Marriage Penalty

A marriage penalty is the amount by which a couple's federal income tax on a joint return exceeds what the two spouses would have paid had they not been married. The authoritative definition is Treasury's, from an Office of Tax Analysis paper by Bull, Holtzblatt, Nunns and Rebelein: "A couple has a marriage penalty if they owe more income tax filing a joint return than the spouses would pay if they were unmarried and each were taxable as a single or head of household filer. Conversely, a couple has a marriage bonus if they owe less income tax filing a joint return than the spouses would pay if they were unmarried and each were taxable as a single or head of household filer."

Two things about that definition are load-bearing. The comparison baseline is "single or head of household," not two single filers, and using two singles systematically understates penalties for couples with children, because a parent filing alone could often use head of household. And penalty and bonus are one measurement read from opposite ends, which is why they get separate pages rather than one: the same rate structure produces both, and which one a particular couple faces depends on how their incomes are distributed rather than on how large they are.

Neither phrase is a defined term in the Internal Revenue Code, though Congress has used the first as a legislative label: section 1(f) is headed "Phaseout of marriage penalty in 15-percent bracket; adjustments in tax tables so that inflation will not result in tax increases," and section 1(f)(8) is headed "Elimination of marriage penalty in 15-percent bracket." Both are now largely dormant for individuals, because section 1(j)(3)(B)(iii) provides that "subsection (f)(8) shall not apply" for years under the current rate structure. The bodies that use the term analytically are Treasury's Office of Tax Analysis, the Congressional Budget Office and the Joint Committee on Taxation.

The rate tables are not where to look, and this is the change that most writing on the subject has not absorbed. Treasury's 1999 paper explains the cause as follows: "Marriage penalties generally arise because the standard deduction and rate brackets for joint filers are less than twice the corresponding amounts for single filers or head of household filers." That was accurate when it was written. It is no longer accurate about the standard deduction or the lower rate bands.

Measured against the 2026 tables as published in Internal Revenue Bulletin 2025-45, the joint boundaries at the top of the 10, 12, 22, 24 and 32 percent bands are each exactly twice the corresponding unmarried boundary, and the joint standard deduction is exactly twice the unmarried one. The standard deduction case is structural rather than coincidental: section 63(c)(2)(A) sets the joint basic standard deduction at "200 percent of the dollar amount in effect under subparagraph (C)," so it cannot diverge without Congress rewriting the paragraph. The doubling breaks at one place in the rate tables, the boundary where the 37 percent band begins, and the analysis of that break belongs to the page on married filing jointly, which works it through with the statutory base amounts.

So the honest headline is that the marriage penalty has become an un-doubled-provision phenomenon rather than a rate-table one. Treasury's paper anticipated this in its next sentence, which has aged better than the one before it: "Marriage penalties and bonuses can also arise because of other tax provisions, such as the Earned Income Tax Credit (EITC) and the taxation of Social Security benefits."

Where the penalty actually lives. Each row below was read in the section's own text. Every provision marked as not indexed was checked by searching the whole section for inflation-adjustment language and finding none.

A note on the earned income credit, because its shape is different from the others. Section 32(b)(2)(B) does not set a separate joint figure at all. It provides that "in the case of a joint return … the phaseout amount determined under subparagraph (A) shall be increased by $5,000." A flat dollar add-on, indexed, rather than a proportional adjustment. For a couple where both work and both earn modestly, that is the largest single marriage penalty available in the code, because the credit phases out against combined income while the phase-out point rises by a fixed amount.

Several of these rows carry a second problem on top of the first. Sections 1211(b), 1411(b), 86(c) and 3101(b)(2) contain no inflation-adjustment mechanism at all, so those thresholds do not merely fail to double, they also fall in real terms every year. Their absence from the list of sections that the 2017 tax act moved onto chained CPI is independent confirmation: that list, in the act's own effective-date note, names dozens of sections and not these, because they were never indexed in the first place. Each of them states its own non-indexing on its own page. Section 121(b), which does double, is also unindexed, so the residence exclusion erodes for everyone rather than penalising couples.

The worst case in the code is not a doubling failure. Section 86(c)(1)(C) sets the Social Security base amount at "zero in the case of a taxpayer who— (i) is married as of the close of the taxable year (within the meaning of section 7703) but does not file a joint return for such year, and (ii) does not live apart from his spouse at all times during the taxable year," and section 86(c)(2)(C) applies the same zero to the adjusted base amount. For someone in that position up to 85 percent of Social Security benefits can be taxable from the first dollar of other income. It is the sharpest, most actionable fact in this area, and it means that filing separately as a way of escaping a joint-return penalty can be far more expensive than the penalty it was meant to avoid. The mechanics of benefit taxation are covered on the page for provisional income.

One structural oddity, worth a sentence because it runs the other way. Marriage neither merges nor doubles anything in the gift tax. The instructions to Form 709 state that "spouses may not file a joint gift tax return. Each individual is responsible to file a Form 709." A couple gets two annual exclusions and two lifetime exclusions, but no joint return and no combined computation.

Measuring a marriage penalty means computing a couple's actual joint liability and comparing it against a hypothetical: what each spouse would owe filing as a single or head of household filer, with income, deductions and dependents allocated between them. The difference is a penalty if the joint figure is larger and a bonus if it is smaller. Both spouses' circumstances matter, which is why no single number can be quoted for "the marriage penalty."

Three patterns predict which side of the line a couple falls on, and the first is the strongest.

Two similar, high incomes tend toward a penalty. Doubling the bands means each spouse's income no longer benefits from the other's unused lower brackets, and the un-doubled provisions above start binding. Two earners at similar levels get the worst of both.

Two similar, modest incomes can face a penalty too, through the credits. The earned income credit's phase-out is the mechanism: combined income runs the credit down while the phase-out point rises by only a flat statutory amount.

Very unequal incomes tend toward a bonus. The lower earner's unused bands absorb part of the higher earner's income. That case is the subject of the marriage bonus page.

A hypothetical, using the provision where the arithmetic is cleanest. Suppose Rosalind and Dev each have $180,000 of wages. Unmarried, neither reaches the additional Medicare tax, whose threshold is $200,000 for an unmarried filer under section 3101(b)(2), so neither pays a cent of it. Married and filing jointly, their combined wages are $360,000 against a joint threshold of $250,000, so $110,000 is exposed to the 0.9 percent tax: $990 a year that exists solely because they married. Nothing about their work, their income or their spending changed.

Run the same couple through the net investment income tax and the shape repeats, because section 1411(b) uses the same pair of thresholds. Run them through the rate tables and, at these income levels, nothing happens at all: the joint bands are exactly twice the unmarried bands through the top of the 32 percent band, so the ordinary income tax is unmoved. That contrast is the whole point of this page. The penalty came from two provisions that were never doubled, not from the brackets everyone assumes are the cause.

What can be done about it, honestly. Not much, and the obvious move usually backfires. Filing separately does not undo it: the separate thresholds are generally half the joint ones, a long list of credits and deductions is reduced or unavailable, and section 86(c)(1)(C) can zero out the Social Security base amount entirely for a couple who live together. Where the penalty runs through the earned income credit or a phase-out, the levers that exist are the ordinary ones that reduce adjusted gross income, such as retirement plan contributions and health savings account contributions, applied with the specific threshold in view rather than in general.

A marriage penalty is an arithmetic outcome of a rate structure rather than a product, so what follows is what is genuinely better and worse about how current law handles it.

What current law gets right

• The rate bands and the standard deduction are doubled for joint filers well up the schedule, so the ordinary case that dominated policy debate for fifty years has largely been solved.
• The standard deduction doubling is written as a formula in section 63(c)(2)(A) rather than as a pair of numbers, so it cannot silently drift apart.
• Where a penalty exists, it is usually traceable to one or two identifiable provisions rather than being diffuse, which makes it measurable in advance.
• The mirror outcome is more common than the penalty, and it is the outcome for most couples with unequal incomes.

What it costs, and where the law is hardest to defend

• A scattered list of thresholds treats a couple exactly as one person, so two earners each comfortably below a limit can cross it purely by marrying.
• Four of the worst offenders have no inflation adjustment at all, so those penalties grow every year with no vote taken.
• The earned income credit's joint adjustment is a flat dollar add-on, which means the penalty falls hardest on two modest earners rather than on high incomes.
• Filing separately, the intuitive escape, generally makes things worse and can be catastrophic for a couple drawing Social Security who live together.
• Because the correct comparison uses head of household where available, couples with children face larger penalties than a "two singles" comparison suggests, and most popular coverage uses the wrong baseline.
• There is no election, no form and no planning technique that removes it. The only lever is the ordinary one of managing income against a specific threshold.