage effects and should make conscious decisions about whether to skew the rules toward penalties or toward bonuses, or to split the difference. As modest as this suggestion surely is, its adoption would constitute a dramatic departure from standard operating procedure in this area. Second: In making those conscious decisions, Congress should start with a rebuttable presumption that the same balance between penalties and bonuses should be struck at all income levels, and with a rebuttable presumption that the most appropriate balance is to split the difference. The burden of persuasion should be on anyone arguing for different treatments of marriage at different income levels, and on anyone arguing for net marriage penalties or bonuses at any income level. There is one area where it is quite possible that both presumptions can be overcome; a case can be made for a tilt toward marriage bonuses for lower-income couples. Third: The problems of marriage penalties and bonuses are so significant and so intractable as to call for a reconsideration of the commitment to the principle of equal tax on equal-income couples (and thus to joint returns). Fourth: Congress should seriously consider ways of decreasing tax marriage effects without sacrificing progressivity, by making greater use of techniques to produce average rate progressivity without using progressive marginal rates (of which the prototype is a system of universal “demogrants” financed by a flat rate tax).
Tax marriage penalties and bonuses are based on counterfactual comparisons; spouses pay more or less tax than if they were unmarried. The standard analysis assumes that each partner, unmarried, would have the same income that she or he has married. Early in the era of income tax marriage effects (beginning with bonuses in 1948, followed by the introduction of penalties in 1951), unmarried cohabitation was rare and generally socially unacceptable. When cohabitation was not a socially available counterfactual to marriage, neither marriage penalties nor bonuses were likely to have much visceral impact. If couples did not live together before marriage, and if going from marriage to unmarried cohabitation was not a realistic option, couples would have had little reason to compare their joint marital tax liability with the sum of their unmarried tax liabilities. Their hypothetical combined tax liabilities as two single people, not living together, would not have been a meaningful reference point for a married couple. And unmarried persons not living together would not have thought of themselves as having a combined tax liability of any amount, so they would not experience any increase or decrease in their combined tax liability when they married and began living together. Even in the rather unlikely event that a two-earner couple did think of their combined tax liability as having been increased by their marriage, they might have accepted it as a reasonable legislative response to the economies of scale of having one household instead of two. In short, in the absence of a realistic cohabitation counterfactual, neither marriage bonuses nor penalties raised very compelling fairness concerns.
The story was much the same for incentive effects. If marriage were the only socially acceptable way to have a life together, two-earner couples in love would not be dissuaded from marriage by the prospect of a tax marriage penalty (in the unlikely event that they even thought in marriage penalty terms). It would have been equally implausible for an earner to marry a non-earner in order to obtain a marriage bonus; the non-tax considerations favoring or disfavoring the married state would overwhelm the married tax advantage (again, in the unlikely event the persons even thought in terms of marriage tax effects).
In recent decades there has been a striking increase in the prevalence and social acceptability of unmarried cohabitation. Cohabitation has become particularly prevalent at lower-income levels, where tax marriage effects tend to be large in dollar terms and very large in percentage-of-income terms. If couples today feel that they have a genuine choice between cohabitation and marriage, then tax marriage effects are a much more pressing policy concern, with respect to both equity and incentives, than in earlier decades. With respect to fairness, penalties and bonuses are more real, and thus more objectionable, when they are based solely on the presence or absence of a marriage license, than when they are bound up with the difference between living together and living apart. And as for incentives, it is much more plausible that a couple committed to living together in any event would decide whether to do so with or without a marriage license based on tax considerations, than that two people in an earlier era would have decided whether to be a couple at all (which they could do only through marriage) based on a tax analysis.
A large and rather inconclusive collection of empirical studies—mostly quantitative, but a few qualitative studies as well—suggests that, even with cohabitation as a socially available alternative to marriage, the influence of tax on the choice between marriage and cohabitation is somewhere in the range of small to nonexistent. It would be a mistake, however, to conclude that if there are no major behavioral effects of tax marriage penalties and bonuses, then there are also no major policy concerns. A generation ago, Boris Bittker suggested that tax incentives create inefficiencies to the extent taxpayers behaviorally respond to the incentives, but that tax incentives create inequity to the extent they do not cause taxpayers to change their behavior. Although Bittker’s observation concerned investment tax incentives, rather than incentives to seek or avoid marriage, his point applies here as well. To the extent taxpayers ignore tax consequences in deciding between marriage and cohabitation, we have avoided a tax influence on marital decisions only by acquiescing in some truly disturbing inequities. If the two-earner couple with $40,000 combined income and two children (Table 1) chooses marriage over cohabitation despite the rather astounding tax marriage penalty of more than $4,200 (more than 10% of their pretax income), that is every bit as troubling as if tax had driven them to unmarried cohabitation. It does not matter whether they chose marriage and the tax marriage penalty out of heroism or out of ignorance; either way the result is grossly unfair.
- Income Tax Marriage Penalties and Bonuses: A Closer Look
- Code Provisions Producing Marriage Penalties and Bonuses
This section offers a closer look at the design features of the income tax responsible for the skewing toward marriage penalties at $40,000 of combined income, toward bonuses at $90,000, and again toward penalties at $1,000,000. It also describes how recent legislation has introduced several significant new tax effects of marriage. Although this section describes the most widely applicable sources of marriage penalties and bonuses in the income tax, it does not attempt a comprehensive survey of the several dozen income tax provisions, the application of which is affected by marital status.
For a couple with $40,000 combined income and no children, the provisions producing marriage effects are the standard deduction, the 10% bracket, and the 15% bracket. The joint return standard deduction is twice as large as the standard deduction for an unmarried taxpayer (other than a head of household), and the joint return 10% and 15% brackets are twice as wide as equivalent brackets for single taxpayers (again, other than heads of households). The marriage tax effects depend critically on the presence or absence of children. First consider the treatment of childless one- and two-earner couples. A childless one-earner married couple with $40,000 income enjoys a marriage bonus of $1,383.75 (the difference between a joint return liability of $2,047.50 and an unmarried tax liability of $3,431.25 on the earning partner, on the assumption that the earning partner claims the non-earning partner as a dependent). In the absence of children, there are no marriage penalties at the $40,000 combined income level. Two childless cohabitants, each with $20,000 income, would each owe tax of $1,023.75, resulting in a combined tax liability of $2,047.50. If they were married, their joint return income tax liability would also be $2,047.50.
The marriage tax effects are different, however, in the presence of children. If a two-earner couple has a child and is unmarried, then one of the partners could file as a head of household and take advantage of the head-of-household standard deduction and 10% and 15% rate brackets. The result is a marriage penalty, although not a very large one. Taking into account the dependency exemption and the child tax credit (but not the EIC), the marriage penalty on an equal-earnings couple with $40,000 of combined income would be $251.25. If one partner files as a head of household and claims the child (for purposes of both the child tax credit and the dependency exemption), that partner will have a federal income tax liability of negative $700 (in other words, that partner will receive a $700 transfer payment from the government). The other partner, filing as a single person, will have a tax liability of $1,023.75. Their combined (net) tax is $323.75. As a one-child married couple filing a joint return, their tax will be $575. The $251.25 difference—all of which is due to the availability of head-of-household status for one partner under the unmarried option—is their marriage penalty.
But the above analysis disregards the EIC, the phaseout rules of which can produce very large marriage penalties. The one-child EIC is reduced by 15.98 cents for every dollar by which adjusted gross income exceeds the phaseout threshold. In 2014 that threshold is $17,830 for an unmarried taxpayer and $23,260 (far less than twice the unmarried threshold) for a married couple. From the maximum one-child EIC of $3,305, the phaseout reduces the EIC of a head-of-household with $20,000 income by $347, to $2,958. The EIC phaseout hits the two-earner married couple much harder. At $40,000 combined income, the phaseout reduces their EIC by $2,675, from $3,305 to $630. The EIC-produced marriage penalty is the $2,328 difference between the unmarried and married EICs. Combining the marriage penalties resulting from the head-of-household rules and the EIC phaseout, the one-child $40,000 income couple faces an income tax marriage penalty of $2,579.25—of which more than 90% is attributable to the EIC’s phaseout rules.
What about a one-earner couple with $40,000 income and one child? Married, their joint return tax liability (taking into account all relevant provisions, including the EIC) will be negative $55. If they were unmarried and the earning partner filed as a head of household (and claimed both the child and the non-earning partner as dependents), the earning partner’s tax would be $1,210. The marriage bonus—the maximum marriage bonus for a one-child couple with $40,000 income—is $1,265. This is slightly less than half of the maximum marriage penalty for a one-child couple at that income level.
To the considerable extent that arguments in favor of pro-marriage public policies are premised on the claimed advantages to children of having married parents, the above results seem perverse. As the above analysis has shown, Congress has eliminated marriage penalties (and provided substantial marriage bonuses) at the $40,000 combined-income level in the case of childless couples, but has tilted the system toward marriage penalties for one-child couples at the same income level.
The tilting is in the same direction, but even more severe, for a couple at the same income level with two children. Again, the marriage penalty is caused much more by the phaseout of the EIC than by the rest of the income tax system. Disregarding the EIC for the moment, an unmarried parent with $20,000 income, filing as a head of household and claiming one of the children for purposes of both the child tax credit and the dependency exemption, would have a negative tax liability (that is, would receive a transfer payment) of $700. Assuming the parents and the children all live together in a single household, the other parent cannot use the other child to support that parent’s also filing as a head of household (because there is only one household, and the first parent has filed as its head). The other parent can, however, claim one of the children for purposes of the child tax credit and the dependency exemption. That parent’s tax liability, filing as a single taxpayer (other than a head of household) will be negative $410. The combined result for the two unmarried partners will be a negative tax (transfer payment) of $1,110. If the couple is married and files a joint return (and again disregarding the EIC for the moment) they would have a negative tax of $820. The marriage penalty is a modest $290 ($1,110 – $820).
But what about the EIC? In this situation, two features of the EIC favor two-earner unmarried cohabitants over two-earner married couples. First, the maximum two-child credit ($5,460 in 2014) is considerably less than twice the maximum one-child credit ($3,305 in 2014). A married couple with two children is relegated to the two-child credit, but unmarried cohabiting parents can divide the children between them for EIC purposes, with each claiming the one-child EIC. Second, as in the one-child situation, the EIC phaseout rules produce large marriage penalties for two-earner couples with similar incomes. Dividing the children between them, each unmarried partner will be entitled to a one-child EIC (after phaseout) of $2,958, resulting in a total EIC of $5,916 for the family. As a married couple, they will start (before phaseout) with a smaller EIC and they will be subject to a much larger phaseout reduction. Their after-phaseout two-child EIC will be only $1,935. Thus, their joint return EIC is almost $4,000 less than the combined EICs they could properly claim if they were not married. Of their total marriage penalty of $4,271, about 93% is attributable to the EIC.
As in the one-child situation, the maximum marriage penalty on the two-child two-earner couple is much larger than the maximum marriage bonus for a two-child one-earner couple. Married, a two-child one-earner couple will have a joint return tax liability (taking into account all relevant provisions, including the EIC) of negative $2,755. If they were unmarried and the earning partner filed as a head of household (and claimed both children and the non-earning partner as dependents), the earning partner’s tax would be negative $1,173.50. Thus, the marriage bonus is $1,581.50. For two-child married couples with $40,000 in combined income, the maximum marriage penalty of $4,271 is 2.7 times the marriage bonus of $1,581.50.
At $90,000 combined income, with the phaseout of the EIC complete and the phaseout of the child tax credit not yet begun, the analysis of marriage penalties and bonuses centers on the tax rate schedule of § 1 of the Code and on the standard deduction. Because the bracket widths (in the relevant brackets) and the standard deduction amount for joint returns are twice those for single taxpayers (other than heads of households), childless partners with $45,000 income each will pay the same tax whether married or unmarried. At income of $90,000 a one-earner childless married couple enjoys a very substantial marriage bonus of $5,283.75 (the difference between their joint return liability of $9,547.50 and the $14,831.25 tax on the unmarried earning partner, assuming the earning partner claims the non-earning partner as a dependent).
If there is a child, the availability of head-of-household status for one unmarried partner creates a moderate-sized marriage penalty of $628.75 for an equal-income couple. As at the $40,000 income level, here too it is ironic that the system eliminates marriage penalties for all childless couples but not for all couples with children. However, with a much smaller maximum marriage penalty at $90,000 than at $40,000, the irony at $90,000 is considerably less striking.
At the $1,000,000 combined income level, marriage penalties return with a vengeance. With the EIC, child tax credit, and dependency exemptions all fully phased out, the presence or absence of children has very little effect on marriage penalties and bonuses. Accordingly, the following example assumes a childless couple, with $500,000 income for each partner. Marriage penalties for equal-income and similar-income couples are large at this income level, because toward the top of the tax rate schedule the joint return bracket widths are nearly the same as the single taxpayer bracket widths. Two childless taxpayers, each with $500,000 income (and claiming the standard deduction), will have a tax liability of $152,591 each, for a total of $305,182. Married, their joint return tax would be $337,842. The marriage penalty of $32,660 is impressive in absolute terms, and is substantial even as a percentage of their combined pre-tax incomes (3.27%). The marriage bonus for a one-earner couple at $1,000,000 is not nearly so large. The unmarried earning partner would owe tax of $350,591, which would be reduced by marriage by only $12,749 (to $337,842).
A few tax marriage effects not implicated in the above examples are also worth mentioning here. First, the alternative minimum tax (“AMT”) features parameters sharply tilted toward marriage penalties. At the $800,000 combined income level, there is no AMT marriage bonus for a one-earner couple, whereas an equal-income two-earner married couple will face an AMT marriage penalty of $3,650. This extreme imbalance is explained by the fact that the only relevant tax parameter—the breakpoint between the 26% and 28% brackets in the AMT tax rate schedule—is identical for joint returns and for single taxpayers.
Of course, the other notable feature of this AMT example is that, even with its maximum tilt toward marriage penalties, the penalty on the two-earner couple is not very large. It is less than half of one percent of the couple’s combined pre-tax income. The explanation for the combination of maximum tilt and small penalty is that the size of tax marriage effects—whether penalties, bonuses, or both—is a function of the extent of the system’s marginal tax rate progressivity. Above the income level at which the phaseout of the AMT exemption is complete, the flatness of the AMT is marred only by the shift from a 26% to a 28% rate at the $182,500 income level. Even with the maximum tilt toward penalties (resulting from the specification of the same breakpoint between the two rates for joint returns and for returns of unmarried taxpayers), the maximum penalty is only $182,500 x (28% – 26%) = $3,650. If Congress had chosen the other extreme, setting the breakpoint at $365,000 for joint returns, the maximum marriage bonus (for a one-earner couple) would also have been only $3,650. And if Congress had set the joint return breakpoint at any amount between $182,500 and $365,000, the sum of the absolute values of the maximum marriage penalty and the maximum marriage bonus would also have been $3,650.
As noted at the beginning of this Article, a progressive marginal rate structure is one of the three horns of the marriage tax trilemma. Decreasing the progressivity of the rate structure does not eliminate tax marriage effects (as long as some marginal rate progressivity remains), but it can greatly reduce the magnitude of the effects. Opinions will differ, of course, as to whether reducing the size of marriage effects is an important enough goal to warrant a significant reduction in the progressivity of the rate structure.
Three other tax marriage effects not implicated in the above examples were introduced in 2010 by the Affordable Care Act. Section 1411 of the Internal Revenue Code imposes a tax at the rate of 3.8% on the lesser of (1) a taxpayer’s net investment income or (2) the amount by which the taxpayer’s modified adjusted gross income exceeds a threshold amount. The threshold amount is $200,000 for single taxpayers and $250,000 for joint returns. Unmarried partners, each with $200,000 of investment income (and no other income), would not be subject to the tax. If they were married, they would owe tax of $5,700 (3.8% of the $150,000 of income above the threshold). The maximum marriage bonus for a one-income couple at the same combined income level is much smaller. A single taxpayer with $400,000 of investment income (and no other income) would owe tax of $7,600 (3.8% of $200,000). Marriage to a spouse with no income (investment or otherwise) would reduce the tax to $5,700, for a marriage bonus of only $1,900. The maximum penalty is thus three times the maximum bonus.
The Affordable Care Act also introduced the “[a]dditional [Medicare] tax,” imposed at the rate of 0.9% on wages and self-employment income in excess of $250,000 (joint return) or $200,000 (unmarried taxpayer). Because the joint return threshold is only slightly higher than the single taxpayer threshold, this tax is also skewed toward marriage penalties.
The premium assistance credit (“PAC”) of § 36B of the Code (another innovation of the Affordable Care Act), which can produce some very large marriage effects, is also strongly tilted toward penalties. The credit is available to low- and moderate-income taxpayers purchasing individual or family health insurance through a state-based health insurance exchange (or through a federally-established exchange, in the absence of a state-run exchange). To be credit-eligible, a taxpayer must have “household income” of at least 100% but not more than 400% of the federal poverty level (as adjusted for family size), and must not be covered by Medicare or by affordable employer-sponsored health insurance. The generosity of the credit decreases as household income increases. The credit amount is set so that the after-credit cost of basic health insurance does not exceed a specified percentage of the taxpayer’s household income, with the specified percentage decreasing as household income (expressed as a percentage of the federal poverty level) increases.
The term “household income” might lead one to suppose that the PAC rules would aggregate the incomes of unmarried cohabitants, with the result that the PAC would never produce either marriage bonuses or marriage penalties (relative to the cohabitation alternative). However, the technical definition of household income includes only the incomes of the taxpayer and any persons whom the taxpayer claims as a dependent (and who are required to file tax returns of their own). Because of this claimed-as-a-dependent rule, the incomes of a two-earner cohabiting couple would not be aggregated as household income.
As an example of the marriage effects of these rules, imagine two childless cohabiting partners, each with income of $17,235 (150% of the federal poverty level for a one-person household, for purposes of PAC determinations for 2014). Assume that each partner satisfies all the eligibility criteria for the PAC, and that the cost of basic individual health insurance for each is $4,000 for the year. Each is entitled to a credit sufficient to reduce the after-credit cost of health insurance to 4.0% of his or her income. This works out to a credit of $3,311 ($4,000 – $689) per person, for a combined credit amount of $6,662. If they were married, their household income of $34,470 would equal approximately 222% of the federal poverty level for a family of two. The increase from 150% to 222% results because marriage doubles household income, while the two-person-household poverty level is only 35% greater than the one-person level. At 222% of the poverty level the after-credit cost of insurance is set at 7.1% of household income, which for this couple is $2,447. If their combined pre-credit insurance cost remains $8,000, their credit will be $5,553 ($8,000 – $2,447). The marriage penalty is $1,069 ($6,662 – $5,553).
What about one-earner couples? At first glance, it might seem that they would enjoy modest marriage bonuses under the PAC, because marriage increases their household size and thus increases their official poverty level. As the statute is written, however, a one-earner couple does not need to be married to be treated as a two-person household for purposes of the PAC. If one partner has no income, the earning partner can claim the non-earning partner as a dependent, and the allowance of the dependency exemption triggers the treatment of the partners as members of a two-person household for PAC purposes. The result of the inconsistent treatment of unmarried cohabitants—as two separate one-person households in the case of a two-earner couple, but as a single two-person household in the case of a one-earner couple—is a combination of large marriage penalties for two-earner couples and marriage neutrality (relative to cohabitation) for one-earner couples.
Most recently, the American Taxpayer Relief Act of 2012 contributed to the skewing of the income tax toward marriage penalties at higher income levels by reinstating and amending the so-called Pease and PEP provisions. The threshold for the application of the “overall limitation on itemized deductions” of Code § 68 (Pease) is only slightly higher for joint returns ($300,000 for 2013, with inflation adjustments in later years) than for heads of households ($275,000 in 2013) and for other unmarried taxpayers ($250,000). The effect, of course, is a tilt toward marriage penalties. The reinstated § 151(d)(3) phaseout of personal and dependency exemptions (PEP) features the same AGI thresholds as Pease, with the result that PEP is also skewed toward marriage penalties.
The overall distribution of marriage penalties and bonuses at any given income level in the population depends, of course, not only on whether the statutory rules are skewed toward bonuses or toward penalties, but also on how incomes are actually distributed between partners at that income level. Imagine, for example, a system skewed toward penalties in the sense that the maximum marriage penalty (on an equal-income couple) at a particular income level is three times the maximum bonus (for a one-earner couple) at the same income level. Despite that skewing, if one-earner couples greatly outnumber two-earner couples at that income level, then couples presented with marriage bonuses would outnumber couples facing marriage penalties, and the total dollar amount of bonuses might exceed the total dollar amount of penalties. What do we know about the actual distribution of penalties and bonuses?
The available studies are less recent than one might wish, but they generally indicate an overall skewing toward marriage penalties, especially in the lower-income ranges. Examining joint filers for tax year 2000, Janet Holtzblatt and Robert Rebelein found that 48.3% of jointly filing married couples were subject to marriage penalties, 41.9% enjoyed marriage bonuses, and 9.8% had neither penalty nor bonus. With total penalties of $30.0 billion and total bonuses of $28.5 billion, the net result was a marriage penalty of $1.5 billion. Holtzblatt and Rebelein also isolated EIC-caused tax marriage effects. Considering all tax provisions except the EIC, they found that 44.6% of joint filers faced marriage penalties, 43.3% received marriage bonuses, and 12.1% had neither penalty nor bonus; the overall effect was a net bonus of $2.0 billion. The EIC, by contrast, produced a net marriage penalty of $3.6 billion. The bulk (about 55%) of EIC marriage penalties fell on couples with combined incomes between $30,000 and $60,000. Of EIC claimants filing joint returns, considering their overall federal income tax situations (not just the EIC), 30.1% had marriage penalties, 28.7% had bonuses, and 41.2% had neither; the net marriage penalty for EIC claimants was $682 million.
More recently, Emily Lin and Patricia Tong have looked at the other side of the coin—the marriage penalties and bonuses that would be experienced by cohabiting couples if they were married. Analyzing data from tax year 2007, they found that 48% of cohabiting couples would have faced marriage penalties of an average of $1,657, 38% would have received bonuses averaging $914, and 15% would have had neither penalty nor bonus. The presence of children was associated with an increased likelihood of potential marriage penalties. Only 38% of childless couples would have faced penalties if married, compared with 52% of couples where one partner had children and 92% of couples where both partners had children. The association of children with would-be marriage penalties was largely due to the EIC. Although 71% of unmarried couples with children claimed the EIC, only 32% of couples with children would have been EIC-eligible if married. For unmarried couples with children, the average avoided EIC marriage penalty was $1,761, which constituted more than 80% of the group’s average overall avoided penalty of $2,175.
Having examined the different balances struck by Congress between marriage penalties and bonuses at different income levels, this Article now considers whether there might be any plausible justification for those differences, and (in Part IV) what should be done if the present differences cannot be defended.
Consider, as a starting point, the explanations Congress and the executive branch have offered over the past few decades for some of the landmarks of marriage-affecting income tax rules. The extreme tilt toward marriage penalties at the high end of the rate structure was introduced by the Omnibus Budget Reconciliation Act of 1993, which added a new top rate of 39.6%, imposed—in the case of both joint returns and single taxpayers—on income above $250,000.
One might have expected some explanation from Congress as to why the new top bracket was designed in this marriage-penalty-maximizing way, but no such explanation was forthcoming. The provision originated with a proposal by the Clinton administration, which offered no explanation at all for the decision to maximize marriage penalties. Without mentioning the difference between joint and individual returns, the administration had simply proposed “an additional 10 percent surtax for those people with taxable income over $250,000, resulting in a 39.6% tax rate for those income levels . . . .” The administration’s proposed high-income marriage penalty attracted only a smattering media and political attention at the time, although it was the focus of a story in the Wall Street Journal. The story noted that a “Treasury Department spokesman didn’t return phone calls” from the Journal inquiring about the marriage penalty effects of the proposal. The relevant publication of the House Ways and Means Committee shed no light on the decision to emphasize marriage penalties for high-income taxpayers. One suspects that the administration (and Congress, following the administration’s lead) simply decided that a quarter of a million dollars worked well, in sound-bite terms, as a threshold for the imposition of the new top rate, and that creating different thresholds for different filing statuses would have been too complicated for sound-bite purposes.
The complete elimination for childless couples of marriage penalties caused by the standard deduction and the 10% and 15% rate brackets was accomplished by the Economic Growth and Tax Relief Reconciliation Act of 2001. In its General Explanation of the 2001 legislation, the Joint Committee on Taxation noted that “Congress was concerned about the inequity that arises when two working single individuals marry and experience a tax increase solely by reason of their marriage.” Moreover, “Congress believed that an increase in the standard deduction for married couples filing a joint return . . . was a responsible reduction of the marriage tax penalty,” and that the expansion of the 15% rate bracket for joint returns “would alleviate the effects of the present-law marriage tax penalty.” As an explanation for the particular policy choices embodied in the 2001 legislation—the skewing toward marriage bonuses, and the elimination of penalties for childless spouses but not for those with children—these conclusory statements are almost laughably inadequate. Again, one suspects the real explanation may have more to do with sound bites than with the policy merits. The 2001 Act enabled legislators to claim Congress was not merely reducing penalties, but actually eliminating them for the typical middle-income, two-earner couple—at least if one’s typical middle-income couple has no children.
What about the opposite skewing—toward marriage penalties rather than bonuses—attributable to the EIC? Before the 2001 legislation, the EIC phaseout thresholds were identical for joint returns and for individual returns, resulting in an extreme tilt toward marriage penalties in the operation of the phaseout. The 2001 Act provided (on a delayed basis) for a joint return phaseout threshold $3,000 higher than the single taxpayer threshold. Even after this increase in the joint return phaseout threshold, the EIC phaseout rules remained heavily tilted in the direction of marriage penalties. The Joint Committee on Taxation’s description of what Congress had in mind fails to elucidate why Congress settled on such limited relief—especially considering that the same Congress had opted for much more extensive marriage penalty relief in the case of the standard deduction and the lower rate brackets. The Joint Committee merely observed that “Congress believed increasing the phase-out amount for married taxpayers who filed a joint return would help to alleviate [the EIC marriage] penalty.” This does nothing to explain why it was appropriate for Congress to eliminate marriage penalties in the standard deduction and lower rate brackets (at least for childless couples), while at the same time only modestly reducing EIC marriage penalties.
In the American Recovery and Reinvestment Tax Act of 2009, Congress provided a modest additional amount of EIC marriage penalty relief, increasing the joint return phaseout threshold to $5,000 more than the single taxpayer threshold. The Joint Committee’s explanation of what Congress had in mind on this occasion was, if anything, even less illuminating that its explanation of the motivation for the 2001 EIC marriage penalty relief. One suspects the real explanation for the skewing of the EIC toward marriage penalties can be found in a combination of early legislative inattention and later-path dependence. The pre-2001 identical phaseout thresholds for married and single taxpayers appear to have been based on a legislative failure even to think about EIC marriage impacts, rather than on a considered decision to maximize penalties and avoid bonuses. In 2001, when Congress finally became aware of the extreme tilt toward penalties and decided to do something about it, Congress naturally defined “doing something” relative to the status quo. Starting from the greatest possible skewing toward penalties in the EIC phaseout rules, Congress could accurately claim to have done something, even as the system remained heavily skewed toward penalties after the enactment of marriage penalty relief. More extensive EIC penalty relief either would have been quite costly to the fisc, or would have required reducing EIC benefits for unmarried recipients (thus making joint return EIC benefits more attractive by comparison, without making them more attractive in absolute terms). Apparently willing to spend only so much on the EIC and apparently unwilling to reduce existing benefits for unmarried credit recipients, but clearly wanting to do something about EIC marriage penalties, Congress settled on EIC phaseout rules still skewed toward penalties (albeit less so than before enactment of the relief).
The very different, pro-marriage, tilt of the standard deduction and lower brackets is best explained by the different status quo in those areas as of 2001 (when Congress made marriage penalty relief a major legislative theme). Having decided to do something about marriage penalties in the standard deduction and the lower brackets, Congress started from a status quo reflecting a compromise between penalties and bonuses, and converted it to a system severely tilted toward bonuses. Having decided at the same time to do something about marriage penalties in the phaseout of the EIC, Congress started with a penalties-only system and converted it into a mostly-penalties regime.
In short, consideration of official explanations and reasonable speculations about actual legislative motivations do not (to put it mildly) inspire confidence that Congress has made informed and thoughtful decisions to favor marriage penalties at lower income levels, to favor bonuses in the middle income ranges, and to favor penalties (again) at the top of the income distribution. Of course, it is possible that the current pattern happens to make good policy sense, despite the fact that Congress stumbled into it. But does it?
Consider first the lower income ranges. When it fundamentally reformed welfare in 1996, Congress set forth at the very beginning of its landmark legislation its findings that “[m]arriage is the foundation of a successful society,” and that “[m]arriage is an essential institution of a successful society which promotes the interests of children.” Congress also noted a number of “well documented” “negative consequences of an out-of-wedlock birth on the mother, the child, the family, and society . . . .” There is, in fact, a considerable amount of evidence that marriage is associated with a wide range of better outcomes (for example, in terms of health, education, and finances) than cohabitation, for both parents and their children.
As is suggested by the fact that Congress chose to extol the benefits of marriage in the context of welfare reform legislation, the public policy reasons to favor marriage over cohabitation would seem to be particularly acute at the lower end of the income distribution. If our major concern is for the well-being of children, it is notable that cohabiting couples with children tend to be less educated and poorer than childless cohabiting couples. Of cohabiting couples (ages 30 to 44) without college degrees, 67% live with children. By comparison, of cohabiting couples in the same age range with college degrees—and with roughly twice the average income of the non-degree couples—only 33% lived with children. All this suggests that Congress should be particularly concerned that the tax system reward marriage—or at least not penalize it—in the case of lower-income couples with children, at least if the primary legislative concern is the well-being of children. And yet this group faces tax marriage penalties that are large in absolute dollar terms, and by far the largest penalties facing any demographic slice in percentage-of-income terms.
To be sure, both quantitative and qualitative studies suggest that lower-income couples are not much influenced by tax penalties (or bonuses) in deciding between marriage and cohabitation. Quantitative studies examining the influence of marriage penalties and bonuses on couples throughout the income distribution have concluded either that penalties and bonuses have no effect on marriage rates, or that the effects are quite small. Quantitative studies focused on lower-income couples similarly find no tax effects on marriage rates, or only small effects. In qualitative, interview-based studies, lower-income couples almost never tell researchers that tax considerations have influenced their choice between marriage and cohabitation.
On the other hand, it is certainly possible that some couples’ decisions are influenced by tax penalties and bonuses, despite the couples’ lack of awareness of that influence. As Adam Carasso and Eugene Steuerle have commented, “People may react to incentives even when they do not calculate them, as when partners choose to cohabit . . . because they simply observe that unmarried couples have a higher standard of living than those who marry . . . .” Moreover, the most recent quantitative study, by Hayley Fisher, finds larger tax effects on marriage decisions than the earlier quantitative studies, especially for low-income couples. Fisher calculates that a $1,000 increase in the marriage penalty is associated with a 1.7 percentage point decrease in the probability of marriage (across the entire income distribution), and that the greatest responsiveness is among persons with the least education; in the lowest educational group a $1,000 penalty increase is associated with a 2.7 percentage point decrease in the probability of marriage.
For two reasons, the studies finding that marital decisions of lower-income couples are not very responsive to tax penalties (and bonuses) are of little help to anyone attempting to defend the legislative tilt toward marriage penalties. First, as suggested by the observation of Carasso and Steuerle and by the findings of Fisher, the jury is still out on the extent of the behavioral effects. Second, if the behavioral effects do turn out to be very small, that merely shifts the basis for the objection from concerns about behavioral effects to concerns about fairness. If a lower-income couple chooses marriage over cohabitation, and thereby suffers a marriage penalty of several thousand dollars, it must be either because the partners are unaware of the tax penalty or because the partners believe so strongly in marriage that they are willing to suffer the penalty as a price of marriage. In the former case the ignorance is non-culpable, given the opaqueness to most taxpayers of the inner workings of the income tax. In the latter case, the partners’ attitude of tax-penalties-be-damned verges on heroism. In neither case does it seem fair for the federal government to adopt a tax system tilted toward penalizing couples choosing marriage. In short, there is no easy or obvious way to construct an argument for tilting toward marriage penalties for lower-income parents; it would be considerably easier to make the case for a tilt in the opposite direction.
What about the severe skewing toward marriage bonuses in the case of couples with incomes in the high-five-figure range? It is not obvious why the income tax rules should be at their most marriage-friendly in this income range. An argument in defense of the tilt might be based on the benefits to children of having married (rather than cohabitating) parents. From that perspective, however, it is strange that the income tax rules in this income range are more pro-marriage for childless couples than for couples with children. Even limiting the analysis to the tax treatment of couples with children, it is mystifying (on the merits) why Congress would lean toward subsidizing marriage for middle-income couples while leaning in the opposite direction for lower-income couples. If Congress is concerned about the well-being of children living with unmarried cohabiting parents, that concern should be greater in the case of children in families with incomes in the $30,000 to $50,000 range than in the case of children in families with incomes in the $75,000 to $100,000 range—both because such children in the lower-income range outnumber such children in the higher range by more than five to one, and because the children in the lower-income families are more economically vulnerable. Yet the pattern of marriage penalties and bonuses—skewed toward penalties at $30,000 to $50,000 and toward bonuses at $75,000 to $100,000—suggests Congress is, for some strange reason, more solicitous of the well-being of the children in the higher-income families. Whatever may be the appropriate balance between marriage penalties and bonuses for middle-income couples, it is difficult to imagine a persuasive justification for tilting toward bonuses at that income level while tilting toward penalties farther down the income distribution. Nor am I aware of any attempted justification—persuasive or otherwise—even having been offered.
Finally, what about the very differently situated couples with incomes at or approaching $1,000,000, for whom marriage penalties predominate? The top rate brackets for joint returns begin at the same or nearly the same income levels as the top rate brackets for unmarried taxpayers; the maximum marriage penalty at $1,000,000 combined income is very large in absolute terms—more than $32,000—and is more than 2.5 times the maximum marriage bonus at that income level. One might weakly defend the current system on the basis that marriage penalties and bonuses at that income level matter very little in terms of children’s well-being. Of all children living with both parents in families with incomes of $100,000 or more (the Census Bureau statistics lump together all six- and seven-figure family incomes), only 0.8% are children of unmarried parents. Even with the skewing toward marriage penalties, very few higher-income parents opt for cohabitation over marriage. Moreover, in that income range most of the 0.8% should survive relatively unscathed despite their parents’ lack of marriage licenses. At that income level, neither parents nor children are likely to suffer undue hardship as a result of a marriage penalty—not even a marriage penalty of $30,000 or more.
All of the above discussion of marriage penalties among the wealthy is in the nature of “it could be worse.” As such, it does not make for a very satisfying defense of the status quo. Even at the high end of the income distribution, a severe tilt toward marriage penalties sends an unfortunate message about how Congress does—or does not—value marriage. Congress has never offered an explanation or defense of this tilt, and it is unlikely that any convincing defense could be developed.
The most important lesson to be gleaned from an examination of the differing tax marriage effects at different points in the income distribution is that Congress should make conscious decisions about the appropriate distributions of penalties and bonuses at various income levels, instead of stumbling into a set of poorly understood and almost-impossible-to-defend results. The current effects are accidents of path dependence and of Congress myopically focusing on one or a few marriage-affecting provisions at a time, rather than thinking about a sensible overall approach to income tax marriage effects.
In this Article I have attempted to demonstrate the need for a legislative rethinking—or perhaps more accurately a first thinking—of the distribution of marriage penalties and bonuses. I have not attempted the next step of prescribing what should be the results of that rethinking. I do, however, tentatively offer a few thoughts on how Congress should approach the rethinking. First, there should be a rebuttable presumption that the balance between penalties and bonuses—however that balance is struck—should be approximately the same at all income levels. This contrasts sharply with the current mix of anti-marriage (at the bottom and top) and pro-marriage (in the middle) policies.
Second, perhaps there should also be a rebuttable presumption that the most equitable approach is to split the difference, by setting the maximum bonus (for a one-earner couple) at each income level equal to the maximum penalty (for an equal-income two-earner couple) at the same income level. A more sophisticated approach to splitting the difference, which would present some data analysis challenges, would be to design the system so that at every income level the dollar amounts of marriage penalties and bonuses actually experienced by affected taxpayers were approximately equal and offsetting. Besides the simple intuitive appeal of splitting the difference, this approach has the virtue of being fair to unmarried taxpayers—the forgotten men and women of tax policy toward marriage. If marriage penalties and bonuses offset, then married people as a group pay the same tax they would pay if they were not married, and single people as a group are neither burdened by having to pay extra tax to make up for a net marriage bonus nor benefitted by a tax reduction caused by a net marriage penalty.
The most likely candidate for the rebutting of both presumptions is a tilting toward bonuses for lower-income parents. This is where the best case can be made for the importance of pro-marriage public policy. This would be, of course, precisely the opposite of the current approach, under which two-income working class spouses are subject to the most severe marriage penalties (in percentage-of-income terms) in the entire system.
Up to this point I have uncritically accepted the legislative commitments to (1) imposing equal tax on equal-income couples regardless of the distribution of incomes within marriages, and (2) progressive marginal rates. However, having considered at length how difficult (or impossible) it is to find a satisfactory distribution of the tax marriage effects that unavoidably follow from those commitments, one might reasonably conclude that it is time to reexamine those commitments.
Marriage penalties and bonuses would both disappear, of course, if the United States abandoned joint returns for married couples, thereby renouncing the goal of equal tax on equal-income married couples. Along with a number of tax commentators, I have long urged exactly that. Anne Alstott has recently offered an especially powerful argument for the end of joint filing, based on the great decline in the societal centrality of marriage from the mid-twentieth century to the present. The unavoidable arbitrariness of marriage penalties and bonuses provides another reason to reconsider our commitment to joint returns.
It is worth noting in this connection that a policy proposal that formally retains joint filing may actually violate the principle of equal tax on equal-income couples. Consider, for example, the possibility of using a two-earner deduction as a technique for reducing or eliminating marriage penalties caused by the phaseout of the EIC. Suppose the maximum amount of the one-child EIC is $4,000, and that for an unmarried taxpayer the credit is reduced by 20% of the amount by which income exceeds $20,000. The standard way of avoiding all marriage penalties in the design of the phaseout would be to set the phaseout threshold for a married couple at $40,000. But that would not only avoid creating a marriage penalty for spouses earning $20,000 each; it would also produce a $4,000 marriage bonus for a one-earner couple with $40,000 income. The two-earner penalty would be avoided, without creating the one-earner bonus as a side effect, if the phaseout threshold were set at $20,000 in all cases (that is, for both single taxpayers and joint returns), but in determining their income for phaseout purposes married couples were allowed a deduction equal to the lesser of $20,000 or the income of the lower-earning spouse. In that case, the $40,000 two-earner couple would be entitled to a $4,000 EIC, while the $40,000 one-earner couple would receive no credit. That may or may not be considered an appropriate result, but note that although this approach does not formally eliminate joint returns, it decisively rejects the principle of equal tax (or transfers) for equal-income married couples. If we are willing to go that far, perhaps we would do better simply to eliminate joint returns.
Marriage penalties and bonuses would also disappear if we gave up on progressive marginal rates. As noted earlier, this has nearly happened already at some income levels under the alternative minimum tax. If one is otherwise committed to progressive marginal rates, the usual—and quite reasonable—response is that the progressivity principle is more important than avoiding tax marriage effects. If a commitment to progressivity implies a commitment to otherwise undesirable marriage penalties and bonuses, then so be it. A progressive marginal rate structure, however, is best understood as a tool for producing progressive average rates, rather than as an end in itself. Might it be possible to produce a desired distribution of progressive average rates without using progressive marginal rates? It is certainly possible to produce progressive average rates through a combination of a universal demogrant (that is, a per-person transfer payment of some specified amount) and a flat-rate tax. If, for example, there is a flat tax (with no exemption) of 20% and everyone is entitled to a demogrant of $10,000, then the average rate at $50,000 is zero, the average rate at $100,000 is 10%, the average rate at $500,000 is 18%, and so on. This system would be progressive despite the absence of progressive marginal rates, and it would feature neither marriage penalties nor marriage bonuses. Married or unmarried, two people would have a net combined tax liability equal to 20% of their combined income minus $20,000.
It may be neither politically possible nor desirable on the merits to adopt a pure version of a flat-tax-plus-demogrant, but it may be possible and desirable to slouch some distance in that direction. We could reduce marginal rate progressivity without reducing average rate progressivity if we introduced small-scale demogrant-type programs as we reduced marginal rate progressivity. If the marginal rate structure remained moderately progressive, there would still be marriage penalties and bonuses, but the size of the tax marriage effects would have been significantly reduced.
I fondly recall my discussions with Bill Turnier years ago about the income tax treatment of marriage. Although I favored (then as now) the elimination of joint returns, Bill’s thoughtful defense of joint filing greatly influenced (and tempered) my thinking on the issue. I appreciate the opportunity afforded me by this well-deserved festschrift in honor of a wonderful friend, colleague, and scholar to revisit the tax treatment of marriage. I hope (and expect) that Bill will not mind that, after having accepted the joint return premise for almost all of this Article, at the end I have smuggled in a bit of an argument for the end of joint filing. I look forward to renewing our friendly debate.
. See, e.g., U.S. Gen. Accounting Office, GAO/GGD-96-175, Tax Administration: Income Tax Treatment of Married and Single Individuals 2 (1996) [hereinafter GAO] (“The trade-offs among these principles exist because, as public finance experts have long recognized, a tax system cannot satisfy all three principles simultaneously.”). For consideration of this trilemma, see, for example, Anne L. Alstott, Updating the Welfare State: Marriage, the Income Tax, and Social Security in the Age of Individualism, 66 Tax L. Rev. 695, 698 (2013).
. See, e.g., GAO, supra note 1, at 2 (“[O]ur current tax system . . . has progressive tax rates and taxes married couples with the same income equally. However, this can result in the income of some married couples falling into different tax brackets than would be the case if they were to file as single individuals. Therefore, the present tax system is not ‘marriage neutral’ . . . .”).
. See Shari Motro, A New “I Do”: Towards a Marriage-Neutral Income Tax, 91 Iowa L. Rev. 1509, 1529 (2006) (“[O]ver the years most critics have regarded the marriage bonus created by income splitting as an undesirable but tolerable side effect . . . .”). But see id. (“[O]utside of academia, the principle of marriage neutrality is not necessarily taken for granted. Some people believe that the government should actively promote marriage, and that the benefits of marriage-based income splitting present an appropriate means of doing so. Though Congress did not create the ‘marriage bonus’ as a marriage-promotion vehicle, many politicians and their constituents have grown to view it and defend it as such.”).
. In a 1996 study, the General Accounting Office identified fifty-nine income tax provisions under which tax liability depends on marital status. GAO, supra note 1, at 3. The number may well be higher today. See Rev. Rul. 2013-17, 2013-38 I.R.B. 201, 202 (“There are more than two hundred Code provisions and Treasury regulations relating to the internal revenue laws that include the terms ‘spouse,’ ‘marriage,’ . . . ‘husband and wife,’ ‘husband,’ and ‘wife.’ ”). Most of those provisions, however, are of limited applicability. See GAO, supra note 1, at 3. This Article focuses on the marriage effects of a small number of basic structural provisions of wide applicability.
. All calculations assume that taxpayers claim the standard deduction rather than itemizing deductions. Compare I.R.C. § 63(c) (standard deduction), with I.R.C. §§ 161–199, 211–224 (itemized deductions). The assumption is rather unrealistic at higher income levels, but it does not dramatically affect the maximum bonus and penalty calculations. See Zelenak, supra note 3, at 56 & nn.220–21 (showing that the popularity of the standard deduction generally declines as income increases). For unmarried one-earner couples, the calculations assume that the earning partner can and does claim the non-earning partner as a dependent. I.R.C. § 151(b).
. If there were a child, then one of the two-earner unmarried taxpayers could file as a head of household, I.R.C. § 2(b), which would create a small maximum marriage penalty at this combined income level.
. Both the child tax credit and dependency exemptions are fully phased out at this income level, I.R.C. §§ 24(b) (child tax credit), 151(d)(3) (dependency exemptions), so the presence or absence of children would be relevant only with respect to the availability of head-of-household filing status in the unmarried case.
. See generally Jane G. Gravelle, Cong. Research Serv., 98-653 E, The Marriage Penalty and Other Family Tax Issues 2 (1998) (“In the case of the EITC, penalties can occur because larger aggregated incomes cause loss of the credit through income phase-outs.”).
. See generally Gravelle, supra note 20, at 2 (“The effect of marriage on tax liability was exacerbated by the higher [AMT] rates introduced in 1993 at the top of the income distribution . . . .”).
. See generally Lawrence Zelenak, Marriage and the Income Tax, 67 S. Cal. L. Rev. 339, 341 (1994) (“The more progressive the rate structure, the more significant the tax effects of joint returns.”).
. Cohabitators represented only 1.1% of all couples in 1960, compared with 7.0% in 1997. Tom W. Smith, Ties That Bind: The Emerging 21st Century American Family, Pub. Persp., Jan.–Feb. 2001, at 34, 35.
The Roper Center’s extensive iPOLL Databank contains no polling on attitudes toward cohabitation from before the 1970s. The lack of pre-1970s polling data is itself quite telling. Apparently cohabitation was so obviously socially unacceptable that pollsters saw no reason to ask about it. In a 1977 survey, only 26% of respondents thought it was “okay” that there had been “an increase in the number of couples living together without being married”; 48% thought it was “wrong” and 23% thought it did not “matter much” or had no opinion. CBS News/N.Y. Times, Oct. 1977, available at http://www.ropercenter.uconn.edu/CFIDE/cf/action/home/index.cfm (enter dataset keyword: USCBSNYT.112677.R39). In the 1994 General Social Survey of the National Opinion Research Center, 41% of respondents disagreed or strongly disagreed with the statement, “It is alright for a couple to live together without intending to get married.” Nat’l Op. Research Ctr., Univ. of Chi., Jan. 1994, available at http://www.ropercenter.uconn.edu/CFIDE/cf/action/home/index.cfm (enter dataset keyword: USNORC.GSS94.Q622B). When the same question was asked in 2012, only 29% disagreed or strongly disagreed with the statement. Nat’l Op. Research Ctr., Univ. of Chi., Mar. 2012, available at http://www.ropercenter.uconn.edu/CFIDE/cf/action/home/index.cfm (enter dataset keyword: USNORC.GSS12C.Q1106B); see Tom W. Smith et al., Nat’l Op. Research Ctr., Univ. of Chi., General Social Surveys, 1972–2012: Cumulative Codebook 2277 (2013), available at http://publicdata.norc.org/GSS/DOCUMENTS/BOOK/GSS_Codebook.pdf (comparing results for the same question from 1972 to 2012).
. Two equal earners might have noticed the marriage penalty if they went from being over-withheld as singles to being under-withheld as spouses, despite the fact that neither their incomes nor their withholding had changed. Thanks to Kathleen Thomas for this point.
. See, e.g., Staff of J. Comm. on Taxation, 91st Cong., General Explanation of the Tax Reform Act of 1969, at 223 (Comm. Print 1970) (defending marriage penalties as appropriate legislative responses to marital economies of living).
. In 1987, only 10% of American women ages 19 to 44 were living with unmarried partners; by 2009–2010, that number had more than doubled to 23%. Wendy D. Manning, Nat’l Ctr. for Family & Marriage Research, Trends in Cohabitation: Over Twenty Years of Change, 1987–2010, at fig.3 (2013), available at https://www.bgsu.edu/content/dam/BGSU/college-of-arts-and-sciences/NCFMR/documents/FP/FP-13-12.pdf. For data on attitudinal changes, as reflected in public opinion polling, see the sources cited supra note 28.
. See Jonathan Vespa, Jamie M. Lewis & Rose M. Kreider, U.S. Census Bureau, Dep’t of Commerce, America’s Families and Living Arrangements: 2012, at 18–19 (2013) (reporting that “37 percent of male spouses and 16 percent of female spouses earned at least $50,000”; the comparable figures for male and female unmarried cohabiting partners were 21% and 12%).
. Although an earning partner can claim a non-earning cohabiting partner as a dependent, I.R.C. § 152(d)(2)(H), the non-earning partner cannot serve as the basis for head-of-household filing status for the earning partner, id. § 2(a)(1)(B).
. See Jackson v. Comm’r, 7154 T.C.M. (CCH) 2022, 2024 (1996) (denying head-of-household status to an unmarried father, who claimed to live with his daughter in one room of a two-bedroom house owned by the child’s mother: “We find that the one room allegedly lived in by petitioner and [the daughter] . . . , without use of a kitchen or telephone, does not constitute a separate household”).
. I.R.C. § 152(c)(4)(B)(ii) provides a tiebreaker rule for the EIC. If both unmarried parents claim the same child for EIC purposes, and if the child resides with both parents for the same amount of time during the year, then the child goes to the parent with the higher adjusted gross income. However, this rule applies only to resolve conflicts resulting from both parents filing returns claiming the same child. If the parents amicably divide their children for EIC purposes, with each parent claiming a different child on his or her return, the tiebreaker rule does not apply. A previous version of the EIC tiebreaker rules did not depend on conflicting claims as a trigger. Under that version, all children were allocated to the higher-income parent (assuming each parent otherwise qualified to claim the children), without regard to whether conflicting claims to the children had been made on the parents’ tax returns. Sutherland v. Comm’r, 81 T.C.M. (CCH) 1001, 1005 (2001).
. The two-child EIC phaseout rules reduce the credit by 21.06 cents for every dollar by which adjusted gross income exceeds the threshold amount. I.R.C § 32(b)(1)(A). The maximum two-child credit of $5,460 is reduced by $3,525, which is 21.06% of the amount by which $40,000 adjusted gross income exceeds the $23,260 phaseout threshold.
. The joint return liability, after reduction by the child tax credit, is $7,955. If the partners are unmarried, the partner filing as head of household (and claiming the child tax credit and the dependency exemption) will owe $2,552.50 and the other partner will owe $4,773.75, for a total of $7,326.25.
. The availability of head-of-household filing status, I.R.C. § 2(b), for an unmarried partner with a child remains a source of marriage penalties at this income level, but the penalties thus created are very small in percentage-of-income terms.
. In 2014, the 35% bracket begins at $405,100 for both joint returns and single taxpayers, while the 39.6% bracket begins at $406,750 for single taxpayers and at $457,600 for joint returns. Rev. Proc. 2013-35, 2013-47 I.R.B. 537, 538–39.
. Suppose, for example, Congress “split the difference” by setting the joint return breakpoint at $273,750. Then the maximum penalty and the maximum bonus would each be $91,250 x (28% – 26%) = $ 1,825, and the sum of the absolute values would be $3,650.
. Patient Protection and Affordable Care Act, Pub. L. No. 111-148, 124 Stat. 119 (2010) (codified as amended in scattered sections of 42 U.S.C.); Health Care and Education Reconciliation Act of 2010, Pub. L. No. 111-152, 124 Stat. 1029 (codified as amended in scattered sections of 42 U.S.C).
. See id. § 151(d)(3)(A) (“In the case of any taxpayer whose adjusted gross income for the taxable year exceeds the applicable amount in effect under section 68(b), the exemption amount shall be reduced by the applicable percentage.”).
. Id. Note that limiting the analysis to married couples claiming the EIC misses the marriage penalties on couples whose EICs were fully phased out under the joint return rules, but who would have been able to claim EICs if not married.
. See generally Emily Y. Lin & Patricia K. Tong, Marriage and Taxes: What Can We Learn from Tax Returns Filed by Cohabiting Couples?, 65 Nat’l Tax J. 807 (2012) (examining tax returns of cohabiting couples).
. See William J. Clinton, A Vision of Change for America 103 (1993) (discussing the tax code change). Clinton’s proposal was enacted in section 13202(a) of the Omnibus Budget Reconciliation Act of 1993. Pub. L. No. 103-66, § 13202(a), 107 Stat. 312, 461 (codified at I.R.C. § 1(a)–(e) (2012)).
. The skewing toward bonuses is a result of the standard deduction, and the 10% and 15% rate brackets, being made twice as large for joint returns as for unmarried taxpayers (other than heads of households). Id. at 26–27. The failure to eliminate all marriage penalties for spouses with children results from the joint return standard deduction not being as large as the combined standard deductions for a single taxpayer and a head of household (and similarly for the 10% and 15% brackets). Id. at 27.
. Even if the $3,000 increase in the joint return phase-out threshold had been immediately effective in 2001, the resulting joint return threshold ($16,090) would have been only 23% higher than the single taxpayer threshold ($13,090). For the 2001 EIC phase-out thresholds, see Rev. Proc. 2001-13, 2001-1 C.B. 337, 339, at § 3.03.
. Staff of J. Comm. on Taxation, 111th Cong., General Explanation of Tax Legis. Enacted in the 111th Cong. 21–22 (J. Comm. 2011) (providing reasons for increasing the generosity of the tax credit for families with three or more children, but providing no reason for the additional marriage penalty relief).
. See, e.g., Linda J. Waite & Maggie Gallagher, The Case for Marriage: Why Married People Are Happier, Healthier, and Better Off Financially 36–46 (2000); Sheela Kennedy & Larry Bumpass, Cohabitation and Children’s Living Arrangements: New Estimates from the United States, 19 Demographic Res. 1663, 1665 (2008) (citing a number of studies in support of the proposition that “[c]hildren raised by a cohabiting parent appear to have poorer outcomes than the children of married parents, across a range of indicators, including academic performance, emotional problems and depression, and behavioral problems and delinquency”).
. David L. Sjoquist & Mary Beth Walker, The Marriage Tax and the Rate and Timing of Marriage, 48 Nat’l Tax J. 547, 556 (1995) (finding no tax effect on marriage rates, although finding some tax-motivated shifting of marriage timing between the end of one year and the beginning of the next).
. James Alm & Leslie A. Whittington, Does the Income Tax Affect Marital Decisions?, 48 Nat’l Tax J. 565, 571 (1995) (“[T]axes affect at least some marital decisions of at least some individuals, but . . . for many individuals taxes are largely irrelevant.”); James Alm & Leslie A. Whittington, Shacking Up or Shelling Out: Income Taxes, Marriage, and Cohabitation, 1 Rev. Econ. Household 169, 182–84 (2003) (finding that cohabiting couples are less likely to marry when marriage would increase their tax liability; the effect is small but statistically significant).
. Stacy Dickert-Conlin & Scott Houser, EITC and Marriage, 55 Nat’l Tax J. 25, 37 (2002) (concluding that marriage penalties and bonuses created by “the EITC expansions during the early- to mid-1990s had little or no effect on the marriage decision . . .”); David T. Ellwood, The Impact of the Earned Income Tax Credit and Social Policy Reforms on Work, Marriage, and Living Arrangements, 53 Nat’l Tax J. 1063, 1099 (2000) (finding some indications that “EITC incentives may influence cohabitation versus marriage decisions”).
. Kathryn Edin & Joanne M. Reed, Why Don’t They Just Get Married? Barriers to Marriage Among the Disadvantaged, 15 Future of Child. 117, 128 (2005) (“[H]ardly any of the mothers or fathers in these studies named . . . the potential loss of the EITC (which poor unmarried parents typically refer to as their ‘tax return’) as a barrier to marriage.”); see Christina M. Gibson-Davis, Kathryn Edin & Sara McLanahan, High Hopes but Even Higher Expectations: The Retreat from Marriage Among Low-Income Couples, 67 J. Marriage & Fam. 1301, 1307–08 (2005) (finding that “the overwhelming majority [of those surveyed] indicated that they needed to get their finances in order before they could get married”); Pamela J. Smock, Wendy D. Manning & Meredith Porter, “Everything’s There Except Money”: How Money Shapes Decisions to Marry Among Cohabitors, 67 J. Marriage & Fam. 680, 686–92 (2005) (reporting that many low-income couples in interviews cited financial concerns as obstacles to marriage, but not reporting that any couples mentioned tax marriage penalties).
. Hayley Fisher, The Effect of Marriage Tax Penalties and Subsidies on Marital Status, 34 Fiscal Stud. 437, 439, 453 (2013) (estimating that “a 1 per cent increase in household income just for being married increases the probability of being married by 1.1 per cent. . . . These estimates are four times larger than those found in the previous literature”).
. See Lawrence Zelenak, Complex Tax Legislation in the TurboTax Era, 1 Colum. J. Tax L. 91, 93, 118–19 (2010) (arguing that complex tax provisions “turn the income tax into a black box, the inner workings of which are incomprehensible to the average taxpayer, thereby undermining . . . the ability of taxpayers to engage in informed tax planning”).
. Vespa et al., supra note 33, at 25 tbl.10 (indicating that 609,000 children were living with unmarried cohabiting parents in the $30,000 to $50,000 family income range, compared with 111,000 children living with unmarried cohabiting parents in the $75,000 to $100,000 family income range).
. Vespa et al., supra note 33, at 25 tbl.10. In that family income range there are 134,000 children living with unmarried cohabiting parents, and 16,836,000 children living with married parents. Id.
. See id. (stating that only 134,000 children live in cohabitating households with incomes over $100,000, while 16,836,000 children in the same income range live in households with married parents).
. See generally Pamela B. Gann, Abandoning Marital Status As a Factor in Allocating Income Tax Burdens, 59 Tex. L. Rev. 1 (1980) (“[A] marriage-neutral income tax system, under which all individuals file separate returns under a single rate schedule, is the most defensible position in the long run.”); Marjorie E. Kornhauser, Love, Money, and the IRS: Family, Income-Sharing, and the Joint Income Tax Return, 45 Hastings L.J. 63 (1993) (“The joint return ought to be abolished. A system that treats each person as a separate taxable unit is more equitable, more consistent with basic tax principles, more efficient, and ultimately better able to accomplish social family goals.”); Zelenak, supra note 26 (“The only way to avoid both marriage bonuses and penalties is to abandon marital status as a tax determinant and to require that spouses file separate returns.”).
Joint filing based on formal marriage is particularly ill-suited to the new patterns of marriage and child-rearing. In the mid-twentieth century, the prevalence, homogeneity, and exclusivity of formal marriage made it a convenient and perfectly sound proxy for family. Today, however, joint filing is not a plausible way of attempting to protect freedom or promote collective welfare.
. From 1981 to 1986 the law permitted a two-earner deduction equal to 10% of the earned income of the lower-earning spouse, up to a maximum deduction of $3,000. See I.R.C. § 221 (repealed 1986). The proposal discussed in the text would be similar, but would apply only for purposes of the EIC phaseout.
. See, e.g., I.R.C. § 32(b)(a)(A) (2012) (providing a phaseout threshold of 15.98% for an eligible individual with one qualifying child). As explained in the Article’s Introduction, the usual method of avoiding marriage penalties is to provide tax parameters (e.g., standard deduction amounts and bracket widths) twice as large for joint returns as for the returns of unmarried taxpayers.
. For this approach to be revenue neutral and distributionally neutral (assuming those neutralities are desired), the decrease in marginal rate progressivity would have to be accomplished by increasing rates in the lower brackets rather than by decreasing rates in the upper brackets.