Income Phase-Out

An income phase-out is a statutory range of income over which a tax deduction, credit, exclusion or eligibility test is reduced to zero rather than ending abruptly. It is the drafting device Congress uses to aim a benefit at a target group without creating a hard edge: below the range the benefit is whole, above it the benefit is gone, and inside it the benefit shrinks by a formula written into that provision. A cliff is the alternative design, where the benefit survives intact up to a single figure and disappears one dollar later. The taxonomy of the two, and the fact that adjusted gross income is the yardstick most of them start from, belongs to adjusted gross income; what this page covers is how a phase-out is actually built and what it does to the cost of your next dollar of income.

Three shapes, and almost everything is one of them. The first is a stated dollar reduction for each increment of excess income. The qualified tips deduction and the qualified overtime compensation deduction are both reduced by $100 for each $1,000 of modified adjusted gross income above $150,000 ($300,000 on a joint return), under Internal Revenue Code sections 224(b)(2)(A) and 225(b)(2)(A). The child tax credit uses the same shape at a different rate, $50 for each $1,000 of excess under section 24(b)(1). The second shape is a flat percentage of the excess: the enhanced deduction for seniors is reduced by 6 percent of modified adjusted gross income above $75,000 ($150,000 joint), under section 151(d)(5)(C)(iii). The third is proration across a band of stated width. The American Opportunity and Lifetime Learning credits are reduced by the ratio the excess bears to $10,000 ($20,000 joint) under section 25A(d)(1), and the traditional IRA deduction is prorated across its own band under section 219(g)(2). The IRA ranges and their two mechanical quirks live on the IRA income phase-out page rather than here.

Rounding is a drafting choice, not a convention, and one IRS form proves it. Schedule 1-A (Form 1040), which carries four of the deductions created in 2025, tells you to divide the excess income by $1,000 and then, for the tips and overtime parts, to "decrease the result to the next lower whole number." In Part IV of the same form, the car loan interest deduction tells you to "increase the result to the next higher whole number." One rounds in the taxpayer's favor and the other does not. Elsewhere the same effect is achieved by the phrase "or fraction thereof," which appears in section 24(b)(1) and rounds a partial increment up. There is no default: read the provision.

Inside a band your marginal rate is not your bracket rate. An extra dollar of income inside a phase-out does two things at once. It is taxed, and it shrinks the benefit. Where the benefit is a deduction, the extra cost is the withdrawal rate multiplied by your bracket rate, because a smaller deduction means more taxable income rather than more tax directly. Where the benefit is a credit, the withdrawal lands on the tax bill dollar for dollar, so the same withdrawal rate hurts several times as much. This is the mechanism behind the gap between a bracket rate and a true marginal tax rate, and it is why two households in the same bracket can face very different costs on the same raise.

Each phase-out measures its own income figure. Very few run on plain adjusted gross income. The tips, overtime and senior phase-outs each define modified adjusted gross income as adjusted gross income increased by amounts excluded under sections 911, 931 and 933, and each writes that definition out separately in its own subsection rather than cross-referencing a shared one. Other provisions add back entirely different items. Modified adjusted gross income is a label rather than a number, so the practical instruction is to find the definition inside the section you are reading and not to reuse a figure you computed for a different rule.

Indexation is provision by provision too. Sections 224 and 225 contain no cost-of-living clause anywhere, so the $150,000 and $300,000 thresholds and the $25,000 and $12,500 caps are fixed in nominal dollars for the whole 2025 through 2028 life of those deductions. An unindexed threshold tightens every year as wages rise, which reduces a benefit quietly and without any vote. The last thing worth knowing is that phase-outs stack: a single raise can run through several of them at once, and the withdrawal rates add.

Every phase-out runs the same four steps, with the details supplied by the provision.

1. Find the income figure that provision uses, which usually means computing a modified adjusted gross income defined inside the section itself.
2. Subtract the threshold. If the result is zero or less, there is no reduction and the benefit is whole.
3. Apply the reduction formula to the excess, using the provision's own rate and its own rounding rule.
4. Subtract the reduction from the benefit, and stop at zero. A phase-out never produces a negative benefit.

A hypothetical example, using the qualified tips deduction because its arithmetic is the most visible. Dana is single, has $28,000 of qualified tips, and modified adjusted gross income of $166,400. Her excess over the $150,000 threshold is $16,400. Dividing by $1,000 gives 16.4, which the form decreases to 16. Sixteen increments at $100 each is a $1,600 reduction. Her deduction is capped at $25,000 before the phase-out, so she deducts $25,000 minus $1,600, or $23,400.

The same figures show what her next dollars cost. Inside the band, each additional $1,000 of income removes $100 of deduction. If Dana's marginal rate is 22 percent, that lost deduction adds $22 of tax on top of the $220 the $1,000 already owed, so $1,000 of extra income costs her $242. Her effective marginal rate is 24.2 percent while she is inside the band, and it drops back to 22 percent once the deduction is fully gone.

Pros

• Targets a benefit at a chosen income group without a cliff, so no single dollar of income can cost a household the whole benefit.
• Keeps the decision to earn more straightforward, because the withdrawal takes a fraction of each extra dollar rather than all of it.
• Makes the cost of a benefit predictable for Congress, which is why the device is used so heavily.

Cons

• Creates marginal rates that appear nowhere in the rate tables, so a household can face a materially higher cost on a raise than its bracket implies.
• Multiplies the number of income definitions a filer has to compute, since each provision writes its own.
• Thresholds that carry no inflation clause shrink the benefit every year without any change in the law.
• Stacks silently: several phase-outs can run on the same dollar, and nothing on a return shows the combined withdrawal rate.