Between early 2021 and early 2024, the United States saw an increase in unauthorized immigration unlike anything in recent memory. Roughly 7 million people entered the country outside formal immigration channels over that stretch, according to Congressional Budget Office estimates, nearly double the number of legal permanent immigrants during the same period. Then, starting in mid-2024, the flow reversed almost as sharply.
That dramatic swing raised a set of questions that have been hard to answer with the data economists usually rely on: How did these inflows change local job markets? What happened to wages? Did rents and home prices rise? And did local governments end up spending more on safety-net programs? A new working paper from the Federal Reserve Bank of Dallas tries to give those questions concrete, empirical answers for the first time.
A gap in the evidence
Daniel J. Wilson of the San Francisco Fed and Xiaoqing Zhou of the Dallas Fed set out to fill what they describe as a hole in the immigration research literature. Prior work has looked closely at legal immigration or at immigration overall, but it has generally not separated out unauthorized flows. That was a reasonable choice historically, when unauthorized net immigration averaged only about 100,000 people per year. But the recent surge, averaging around 1.75 million per year, changed the picture enough that older estimates may not apply.
The researchers note that unauthorized immigrants tend to differ from legal immigrants in ways that matter for economic analysis. They are more likely to be of working age, more concentrated in lower-skilled occupations, and often live in higher-occupancy households. Standard survey data also does a poor job of counting them, which means much of the existing evidence about their labor market impact has come from indirect estimates or from theoretical models rather than direct measurement.
Building a new dataset
To get around the measurement problem, Wilson and Zhou pulled together administrative microdata that had only recently become available through Freedom of Information Act requests filed by the Transactional Records Access Clearinghouse at Syracuse University. Those records include individual-level immigration court files from the Department of Justice, which contain nationality, age, entry date, and zip code of residence for people issued a “Notice to Appear” (the document that starts removal proceedings and lets someone seek asylum). They also include records on parole entries from Customs and Border Protection.
The authors combined those sources with Department of Homeland Security figures on “got-aways” (people who entered without being apprehended) and on deportations. From these, they built monthly county-level measures of entries, exits, and net inflows of working-age unauthorized immigrants, then adjusted for typical employment rates among recent arrivals from the same origin countries to produce what they call unauthorized immigrant worker flows, or UIWF.
The national picture from their data matches what the CBO and other researchers have found: net inflows rose sharply beginning in early 2021, peaked in January 2024, and then dropped steeply after the Biden administration tightened processing rules in June 2024. By February 2025, net entry turned negative.
Sorting out cause and effect
Simply comparing places with more immigrants to places with fewer doesn’t reveal causation, since immigrants tend to move toward areas that are already growing. To address that, Wilson and Zhou used a version of a standard technique in the immigration literature called a shift-share instrument. The idea is to predict how many immigrants a local area would receive based on two things: national inflows from each origin country, and the historical pattern of where earlier immigrants from those countries settled.
Their version, which they call a “two-way leave-out” design, drops the local area’s own contribution from the national inflow figure and drops each country’s own share from the historical settlement pattern. That helps rule out the possibility that persistent local economic conditions are secretly driving both past settlement and current inflows. The authors ran a battery of statistical tests and reported that the design holds up to standard validity checks.
Jobs went up, wages held steady
The central labor market finding: during the boom period, an inflow of unauthorized immigrant workers equal to 1 percent of a local area’s starting employment raised local employment by about 0.96 percent. That’s close to a one-for-one increase, suggesting these workers were, in fact, showing up in official payroll data (which comes from state unemployment insurance records).
Wages were a different story. The estimated effect on average weekly wages was negative but statistically indistinguishable from zero. The authors note one caveat: their wage data reflects average weekly earnings, which combines hourly pay and hours worked. If unauthorized immigrant workers pushed up total hours while pulling down hourly rates, the two effects could cancel out in weekly wage figures.
By industry, the employment effect was concentrated in Leisure and Hospitality. For the average local labor market, UIWF explained roughly 30 percent of total employment growth during the boom.
Results for the slowdown period from mid-2024 through mid-2025 pointed in the same direction, though with wider margins of error given the shorter time frame and smaller variation across places.
Housing: demand up, supply flat
The housing findings tell a different kind of story. An inflow of unauthorized immigrant workers equal to 1 percent of initial employment was linked to a 2.2 percent rise in local home prices and a 1.4 percent rise in market rents. The rent effect was slightly larger for multi-family units than for single-family homes. These magnitudes are broadly similar to what earlier research found for legal immigration in the 1980s and 1990s.
On the supply side, the authors found essentially no effect on new housing permits, whether for single-family or multi-family construction. Combined with the modest employment effect they measured in the construction sector, this suggests that the immigration boom acted as a demand shock hitting relatively fixed short-run housing supply, which is a straightforward way to get rising prices and rents without new building.
For the average metro area, UIWF explained about 30 percent of home price growth and 20 percent of rent growth during the boom.
Personal income and government transfers
The authors also examined broader measures of local economic activity. They found no significant effect on total personal income, but a significant negative effect on labor income per capita. The researchers interpret this as a composition effect: because unauthorized immigrants tend to earn less than native-born workers, adding more of them to a local workforce pulls down the average, even if the total pie grows.
The finding that may generate the most attention concerns government transfers, meaning payments like unemployment insurance, food assistance (SNAP), and Medicaid. An increase in UIWF equal to 1 percent of initial employment was associated with a 4.5 percent reduction in total local government transfers and a 5 percent reduction in transfers per capita.
The authors offer two possible explanations. First, higher local employment reduces eligibility and demand for safety-net programs across the board. Second, unauthorized immigrants themselves may be less likely than other residents to receive these benefits, given their legal status and the age profile of recent arrivals, who skew heavily working-age.
Wilson and Zhou acknowledge that this finding sits in tension with recent survey work by the Center for Immigration Studies, which reported higher welfare program participation among non-citizen households. They point out that the survey measures whether households received any benefit at all, not the dollar amounts, and does not distinguish legal from unauthorized immigrants or account for households containing U.S.-citizen children.
Caveats worth noting
The authors flag several limits on their work. Their slowdown-period estimates are noisier because there’s less cross-area variation to work with. Measuring unauthorized outflows is harder than measuring inflows, especially for people who left voluntarily. And while their local estimates are informative, national-level effects could differ because of general equilibrium responses, such as native-born workers moving between regions in response to immigration, that get absorbed by their statistical controls.
The researchers also caution that their findings speak to the recent episode specifically. Housing supply might respond differently over a longer horizon, and the composition of unauthorized immigration, dominated during this period by Central and South American countries, could shift again.
