The $9 Trillion Question: Who Really Pays — and Who Really Benefits — From American Immigration

The United States is home to 53.3 million foreign-born residents as of January 2025 — the largest number ever recorded and 15.6% of the total population [1]. This share has not been seen since 1890, and it has nearly tripled since hitting a modern low of 4.7% in 1970 [1]. The economic effects of this demographic transformation are real, large, and distributed unevenly across classes, regions, and racial groups. The debate over those effects has generated more heat than light, with both sides routinely cherry-picking data to support predetermined conclusions.

This report presents the evidence as it stands — contested, uncomfortable, and resistant to the clean narratives that politicians prefer.

The Federal Fiscal Ledger: A Net Positive

At the federal level, the math favors immigration. The Congressional Budget Office estimated in 2024 that the immigration surge between 2021 and 2026 would reduce federal deficits by $900 billion over the 2024–2034 period [2]. The mechanism is straightforward: immigrants in the surge population are projected to pay $788 billion in income and payroll taxes, while adding $177 billion in mandatory spending — a clear net positive for federal coffers [2].

The CBO further projects this immigration will add $8.9 trillion to nominal GDP over the same decade, a 2.4% increase [2]. The economic ripple effects of a larger workforce — more output, more consumption, more tax revenue — account for an additional $285 billion in deficit reduction beyond the direct fiscal contributions [2].

Source: FRED / Bureau of Economic Analysis

Data as of Mar 20, 2026CSV

These numbers need context. The federal government captures the revenue side of immigration more efficiently than it bears the cost side, because the most expensive public services — schools, hospitals, police — are funded primarily at the state and local level.

State and Local Costs: A Different Story

The CBO published a companion analysis in 2025 examining the immigration surge's effect on state and local budgets in 2023 and found the opposite picture: costs exceeded revenue gains [3]. The largest cost increases were for primary and secondary education, shelter and related services, and border security [3]. State and local tax revenues grew, but not enough to cover the additional service demands.

This fiscal asymmetry — federal gains, local losses — is the central structural problem in the immigration debate. A meatpacking plant in rural Iowa generates payroll tax revenue for Washington while the local school district scrambles to hire ESL teachers. The federal government collects; the county pays.

The Manhattan Institute's 2025 fiscal analysis adds granularity by education level [4]. A college-educated immigrant arriving at age 30 contributes roughly $215,000 net to federal finances over 10 years and $1.7 million over 30 years. An immigrant without a high school diploma represents a net fiscal loss of approximately $20,000 over 10 years and over $130,000 over 30 years when descendants are included [4]. College-educated immigrants pay 3.3 times more in taxes than those without a high school diploma, while less-educated immigrants receive 2.3 times more in government benefits [4].

Education at arrival, in other words, is the single strongest predictor of whether an immigrant will be a net fiscal contributor or a net fiscal cost.

The Welfare Debate: Both Sides Have a Point

Few topics generate more mutual incomprehension than immigrant welfare use. The Cato Institute and the Center for Immigration Studies — the two most prominent think tanks on opposite sides of the immigration debate — present findings that appear contradictory but are both methodologically defensible.

Cato's analysis of 2023 data found that on a per capita basis, immigrants consumed 24% less in welfare and entitlement benefits than native-born Americans [5]. Immigrants were 14.8% of the population but consumed only 10.4% of all means-tested welfare and entitlement benefits [5]. Native-born Americans averaged $7,134 in old-age entitlements and $3,638 in means-tested benefits per capita, compared to $4,864 and $3,370 respectively for immigrants [5]. A 30-year Cato study found that immigrants reduced federal deficits by $14.5 trillion since 1994 [5].

The Center for Immigration Studies, using the same Census Bureau survey data but measuring at the household level rather than per capita, reported that 61% of households headed by undocumented immigrants used some form of welfare in 2024, compared to 37% of native-born households [6]. Legal immigrant households fell in between at 51% [6]. The programs most commonly accessed were Medicaid, school lunch and breakfast programs, WIC, and the Earned Income Tax Credit [6].

The reconciliation of these findings lies in methodology and what gets counted. Cato measures per-person benefit consumption in dollar terms — and since immigrants skew younger and use far less Medicare and Social Security (the most expensive programs), their per capita costs are lower. CIS measures household participation rates — and since undocumented immigrant households tend to be lower-income and include U.S.-born children who are legally entitled to benefits, their participation rates are higher. CIS itself notes this is not evidence of fraud; it reflects low incomes and the presence of citizen children [6].

The uncomfortable truth is that both findings can be simultaneously true. Immigrants consume less per person in total benefits. But undocumented immigrant households participate in safety-net programs at higher rates, primarily through their U.S.-citizen children, who are legally eligible.

Wages: The Borjas-Card Divide

The academic debate over immigration's wage effects has been dominated for two decades by Harvard's George Borjas and Berkeley's David Card, who reach substantially different conclusions using different methodologies.

Borjas, analyzing national labor markets by education and experience cohorts, finds that immigration reduces wages for competing native workers. His most cited estimate: a 10% increase in labor supply from immigration reduces wages by 3–4% for the affected group [7]. For native-born high school dropouts — the group most directly in competition with low-skilled immigrants — the estimated wage reduction is 2–5%, depending on whether the effect is measured in the short or long run [7].

Card, studying local labor markets and natural experiments (most famously the 1980 Mariel boatlift), finds negligible wage effects [8]. His work suggests that local economies absorb immigrant labor through capital adjustment, firm creation, and complementary task reallocation without significant native wage depression [8]. The Ottaviano-Peri framework, building on Card's approach, argues that immigrants and native workers are imperfect substitutes — they tend to perform different tasks even within the same occupation — and that immigration may actually raise wages for most native workers while depressing them only for prior immigrants [7].

The distributional question here is critical and often evaded. Even under the more optimistic Ottaviano-Peri estimates, low-skilled native workers — disproportionately Black and Hispanic Americans without college degrees — bear real wage costs. The benefits of cheaper labor flow to employers (higher profits), consumers (lower prices for food, construction, and services), and high-skilled workers (complementary productivity gains). The costs concentrate among those least able to absorb them. This is a genuine policy trade-off, not a talking point.

The Industries That Cannot Function Without Immigrants

The sectoral dependence on immigrant labor is not hypothetical. Foreign-born workers make up roughly 17% of the total U.S. labor force but far higher shares in specific industries: approximately 37% of agricultural workers, 25% of construction workers, and 22% of food service workers [9]. In some subsectors — fruit and vegetable harvesting, meatpacking, hotel housekeeping — the share exceeds 50%.

Source: Bureau of Labor Statistics

Data as of Mar 20, 2026CSV

The consequences of rapid removal would be concrete. Agricultural economists have estimated that removing undocumented workers from U.S. agriculture would reduce domestic production by $30–60 billion annually and increase food prices by 5–6% [9]. Construction, already facing severe labor shortages, would see project timelines extend and costs rise further. Healthcare — particularly direct care work like home health aides and nursing assistants — relies heavily on immigrants to staff an industry serving an aging population [10].

Social Security: The Demographic Time Bomb

The worker-to-beneficiary ratio for Social Security has fallen from 5.1:1 in 1960 to approximately 2.8:1 today, and the CBO projects it will reach 2.1:1 by 2055 [11]. Without immigrants and their U.S.-born children, the prime working-age population (ages 25–54) would have shrunk by more than 8 million between 2000 and 2023 [10].

Immigration is the only factor currently preventing the working-age population from outright decline. The Bipartisan Policy Center found that cutting immigration by 50% would accelerate Social Security insolvency by several years, while increasing immigration by 50% would extend the trust fund's solvency, though not indefinitely [11]. The CBO has noted that immigration would need to rise to five times its current annual level just to maintain today's working-age share through 2060 — making it clear that immigration alone cannot solve the entitlement funding crisis [10].

Source: World Bank

Data as of Mar 20, 2026CSV

The international comparison is instructive. Japan, with a foreign-born population share under 3% and net migration that turned negative in 2021, faces the most acute demographic crunch among developed nations [12]. Germany absorbed over 1.1 million net migrants in 2015 alone during the refugee crisis and has maintained levels above 300,000 annually since [12]. Canada has pursued the most aggressive immigration expansion among peer nations, with net migration exceeding 430,000 in 2022–2023 [12]. Australia runs a points-based system that produced net migration averaging 140,000–250,000 annually pre-COVID [12].

The U.S. net migration of 1.3 million in 2022–2023 is the highest in absolute terms but represents a smaller per-capita rate than Canada or Australia [12].

Innovation: The Outsized Immigrant Contribution

The data on immigrant contributions to innovation and entrepreneurship is among the least contested in this debate. In 2025, 231 of the Fortune 500 companies — over 46% — were founded by immigrants or their children, the highest share since tracking began in 2011 [13]. These companies generated $8.6 trillion in revenue in fiscal year 2024 and employed over 15.4 million people worldwide [13]. Immigrants and their children founded 80% of Fortune 500 companies in professional services, 65.6% in manufacturing, and 57.5% in information technology [13].

Patent data tells a similar story. Immigrants comprise roughly 16% of U.S. inventors but produce 23% of patents [14]. A Stanford study found that 32% of total U.S. innovative output since 1990 can be attributed to immigrant inventors, including spillover effects on their native-born collaborators [14]. Immigrants have won 38% of U.S. Nobel Prizes in physics since 1901, and in 2019, 50% of American Nobel laureates were foreign-born [14].

These contributions are concentrated among high-skilled immigrants, which frames the policy question sharply: the current U.S. system allocates roughly two-thirds of permanent visas through family reunification and the diversity lottery, with only about 15% through employment-based categories. Canada and Australia allocate the inverse ratio, selecting primarily for skills [14]. Whether this matters depends on what you think immigration policy should optimize for.

Crime: What the Data Shows and What It Doesn't

Academic research on immigrant crime rates is remarkably consistent. A 2020 PNAS study using comprehensive Texas arrest data found that undocumented immigrants were roughly 50% as likely as native-born citizens to be arrested for violent crimes, 40% as likely for drug offenses, and 25% as likely for property crimes [15]. The Cato Institute, analyzing incarceration data from 2010–2023, found that undocumented immigrants had consistently lower incarceration rates than native-born Americans [16]. Between 1980 and 2022, the immigrant share of the U.S. population doubled while total crime dropped 60% [15].

Research on sanctuary policies — jurisdictions that limit cooperation with federal immigration enforcement — found no detectable increase in crime rates. A PNAS study showed that sanctuary policies reduced deportations by about one-third, primarily by reducing deportations of people with no criminal convictions, without affecting deportations of people with violent convictions [17]. A Center for American Progress study found 35.5 fewer crimes per 10,000 people in sanctuary counties compared to non-sanctuary counties [17].

But aggregate statistics do not address what drives public anxiety. Springfield, Ohio — a city of under 60,000 — absorbed between 12,000 and 20,000 Haitian immigrants over a four-year period, nearly all with some form of legal status [18]. The resulting strain on housing, schools, and city services was real and visible: housing prices spiked, the school system scrambled for ESL resources, and long-time residents experienced the social fabric of their community changing faster than institutions could adapt [18]. That these immigrants were largely law-abiding and filling genuine labor shortages did not make the disruption less tangible for existing residents.

Similarly, claims about Venezuelan gang activity in Aurora, Colorado, fueled national media coverage disproportionate to the actual documented incidents. The pattern is consistent: rapid, concentrated resettlement — regardless of the immigrants' legal status or individual behavior — creates legitimate community friction that national-level statistics cannot capture.

Fraud: Real but Contextual

Documented cases of benefits fraud involving immigrant communities exist and should not be dismissed. Large-scale childcare funding fraud schemes, worksite identity theft operations, and organized benefits fraud have been prosecuted in multiple states. Congress responded with the Deporting Fraudsters Act, which passed the House in early 2026 and would make conviction for federal benefits fraud, identity document fraud, or Social Security number misuse a deportable offense [19].

The scale matters for context. Estimates of total federal fraud range from $233 billion to $521 billion annually across all programs and populations [19]. Immigrant-specific fraud, while real, represents a fraction of this total. Native-born Americans commit the vast majority of benefits fraud by volume simply because they constitute the vast majority of beneficiaries. The policy question is whether immigrant fraud warrants specific deportation consequences beyond the criminal penalties that already apply to all offenders.

The H-1B Squeeze

The H-1B visa program, designed for specialty occupations, has become a focal point for the high-skilled immigration debate. In September 2025, the Trump administration imposed a $100,000 supplemental filing fee on certain H-1B petitions, primarily those for beneficiaries outside the United States [20]. Combined with existing fees — the base $780 filing fee, the $4,000 surcharge for large H-1B-dependent employers, and the premium processing fee of $2,965 — the total cost of sponsoring a single H-1B worker can now exceed $108,000 [20].

For large technology companies, this is an operating cost. For small employers, startups, and universities, it is prohibitive [20]. The practical effect is to consolidate H-1B access among the largest firms while pricing out the smaller employers that account for a disproportionate share of innovation. Whether this is an intended feature or an unintended consequence depends on whom you ask.

The Skills Composition Question

The composition of immigration flows has shifted significantly. Among recent arrivals, roughly 30% hold bachelor's degrees or higher, while approximately 25% lack a high school diploma [4]. This bimodal distribution — heavy at the top and bottom, light in the middle — creates different economic effects in different sectors.

Merit-based systems like Canada's and Australia's produce immigrant populations with higher average education levels and faster economic integration. The U.S. system, weighted toward family reunification, produces a more diverse skills distribution that reflects existing immigrant communities rather than labor market needs [4]. The Manhattan Institute analysis found that education at arrival is the dominant predictor of lifetime fiscal impact: high school dropouts produce descendants of whom only 7% earn college degrees, while advanced-degree holders produce descendants of whom 90% earn at least a bachelor's [4].

This does not mean family reunification has no economic value — family networks provide settlement support that reduces public costs and accelerates employment. But the fiscal math is clear: the skill level of admitted immigrants has enormous consequences for public budgets over generational time horizons.

What This Means

The honest summary is that immigration produces large aggregate economic gains that are distributed unequally. The federal government benefits. Employers benefit. Consumers benefit through lower prices. High-skilled immigrants generate substantial fiscal surpluses and outsized contributions to innovation.

The costs fall on state and local governments funding schools and services, on low-wage native workers competing for the same jobs, and on communities absorbing rapid population growth without adequate infrastructure. These costs are real, measurable, and borne disproportionately by people with the least political power.

The policy question is not whether immigration is "good" or "bad" for the economy — that framing is too crude to be useful. The question is: who bears the costs, who captures the benefits, and whether the political system is willing to compensate the losers. So far, the answer to that last question has been no.

Source: U.S. Census Bureau / American Community Survey

Data as of Mar 20, 2026CSV