Not Your Father’s Marriage Penalty
The Impact of AGI-Linked Programs on Household Incentives
David Bernstein Author and Consultant | Bernstein.book1958@gmail.com | 202-413-5492
Abstract
Federal income tax reform has largely mitigated traditional statutory marriage penalties for middle-income households. However, a new generation of “implicit” penalties has emerged. By linking health insurance subsidies and student loan repayments to Adjusted Gross Income (AGI), current policy creates steep marginal burdens. This paper demonstrates that the Affordable Care Act (ACA) premium tax credits and the 2025 Repayment Assistance Plan (RAP) create “de facto” tax rates that materially distort incentives for marriage and childbearing. Using three numerical case studies, the analysis shows that the interaction of these programs can produce effective marginal tax rates exceeding 55%, driven by eligibility cliffs and non-marginal rate jumps.
1. Introduction: The New Tax Landscape
For decades, the “marriage penalty” was defined by statutory tax brackets—the risk that a couple’s combined income would push them into a higher tax tier than they faced as individuals. While the 2017 tax reforms and subsequent adjustments aligned brackets for most joint filers, the financial reality for working-age adults has become more complex.
Modern household formation occurs within a “stacked” system where health insurance costs and student debt are indexed to income. Because these AGI-linked programs operate independently of the IRS tax code but rely on the same income definitions, they function as a parallel tax system. Marriage aggregates income for eligibility, while children alter poverty thresholds and benefit formulas. The resulting Implicit Marginal Tax Rate (IMTR) represents the true share of an additional dollar that a household loses to taxes, lost subsidies, and increased mandatory payments.
2. Literature Review: From Statutory to Implicit Penalties
The economic analysis of marriage penalties originates in Gary Becker’s (1973) household framework, which established the family as a single utility-maximizing unit. This framework implies that when tax systems apply progressive rates to joint income, they inadvertently create penalties for dual-earner couples with similar incomes while providing bonuses for those with asymmetric earnings.
Subsequent empirical research, notably by Hausman (1981) and Eissa & Hoynes (1998, 2004), built on this by demonstrating how these marginal rates distort the labor supply of secondary earners.
By the early 2000s, policy reforms reduced explicit statutory marriage penalties for many through bracket alignment and expanded standard deductions. Martin Feldstein (1995, 1999) argued that taxable income is highly responsive to these marginal rates, emphasizing the “deadweight loss” created when tax design alters economic behavior. However, as statutory penalties narrowed, a parallel literature emerged regarding benefit reduction rates.
Robert Moffitt (1986) formalized how phase-outs in means-tested programs (like SNAP or Medicaid) create high implicit tax rates even when statutory rates are modest. The Affordable Care Act (2010) extended this architecture to the middle class by tying health insurance contributions to federal poverty thresholds. Recent scholarship (Bernstein, 2025) has begun to document how these “stacked” AGI-linked obligations—specifically health subsidies and modern student loan repayment plans—create a new, non-statutory marriage penalty that rivals the distortions of the pre-reform era.
3. Institutional Mechanics: ACA and RAP
The ACA “Subsidy Cliff”
The Affordable Care Act calculates premium tax credits based on household income relative to the Federal Poverty Level (FPL). Under current law, a sharp “eligibility cliff” exists at 400% of the FPL.
The Mechanism: Below 400% FPL, households pay a sliding-scale percentage of income toward a benchmark plan. At 400.01%, the subsidy drops to zero instantly.
The Marriage Interaction: For a two-person household in 2026, the cliff is $86,560. Two individuals earning $45,000 and $50,000 are subsidized as singles but lose all credits upon marriage, as their $95,000 joint income clears the cliff.
The RAP “Percentage Jump”
The 2025 Repayment Assistance Plan (RAP) for student loans uses a graduated percentage schedule that applies to the entire AGI once a threshold is reached.
The “Jump” Effect: Unlike a tax bracket, entering the 8% tier means you pay 8% on every dollar of your AGI, not just the income above the threshold. A marginal raise can trigger a massive jump in total required payments.
Inflation and Indexing: These RAP brackets are not indexed to inflation. As nominal wages rise, households are pushed into higher percentage tiers, creating a “bracket creep” that raises de-facto taxes without legislative action.
Unlike previous income-driven plans that protected a large percentage of income (up to 225% of the Federal Poverty Level), RAP applies a percentage to the total Adjusted Gross Income (AGI). While this results in higher monthly outlays, it mechanically functions as an amortization accelerator. By forcing a higher payment early in the loan lifecycle, the program reduces total interest accrual and shortens the path to zero balance. However, this acceleration is involuntary and directly consumes household liquidity during the critical early years of marriage and family formation.
This liquidity squeeze creates a “Reversal Trigger.” Under the 2026 OBBBA framework, households retain the right to opt out of RAP and convert to a Standard Federal Repayment Plan—a fixed-term, non-AGI-linked schedule (typically 10 to 25 years depending on the balance). When the RAP “percentage jump” causes the income-linked payment to exceed the cost of a fixed-term monthly payment, the household is incentivized to “reverse” their strategy.
This conversion can lead to a higher loan balance if borrowers have accrued unpaid interest or are behind on their RAP payments, as that interest is immediately added to the principal.”
4. Numerical Framework: The Case Study
To quantify the impact of AGI-linked programs, we track a “Base Case” of two individuals who then choose to marry. Person One earns $45,000 with $20,000 in student debt, and Person Two earns $50,000 with $25,000 in student debt.
Example One: The Marriage Penalty (No Children)
In the Base Case as singles, both individuals are well below the 2026 single-person ACA subsidy cliff of $63,840. They receive subsidies that cap their combined health insurance costs at approximately $9,462 per year. Under RAP, they face repayment rates of 4% and 5% respectively, totaling $4,300. Combined with a total federal tax liability of $7,360, their total annual required outflows as singles equal $21,122.
When they marry, the aggregation of their $95,000 income creates two distinct programmatic shocks despite the fact that their federal income tax remains unchanged at $7,360.
First, they face a reduction in the ACA subsidy. For a two-person household, the 2026 subsidy cliff is only $86,560. Because their $95,000 joint income exceeds this threshold, they lose 100% of their premium tax credits. Their out-of-pocket health insurance costs jump to the full market “sticker price” of $12,000—an immediate loss of $2,538 in annual disposable income.
Second, they face the RAP Jump: their $95,000 joint AGI places them in the 9% repayment bracket. Because RAP applies this percentage to the entire income base rather than just the marginal amount, their required student loan payments rise to $8,550—an increase of $4,250 over their combined single payments. Consequently, their total annual required outflows rise to $27,910. Marriage has cost this couple $6,788 per year in disposable income, effectively functioning as a 7.1% “de facto” tax on their household formation.
Example Two: The Child as a “Cliff Buffer”
Adding one child to the married couple at the same $95,000 income level demonstrates how household size temporarily mitigates the penalty. For a family of three, the 2026 ACA cliff moves to $109,280. Because $95,000 is now below the threshold, the household regains subsidy eligibility, and premium costs fall to $9,462. When combined with the $2,000 Child Tax Credit and a $600 RAP dependent adjustment, total required outflows drop to $22,772. While the child “cures” the marriage-related subsidy loss, it creates a new “marginal trap”: if this family earns a $6,000 raise (to $101,000), they hit the 10% RAP bracket, resulting in an Implicit Marginal Tax Rate (IMTR) of 48%.
Example Three: The 55% Marginal Rate (Higher Income)
As household income rises to $114,500, the family clears the three-person FPL cliff of $109,280, losing all health subsidies again. This $19,500 increase in gross income is subjected to a “stacked” set of claims that severely erodes the gain. First, federal income tax liability rises by $2,340 as a portion of the income moves from the 12% to the 22% statutory bracket. Second, the household hits the maximum RAP tier of 10%; because this higher rate applies to the entire income base, required payments rise by $2,900. Third, the household crosses the ACA subsidy cliff, increasing out-of-pocket premiums by $5,538 as they move to full market-price family coverage.
In total, these three programs claim $10,778 of the $19,500 raise. The household retains only $8,722 of their additional earnings, yielding an IMTR of 55%. For every dollar earned in this range, the household sees only 45 cents in realized gains, while the remaining 55 cents is consumed by the interaction of the subsidy cliff, the non-marginal repayment jump, and statutory tax progression.
The rational escape from this 55% trap is the ‘Reversal Trigger’—electing a Standard Plan to trade long-term interest capitalization for immediate cash-flow survival.”
5. Interpretation and Policy Implications
The findings confirm Feldstein’s concerns regarding the deadweight loss of high marginal rates but relocate the source of the problem from the IRS to the Department of Education and Health and Human Services.
The RAP Design Flaw: Applying rates to total AGI rather than marginal income creates “kinks” that discourage career advancement. A modest raise that pushes a household into a new $10,000 bracket can trigger a payment spike that exceeds the value of the raise itself.
The Indexing Erosion: Because the RAP’s $10,000 bracket thresholds are not indexed to inflation, the program’s relative benefit erodes for all borrowers over time. While the marriage penalty remains mathematically constant in nominal terms, the “escape” of lower-tier payments vanishes as inflation pushes more households—both single and married—into the maximum 10% tier. This creates a “ratchet effect” where the high-IMTR environment of Example Three eventually becomes the universal default for the middle class.
Correctable Choices: These are not inevitable results of progressivity. Replacing the 400% FPL cliff with a smooth phase-out and applying RAP rates marginally would align these programs with the goal of family stability.
6. Conclusion
The modern marriage penalty is no longer a matter of simple tax brackets. It is a ‘stacked’ phenomenon where independent policy choices in health and education aggregate into a punitive burden on household formation. Restoring the economic incentive for marriage requires more than tax reform; it requires a smoothing of the cliffs and jumps that define the current AGI-linked landscape.
Authors Note:
To look back: “For a foundational look at how RAP’s design creates these misaligned incentives, see my initial memo: Student Loans and the Marriage Incentive Problem..
To look ahead: “To see these ‘stacked’ penalties applied to a real-world scenario with specific debt loads, read my follow-up: Case Study One: Identical Incomes, Unequal Debt.
7. Bibliography
Becker, G. S. (1973). A Theory of Marriage: Part I. Journal of Political Economy, 81(4), 813-846.
Bernstein, D. P. (2025). Liquidity Today, Tax Traps Tomorrow: The Interaction of AGI-Linked Benefit Programs. Economic Memos Research Series.
Eissa, N., & Hoynes, H. W. (2004). Taxes and the labor market participation of married couples: The earned income tax credit. Journal of Public Economics, 88(9-10), 1931-1958.
Feldstein, M. (1995). The Effect of Marginal Tax Rates on Taxable Income: A Panel Study of the 1986 Tax Reform Act. Journal of Political Economy, 103(3), 551-572.
Feldstein, M. (1999). Tax Avoidance and the Deadweight Loss of the Income Tax. Review of Economics and Statistics, 81(4), 674-680.
Hausman, J. A. (1981). Labor Supply. In H. J. Aaron & J. A. Pechman (Eds.), How Taxes Affect Economic Behavior (pp. 27-83). Brookings Institution.
Moffitt, R. (1986). The Econometrics of Piecewise-Linear Budget Constraints: A Survey and Exposition of the Maximum Likelihood Method. Journal of Business & Economic Statistics, 4(3), 317-328.
Patient Protection and Affordable Care Act, Pub. L. No. 111-148, 124 Stat. 119 (2010).