Should the Government Redistribute Income? The Economics of Taxes, Transfers, and Trade-Offs

Picture two households. In one, a retired executive receives $480,000 a year in investment income and Social Security. In the other, a single mother working full-time at a warehouse earns $28,000. A dollar transferred from the first household to the second almost certainly improves total human welfare — because the satisfaction from a $1,000 rent payment is worth far more to someone who cannot cover it than to someone who would not notice the deduction. This intuition is the entire moral foundation of income redistribution, and it has a formal name: the diminishing marginal utility of income.

But the intuition runs into a complication almost immediately. To move that dollar, you need a tax system, a transfer agency, an eligibility determination, a delivery mechanism. Along the way, some of the dollar leaks out — in administrative costs, in the behavioral changes that taxes produce, in the work hours that generous benefits sometimes displace. The late Yale economist Arthur Okun gave this problem its most memorable formulation in his 1975 book Equality and Efficiency: The Big Tradeoff: the money must be carried from rich to poor in a leaky bucket. The policy question was never whether to carry the bucket — it was how much leakage was tolerable.

Nearly fifty years on, that question remains genuinely contested, and the evidence on all sides is richer than either camp typically acknowledges.

The case for redistribution: three distinct arguments

The welfare argument. Diminishing marginal utility means that equal amounts of money produce unequal amounts of satisfaction depending on who holds them. If you accept that welfare comparisons across people are meaningful — a contested but practically necessary assumption — then transferring income from high to low reduces welfare loss for the giver less than it raises welfare for the recipient, increasing total social welfare. This is not a left-wing invention; it follows from standard utility theory.

The market failure argument. Some forms of redistribution do not just shift income — they correct inefficiencies that markets produce on their own. Universal public education and basic healthcare access generate positive externalities (a better-educated, healthier workforce raises productivity for everyone) and address insurance market failures that would leave millions without coverage in a purely private system. These programs can simultaneously improve both equity and efficiency — which means the trade-off Okun described is not universal.

The social insurance argument. Unemployment insurance, Social Security, and disability insurance redistribute not only across income levels but across life circumstances. A risk-averse person would pay a fair premium to insure against unemployment or disability — these programs provide that insurance at social scale. Even a person who never uses them benefits from living in a society where economic misfortune does not produce destitution, with the social frictions that accompany it.

The case for caution: where efficiency costs are real

Labor supply distortions from progressive taxation. When marginal tax rates rise with income, the after-tax return to working additional hours falls. Whether high earners actually work less in response is an empirical question — the evidence suggests high-income earners are somewhat more responsive than lower-income workers, but the effect is moderate. The Congressional Budget Office models these behavioral responses when scoring tax proposals; they are real but not catastrophic at the rates the U.S. currently applies.

For context: the U.S. federal income tax system has seven brackets ranging from 10% to 37% for the 2025 tax year. The top rate of 37% applies only to income above $626,350 for single filers. Historically, the top marginal rate has been as high as 91% in the 1950s — and economic growth during that era was robust, suggesting the relationship between top rates and growth is far less deterministic than advocates on either side imply.

The benefit cliff problem. This is where redistribution policy does its clearest damage — and it is a design flaw, not an inherent feature of redistribution itself. Many means-tested programs are structured so that as a household's income rises, benefits phase out. When multiple programs phase out simultaneously, the result is an implicit marginal tax rate that can exceed 50, 70, even 100 percent on incremental earnings.

Research from the HHS Office of the Assistant Secretary for Planning and Evaluation (ASPE) found that households with children just above the poverty line face a median marginal tax rate of 51 percent — meaning each additional dollar earned reduces net income by half a dollar once benefit reductions are factored in. A family receiving SNAP, Medicaid, and a housing subsidy who takes a raise may lose more in benefits than the raise provides.

This is not an argument against redistribution — it is an argument against poorly designed redistribution.

The main tools and how their incentives differ

Progressive income taxation is the foundation of redistribution in every wealthy country. Higher rates on higher incomes fund the programs that support lower-income households. The efficiency cost is the distortion to labor supply and investment decisions — real but manageable at moderate rates.

Means-tested cash and in-kind transfers — SNAP, Medicaid, housing vouchers, TANF — provide direct material support to low-income households. Their efficiency cost is the work-disincentive embedded in phase-outs, plus administrative overhead. The empirical evidence on work disincentives is more nuanced than critics suggest: SNAP recipients have broadly maintained labor force participation, and Medicaid expansions have not produced large exit from work.

The Earned Income Tax Credit (EITC) is the most well-designed redistribution tool in the U.S. system, for one specific reason: it subsidizes work rather than replacing income. For a working family with three or more children in 2024, the maximum EITC credit was $7,830 — a direct income boost that rises with earnings through the phase-in range, rewarding entering the labor force. The IRS EITC program delivers this credit to tens of millions of working families annually, making it the largest federal anti-poverty program for working-age adults by expenditure. Research by economists including Raj Chetty consistently finds that the EITC has among the lowest deadweight loss per dollar of poverty reduction of any federal transfer program.

Social insurance programs (Social Security, unemployment insurance, disability insurance) are the largest redistributive programs by spending. Social Security alone lifted 28 million people above the poverty line in 2023, according to Census Bureau Supplemental Poverty Measure data — more than any other government program.

What the evidence shows: does redistribution harm growth?

The empirical record does not support the claim that moderate redistribution crushes economic growth. IMF research published in Finance & Development (Berg and Ostry, 2011) found that high inequality itself can be harmful to growth sustainability — making redistribution a potential complement to growth rather than a pure trade-off. Cross-country comparisons consistently show that high-redistribution countries (Nordics, Germany, Netherlands) produce strong long-term growth, high social mobility, and living standards that match or exceed those in lower-redistribution peers.

The U.S. redistributes substantially less than most OECD peers. Pre-tax income inequality — measured by the Gini coefficient — is similarly high across most wealthy nations. What differs sharply is post-tax-and-transfer inequality. The U.S. ends up with higher after-transfer inequality not because markets produce more inequality here but because the government intervenes less to offset it.

A framework for evaluating redistribution proposals

The honest economic framework for evaluating any redistribution proposal is not "is this fair?" — that is a values question — but "how much poverty reduction per dollar of efficiency cost does this achieve?" Apply three tests:

1. What is the implicit marginal tax rate? Any program that phases out rapidly creates work-disincentive damage. Evaluate phase-out design, not just generosity.

2. Does it subsidize work or replace it? The EITC model (subsidize the transition to work) produces better labor market outcomes than programs that provide equivalent income unconditionally in the phase-out zone.

3. How does the administrative overhead compare to the benefit? A program that costs $0.70 in administrative and behavioral distortion costs to deliver $1.00 in benefits to recipients is meaningful redistribution. One that costs $0.95 may not be.

The most defensible redistribution policy is not the one that transfers the most income. It is the one that achieves the most human welfare improvement — more people above subsistence, more children in stable housing, more workers able to take the job offer — per dollar of efficiency cost. That is an empirical question, and the evidence points clearly toward work-subsidizing, low-cliff-creating designs over unconditional income replacement at high marginal phase-out rates.

Okun's bucket still leaks. The goal is to make it leak less — not to stop carrying it.