Immigration and Citizenship: what the evidence shows

Evidence on immigration and citizenship in the United States is strongest for counted events, such as the 1,364,090 people who became lawful permanent residents in fiscal year 2024 and the 818,500 who naturalized that year, and weakest for the unauthorized population, which has no direct count.⁠^1,7,2 Research finds small average effects on native-born wages and mixed fiscal effects by level of government, but those averages are disputed for particular groups, places and years.

In plain words

Some immigration numbers are counted one by one, like green cards and citizenship ceremonies. Those numbers are fairly solid.

Other numbers are estimates. Nobody asks people in a national survey whether they are in the country legally, so experts estimate the number of people without legal status by subtraction. Different experts get different answers.

Research on jobs, taxes and crime usually finds small effects on average. But averages can hide who gains and who loses. This page shows what is known, where experts disagree, and what nobody knows yet. The companion pages cover what the public prioritizes and who decides.

Defined terms

Residual method. The way the unauthorized population is estimated: start with the total foreign-born population from a survey, subtract the estimated legally resident foreign-born population from administrative records, and treat the remainder as unauthorized.⁠²

Apprehension, inadmissible, encounter. CBP defines apprehensions as “the physical control or temporary detainment of a person by USBP between POEs,” inadmissibles as people encountered at ports of entry “seeking lawful admission” and found inadmissible, and reports both as encounters.⁠⁶

Net international migration. The Census Bureau’s measure of people moving into the country minus people moving out over a year.⁠⁴

Fiscal impact. Taxes paid minus the cost of public services received, which can be measured for one year or over a lifetime and for federal, state or local budgets separately.

First and second generation. Immigrants themselves (first generation) and their U.S.-born children (second generation).

Synthetic control. A method that builds a comparison group from a weighted mix of places that tracked the studied place before an event, used in the Mariel boatlift debate.⁠¹⁰

Evidence

Evidence on immigration and citizenship from federal statistics falls into several categories, each with different reliability.

Legal permanent immigration

The DHS Office of Homeland Security Statistics reports 1,364,090 new lawful permanent residents in fiscal year 2024.⁠¹ Of these, immediate relatives of U.S. citizens were 672,240 (49.3%), family-sponsored preferences 188,190 (13.8%), employment-based preferences 171,120 (12.5%), refugee adjustments 178,200 (13.1%), asylee adjustments 41,650 (3.1%) and diversity visas 49,890 (3.7%).⁠¹ More than half, 782,800 (57.4%), were already in the country and adjusted status; 581,290 (42.6%) were new arrivals.⁠¹ The leading countries of birth were Mexico (202,600) and Cuba (184,040).⁠¹

Naturalization

USCIS reports 818,500 new citizens in fiscal year 2024, a 7% decrease from the prior year, and more than 7.9 million over the last decade.⁠⁷ It reports that 89.7% of applicants passed the naturalization test on the initial exam and 94.4% including re-exams.⁠⁷ Those rates predate the 2025 civics test described on the who decides page.

The unauthorized population

The most recent official estimate is 11.0 million unauthorized immigrants on January 1, 2022, down from 11.6 million in 2010 and up from 10.5 million in January 2020.⁠² In that estimate, 44% were from Mexico, 79% had entered before January 2010, and 42% lived in California or Texas.⁠² No estimate exists for January 2021 because the 2020 American Community Survey was not considered reliable.⁠²

Pew Research Center estimates a record 14 million in 2023, up 3.5 million from 2021, equal to 4.1% of the U.S. population and 27% of the foreign-born population.⁠³ Pew’s preliminary data suggest the population fell in the first half of 2025, “possibly by as much as 1 million,” and the Current Population Survey showed the foreign-born population falling from 53.3 million in January 2025 to 51.9 million in June 2025, a change Pew says is not fully explained.⁠³

Border encounters

CBP reported 237,565 Southwest Border apprehensions in fiscal year 2025, which it called the lowest total in 55 years, compared with 201,780 in fiscal year 1970, and 87% below the average of the prior four fiscal years, which it put at 1.86 million.⁠⁵ The release was issued October 7, 2025 and said finalized end-of-year reports would follow.⁠⁵

Population change

The Census Bureau estimated net international migration at 1.3 million for the year ending June 30, 2025, down from 2.7 million the year before, a 53.8% decline, and projected about 321,000 by July 2026 if trends continue.⁠⁴ Natural change (births minus deaths) was about 519,000.⁠⁴ The Bureau warns that year-to-year comparisons should be made only within the same vintage of estimates.⁠⁴

Labor force

The Bureau of Labor Statistics reports that foreign-born workers were 19.1% of the U.S. civilian labor force in 2025.⁠¹¹ Their labor force participation rate was 66.3%, compared with 61.6% for the native born, and median usual weekly earnings for full-time wage and salary workers were $1,059 for the foreign born and $1,236 for the native born.⁠¹¹ The 2025 averages exclude October because that month’s survey was not collected during the federal shutdown.⁠¹¹ These figures include both lawful and unauthorized immigrants and are not adjusted for age or education.

Economic and fiscal research

The National Academies’ 2017 consensus report concluded that “the long-term impact of immigration on the wages and employment of native-born workers overall is very small,” that negative effects fall mostly on “prior immigrants” and “native-born high school dropouts,” and that immigration “has an overall positive impact on long-run economic growth”.⁠⁸ On budgets, it found effects “generally positive at the federal level and negative at the state and local levels,” that “state and local governments bear the burden of providing education benefits to young immigrants,” that first-generation immigrants are “more costly to governments than are the native-born,” and that the second generation are “among the strongest fiscal and economic contributors”.⁠⁸ On skilled immigration, it reported studies finding positive wage effects and also noted that “high-skilled immigration can have adverse effects on the wages or productivity of natives working in those fields”.⁠⁸

Crime

A 2020 study in the Proceedings of the National Academy of Sciences linked Texas Department of Public Safety arrest records with DHS status data for 2012 to 2018 and found that “undocumented immigrants have substantially lower crime rates than native-born citizens and legal immigrants across a range of felony offenses”.⁠¹² The authors state that “arrest rates represent only one metric of criminality,” that the undocumented population is estimated, and that the federal status data may misclassify some people.⁠¹²

Comparative notes

The OECD reports 6.2 million new permanent immigrants to its member countries in 2024, 4% fewer than in 2023 but 15% above 2019; family migration was 44% of permanent flows, labor migration 17% and humanitarian 13%.⁠¹³ New asylum applications in OECD countries reached 3 million in 2024, 13% more than in 2023.⁠¹³ Across the OECD, immigrants entering the labor market earn 34% less than native-born workers, “mainly due to them working in lower paying sectors and firms,” a gap that narrows over time.⁠¹³ OECD categories differ from U.S. visa classes, so these shares should not be compared directly with the DHS figures above.

Where the data are disputed

Immigration and citizenship data are disputed in at least four places.

The size of the unauthorized population. The official January 2022 estimate (11.0 million),⁠² and Pew’s 2023 estimate (14 million)⁠³ refer to different dates, during a period of rapid change, so they do not necessarily conflict. Both methods count people with temporary protections as unauthorized: OHSS explicitly includes TPS and DACA recipients, parolees and people awaiting removal proceedings⁠² and Pew reports that about 6 million of its 14 million had some protection from deportation.⁠³ Whether such people should be called unauthorized at all is a definitional dispute, not only a statistical one.

What border numbers measure. Apprehensions, inadmissibles and expulsions are different events,⁠⁶ and a fall in apprehensions can reflect fewer attempts, more deterrence, changes in where people cross or are processed, or policy changes at ports of entry. CBP’s own release does not separate those causes.⁠⁵

Wage effects on low-skilled workers. The Mariel boatlift of 1980 is the most cited natural experiment, and the same event yields opposite conclusions. George Borjas found that “the wage of low-skill (non-Hispanic) working men in Miami declined substantially in the years after Mariel”.⁠⁹ Giovanni Peri and Vasil Yasenov, using synthetic control, found “no significant difference in the wages of workers in Miami relative to its control after 1980”.⁠¹⁰ The disagreement turns on which workers are counted as comparable and which cities serve as controls.

Crime. The Texas study is one of the few with direct status data.⁠¹² Its findings rest on arrests, not offending, and on matching to DHS records. Skeptics of such findings point to the limits the authors themselves acknowledge, especially the reliance on arrests and on status matching; others treat the study as strong evidence because it uses direct status data rather than proxies. A single state over seven years cannot settle the national question.

What is unknown

Several basic facts about immigration and citizenship in the United States are unknown or out of date.

• The current unauthorized population. The latest official estimate is for January 2022,⁠² before the largest recent changes in both directions. Pew’s mid-2025 figures are labeled preliminary.⁠³
• Emigration. The Census Bureau says its latest migration estimates reflect methodological improvements and warns against comparing across vintages;⁠⁴ how many people left in 2025, voluntarily or by removal, is not precisely known.
• Fiscal effects of recent arrivals. The National Academies report is from 2017.⁠⁸ An updated, independent lifetime fiscal estimate covering the 2021 to 2025 period was not found for this page.
• Long-run effects of the 2025 to 2026 decline. Effects on labor supply, prices, housing and local budgets will take years to measure.
• Integration outcomes for the newest citizens under the 2025 civics test, for which no published pass rate was found.

Competing interpretations

Immigration and citizenship evidence supports more than one reasonable reading.

Small average effects as reassurance. Supporters of current or higher levels read the National Academies findings,⁠⁸ and the Texas crime study⁠¹² as showing that common fears are overstated.

Distributional effects as the real story. Supporters of lower levels read the same report as showing that losses fall on specific groups, prior immigrants, workers without a diploma and state and local taxpayers⁠⁸ and argue that averages are the wrong test.

Data gaps as a reason for caution. A third reading holds that, with the unauthorized population estimated rather than counted and the official estimate four years old, strong claims in any direction outrun the evidence. See measurement.

Application to this framework

My Vote Forward’s approach to evidence applies directly to immigration and citizenship.

1. Every number needs a date and a denominator. “Encounters,” “apprehensions” and “unauthorized population” each mean something specific; a framework question should define the term it uses. See question design.
2. Separate importance from effectiveness. A participant can rank immigration as important without believing any particular program works. See tradeoffs.
3. Show both sides of a disputed study. When the framework links evidence for a proposal, it should link the strongest opposing evidence, as in the Borjas and Peri and Yasenov exchange.⁠^9,10 See proposals.
4. Participation results are not population facts. Opinions recorded in the framework describe participants, not immigrants or the public. See representation.

Unresolved questions

• When will an official estimate of the unauthorized population after January 2022 be published, and how will it reconcile with private estimates?
• How much of the 2025 decline in net migration reflects fewer arrivals, more departures or measurement changes?
• Do the National Academies’ 2017 fiscal conclusions hold for the 2021 to 2025 arrivals, many of whom entered through parole or asylum processes?
• Can wage research agree on how to define comparable workers, so that natural experiments like Mariel yield a common answer?
• What national data would allow crime comparisons by immigration status beyond a single state?