In light of ICE stepping up interior “enforcement”,1 I thought I would republish a piece I wrote last year about what we know about the undocumented population of the United States.
The United States has a large population of undocumented immigrants, making up perhaps 3% of the US population. Close to one in four immigrants in the US are not documented.
This is relatively high for a rich country, and much US political discussion is devoted to the issue of undocumented migrants. Therefore, it seems worth understanding who undocumented migrants are, and how they fit into both US society and the US economy.
Two notes to start:
A note on what this is: This is designed as an overview of what the academic literature says about the US’ undocumented population at the time the constituent papers were written. It is descriptive, rather than prescriptive. It does not attempt to recommend policies, because that is not what a literature review is for.
In other parts of this Substack, I make arguments based on personal opinion; for the migration living literature review, I try to stick closely to only what the evidence says. I show that undocumented migrants in the US are largely law-abiding people who participate in the labor market, because that is what the evidence shows; it is not an attempt to make a political point.
A note on data: somewhat by definition, the empirical evidence on the undocumented population is weaker than on documented residents of a country. There is limited information available on a population that has limited interaction with the legal institutions of a country.
For instance, the method for determining how many undocumented people are in the US is fairly rudimentary. It simply involves taking the number of temporary legal immigrants2 and applying mortality table data to the number of green cards issued, to determine how many documented permanent residents there “should” be in the country. One then sums the number of permanent residents and temporary immigrants, and subtracts the sum off the total population to find the number of undocumented people.
There are several ways this could be wrong, though:
Our population estimates might be wrong. Undocumented people typically try to avoid interacting with the state; this is likely to include census takers. The Census makes some assumptions on what percentage of the undocumented population they are able to reach, but they could be wrong.
Mortality life tables might not accurately reflect the survival rates of green card holders. If immigrants live longer than expected, one might see “too many” people in the population data and assume that they’re undocumented - but actually, they are green card holders who are unexpectedly healthy.
The difficulty in accurately counting this population means that even descriptive statistics have significant error bars. For instance, the Pew Center is one of the major organizations that attempts to measure the undocumented population, and several economics papers (e.g. Borjas 2016, Borjas 2017) build on their work. But Bhandari, Feigenberg, Lubotsky and Medina-Cortina 2021 believes the Pew Center underestimates the population of undocumented migrants from Mexico by 35%; another paper argues that Pew underestimates the undocumented population by 50%.
Given the uncertainty of even basic descriptive statistics, then, this post relies less on causal inference than most of my migration literature review. This post is less about establishing the effect of changes in policy on undocumented people and more about trying to establish who undocumented people in the United States even are.
Descriptive Statistics
With that out of the way, who are undocumented immigrants in the US? Or at least, who are they (to the best of our knowledge)?
Demographic Characteristics
Borjas 2016 works with a (de-identified) version of the Current Population Survey to find that (as of the mid-2010s), undocumented migrants are:
Disproportionately male (54%)
Younger (on average) than either the native-born or legal immigrants.
Have much lower education levels than either the native-born or legal immigrants. Notably, nearly half of undocumented migrants (42%) do not have a high school diploma, while only 7% of the native-born do not have a high school diploma.
… and are concentrated in states near the border3
Country Of Origin
About 70% of undocumented migrants in the US come from Latin America, with about a third to half coming from Mexico and another ~20% from the Northern Triangle of El Salvador, Honduras and Guatemala.
However, the Discourse often neglects that perhaps one in three undocumented migrants in the US is not from Latin America. About 6% of undocumented migrants come from India; another 6% of undocumented migrants come from Canada and Europe.
Tenure
Most have been in the US a while. Only about 20% have been in the US less than five years; 40% have been in the US more than 15 years.
Some of this may (ironically) be due to increased enforcement over the last several decades. Many undocumented migrants would like to be circular migrants - coming to the US, earning some money, going back to their home country, and repeating the cycle. But as the border becomes riskier and crossing becomes more difficult, it becomes more dangerous to go back and forth - migrants are more likely to pick one side and stay there.4
Furthermore, if a migrant now does not think trying the border again is a good idea, they are more likely to bring their family with them. For a six-month stint away, it is likely one would leave a wife and children at home; for a ten-year stint abroad, one is more likely to try to bring them along.
Given the long tenure of migrants, it is also very common for households to be mixed-status. Around 70% of undocumented migrants live in a household where at least one person is either a citizen or documented migrant.
The most common situation appears to be a US citizen child living with undocumented parent(s); there are about 4.4 million such children. This means about 6% of US children live with an undocumented parent.
Population Size
As noted above, there is some disagreement on how many undocumented people live in the United States. The most commonly cited number is around 11 million, though other sources find up to 16 million.
Most sources do agree that the undocumented population likely peaked around 2007, and then decreased by several hundred thousand per year through 2021 or so. This is likely for two reasons:
Employment in construction - the second-largest sector for undocumented immigrants in the US - dipped from 2006 to 2022.
Demographic shifts - and lower fertility - in Mexico and Central America have meant there are simply fewer young people who want to move (Hanson, Liu and McIntosh 2017).
In recent years, it looks like undocumented migration is on the upswing again, though the total is still significantly below its 2007 high.
Employment
Undocumented men are much more likely to work than documented men. Borjas 2016 finds that 87% of undocumented men are employed, compared to 78% of legal immigrants and 73% of native-born men.
Undocumented women, however, are much less likely to work than documented women. Only 55% of undocumented women are employed, compared to 59% of legal immigrants and 67% of native-born women.
Doing the weighted average,5 this means undocumented migrants are, overall, more likely to be employed than either legal immigrants or the native-born. Nor does this seem to have changed since 2014 (the last year in the Borjas sample); in 2022, Pew estimated that 75% of undocumented migrants were working, while only about 60% of the overall US population is currently working.
The high labor force participation of undocumented migrants makes sense for several reasons:
Legal immigrants and the native-born have access to social safety net programs that undocumented people do not. For instance, undocumented people cannot access unemployment or disability benefits. Without such benefits, work is likely to be not very optional.
Furthermore, undocumented immigrants are in a considerably worse financial state than documented immigrants or the native-born and are thus unlikely to be able to survive on savings. In 2008, the average undocumented Latino immigrant had a net worth of $38,000; the average documented Latino immigrant had a net worth of $66,000. The gaps for other racial groups are similarly large (see table 2).
Undocumented migrants come to the US largely for economic opportunity, balancing the higher wage-earning potential with the legal risks. If one is not currently pursuing that economic opportunity, being in legal jeopardy is less worthwhile.
However, this high labor force participation is still worth noting because of the relatively low levels of education in the group. Among legal immigrants and the native-born, employment rates are lowest for those with the least education. An undocumented man is 17 percentage points more likely to be employed than a native-born person, controlling for education, age, and socioeconomic status.6
I’ll discuss aggregate effects in more detail in the fiscal effects section, but this high labor force participation rate is a significant reason that undocumented migrants are a net fiscal positive for the US.
Wages
Borjas 2017 uses Current Population Survey data to look at the wages of undocumented migrants. Perhaps unsurprisingly, they are low; at the age of 25, perhaps 18% below documented immigrants and 26% below the native-born.
More surprisingly, they are completely flat over time. The average documented immigrant makes considerably more at 45 than at 25; the average undocumented migrant does not.
This also means the hourly wage penalty for being undocumented rises over time; by age 45, the average documented immigrant makes about double what the average undocumented migrant does. Most of this wage penalty does not appear to be due to status; documented immigrants are a different population with different (and higher) expected earnings, largely due to higher education levels among the documented population.
But there is a significant wage penalty for status alone - around 10%. Put another way, if an undocumented worker became documented tomorrow, they could probably get a job paying about 10% more.
Interestingly, this wage penalty is much lower than it used to be; Borjas estimates that in 2007, the wage penalty for being undocumented was about double what it was in 2014. He does not really have an answer for why this is, though.
Fiscal Effects
Undocumented immigrants are sometimes considered to be a drain on the US economy. This appears to be exactly the opposite of what actually happens - since undocumented immigrants often pay taxes, but are entitled to few benefits.
How do undocumented migrants pay taxes?
In order to work in many US jobs,7 you have to fill out a W-9 where you list a Social Security Number (SSN) or an Individual Taxpayer Identification Number (ITIN).
Acquiring a SSN requires legal status, so some undocumented people list a fake SSN or list a SSN belonging to someone else on their W-9. Furthermore, you don’t need legal status to acquire an ITIN. Historically, the IRS does not care about your documentation status;8 even as an undocumented person, you very much still have to pay federal income tax.
Not all undocumented people work a job that does federal tax withholding; some work solely cash-in-hand. However, even they cannot escape the long arm of taxation, as consumption taxes are levied at the point of consumption. No one checks your immigration status when you buy clothes, food or other goods; sales tax is always levied.
What do undocumented people contribute to the economy (besides taxes)?
Per Edwards and Ortega 2017, undocumented people contribute about 3% to the GDP of the US.9 This is somewhat less than their percentage of the workforce, as they are paid less than the average worker.
Still, this is a substantial amount; undocumented people contribute (roughly) similar amounts to the economy as the defense industry.
The Institute for Taxation and Economic Policy estimates that undocumented people pay about $60B in federal taxes each year, and about $30B in state and local taxes each year.
What is the overall impact of undocumented migrants on the US’ fiscal policy?
Undocumented migrants clearly pay some taxes in the US. But do they pay more in taxes than they receive in benefits? There have been a number of attempts to figure this out.
I’ll start with the most negative findings:
The Heritage Foundation finds that the average undocumented migrant household is a net tax recipient, because they receive more benefits than they pay in taxes. This costs the state around $15,000 per year per household.10
In 2006, the Texas Office of the Comptroller calculated that undocumented migrants generated $424M for the state but cost local governments $1.44B.11
However, most other estimates are positive.
Hanson 2009 estimates that undocumented migrants raise native welfare by 0.03 percent of GDP. This is basically indistinguishable from zero, as it would be about $26/person/year.12
Alex Nowrasteh, Sarah Eckhardt, and Michael Howard have modeled the economic contributions of immigrants by education level.
If we use Borjas 2016’s decomposition of the educational makeup of undocumented migrants, 42% of undocumented migrants have less than a high school diploma (Nowrasteh et al NPV -$85,000), 29% have a high school diploma only (NPV $35,000), 13% have some college (NPV $141,000) and 16% have a college degree (NPV $375,000).
If an undocumented migrant had average earnings for someone of their education level, their NPV would be +$52,780, and thus, they would be a net contributor to the US.
However, we know that undocumented migrants make less than the average immigrant, so this is probably somewhat an overestimate.13 Given the legal-undocumented wage gap is about 10%, I’d adjust all contributions downwards by 10%. In this case, you’d end up with a still positive but smaller net contribution of about $40,000 per undocumented immigrant (in NPV).14
Chassamboulli and Liu 2024 uses a calibrated model to find undocumented migrants increase native welfare, but there’s not an obvious way to convert this into a dollar value.
Notably, this result does not come directly through their fiscal contributions in the form of taxes, but through increased firm profits and an overall increase in jobs.
It’s worth taking a second to dwell on this point. Undocumented migrants do pay taxes and consume benefits, as do all members of a society. But looking at only this may underestimate the impact undocumented migration has on an economy, because immigrants also consume goods and services. They participate in the economy, and it adjusts around them. Other people - who may not be undocumented - are employed in order to sell them things.
A strict Heritage-style “here are taxes paid and benefits received” accounting may underestimate what undocumented people bring into the US economy. This is discussed in more detail in Clemens 2022, but suffice to say that he finds that adding an immigrant with less than a high school diploma - the lowest-earning education group and thus, the group that pays the smallest amount of taxes - has significant and positive NPV.15
Deportations
Another way one can examine the economic impacts of undocumented migrants is to look at what happens when they aren’t there - or rather, are forcibly removed.
In 2008, a new data sharing program between local law enforcement and Immigration and Customs Enforcement (ICE) was introduced. This program, called Secure Communities, increased the likelihood that undocumented migrants would be detained and deported.16
East, Hines, Luck, Mansour and Velasquez 2022 and Velasquez and East 2024 examine the effects of this program (and its quasi-random rollout). They find that undocumented immigrants were less likely to work - but also that wages and employment for US citizens declined in response to deportations. For every 1% that employment of likely-undocumented people declined, wages for the US-born declined 0.1%.17
There are several reasons this could be true:
As firms hire more documented workers, this increases their labor costs (since as previously noted, undocumented workers are cheaper). This in turn reduces their ability to create new jobs.
Undocumented people consume goods and services, and if there are fewer of them, there is less employment among US citizens providing said goods. And indeed, East et al find that employment drops particularly in local consumption.
Velasquez and East 2024 adds another explanation: employing undocumented women - in childcare - allows US citizen women to work. The implementation of Secure Communities reduced the amount of childcare available, and thus, particularly affected the wages of women with young children.
Thus, undocumented people seem to fill an important role in the labor market. Combining these papers with the macro evidence above, I would say it is more likely than not that the presence of undocumented people increases the income of the US-born.
Crime
Money is, of course, not everything; as noted in my crime and immigration post, many people worry that undocumented immigrants are more likely than the native-born to commit crimes.
The evidence suggests they are not, though. There is surprisingly strong evidence for this: this is a consistent result across a variety of contexts. Undocumented migrants are less likely to be arrested and incarcerated in Texas and Georgia; another study in Texas finds the same results. There is no association between the size of the undocumented population and overall crime rates.
When cities adopt sanctuary policies (such that they may attract more undocumented immigrants), they do not experience an increase in crime rates (and may even experience a decrease). Nor does increasing deportations decrease crime - not in North Carolina, not in Arizona. There is simply no evidence that “[undocumented migrants are] bringing drugs. They’re bringing crime. They’re rapists” - and reasonably strong evidence that really quite a lot are “good people, I assume”.18
Alex Nowrasteh has written a longer post going through possible reasons why undocumented people are less likely to commit crime than the native-born, but I think the most likely explanation is simply that the punishments are worse for an undocumented person. If a native-born person commits a crime, they face possible fines and imprisonment. If an immigrant commits a crime, they face possible fines, imprisonment, and deportation. Undocumented migrants, in particular, have often faced considerable risk to move to the country they are now living in; it seems very rational to avoid anything that might put that in jeopardy.
It is undoubtedly true that there are individual undocumented people that commit crimes, just as there are individual citizens that commit crimes. As a group, though, undocumented people in the US seem to be pretty law-abiding once they arrive in the US.
Legalization
There is one last topic that is of relevance to the economics of undocumented migration. What if undocumented migrants are given legal status?
The US has done this via several programs. In 1986, the Immigration Reform and Control Act (IRCA) granted legal status to 1.7 million undocumented migrants; more recently, Deferred Action for Childhood Arrivals allowed some undocumented migrants who arrived in the US as children to gain work authorization.
Not surprisingly, having the ability to work legally increases one’s likelihood of having a job at all. It also increases wages among those who are working - consistent with the wage evidence previously discussed.
Legalization has somewhat more contradictory results in terms of educational attainment. Legalization does make youth more likely to finish high school, but it’s unclear what effects it has on college attendance. It’s possible that DACA recipients are more likely to attend college - but it also might make university less attractive to newly legalized immigrants.19
Even if it does, though, I don’t think this should be an argument against legalization. One framing of this might be “undocumented kids drop out of college when they get status”; another, equally-valid framing would be that “when undocumented people are given the opportunity to earn more money, become productive members of society20 and pay more in taxes, they take it”.
Indeed, this appears to be the through line for what we know about undocumented people in the US. Undocumented people want to work, and indeed they do. They work, they pay taxes, and (largely) abide by the laws of the country in which they live. The US benefits from their presence.
Many thanks to Akshay Narayanan and Ruth Grace Wong for their edits. As always, all mistakes and typos are mine.
Generally, law enforcement should not come with a body count.
Students, H-1B holders, etc.
It is not overly surprising that both Texas and California have (relatively) more undocumented people than would be expected given their share of the population
Should you want a formal model for this, Kemnitz and Mayr 2012 and Basu, Chau, and Park 2022 have them, but that’s the general intuition.
About 49% of both the native-born and legal immigrants are male. Note that the population isn’t 50/50 - this is because men die younger.
Though an undocumented woman is similarly less likely to be employed; the overall effect that undocumented people are more likely to be employed is largely because more undocumented people are men.
The ones that are not under the table / cash-in-hand / self-employment.
Around $600 billion a year
Though this includes education for minor children in the household, which seems like an odd thing to include; educational spending is not supposed to be paid for at time of consumption.
Again, largely because of education spending.
Based on current GDP/capita. This is roughly one delivery pizza in my hometown of Charlottesville, VA.
They’re, of course, also eligible for fewer benefits, so it’s entirely possible the lack of benefits outweighs the lower contributions. This is a bit of a guess; I could also see adjusting the net contribution of undocumented people upwards rather than downwards.
Note that Alex has not signed off on this analysis; this is entirely my supposition. His conclusion would likely be actually somewhat more positive; he writes that “net fiscal impact of illegal immigrants would almost certainly be more positive than that of legal immigrants at the same age and education level, but we were unable to verify that because of the small sample sizes”.
Like other papers in this vein, he finds that more educated immigrants are more net positive; the point here is that even the most “costly” type of immigrant is still net positive.
Despite the fact that Secure Communities focused on cooperation between law enforcement and ICE, it did not only result in the removal of criminal migrants. About 21% of those deported through Secure Communities had no criminal record; even for law-abiding undocumented people, the mere existence of Secure Communities increased their likelihood of deportation.
Scaling linearly, if you were to decrease the employment of 100% of likely-undocumented people by deporting everyone, wages for the US-born would decline 10%. This is unlikely to be literally true, though, as this paper looks at marginal changes; I suspect removing 8 million people from the US labor market would have general equilibrium effects that this paper cannot simulate.
Universities in the US do not require documented status, while most jobs do; it appears that some undocumented students would rather be working, but are going to school because that’s what they have access to at the time. Some of these people then drop out of university when they get work authorization.
I’m not saying that students aren’t productive members of society, but as someone with more degrees than good sense, I’m not not saying it.
Looking at the economic facts, immigrants of basically all kinds are so clearly beneficial to the US. It's hard to square the obvious economic benefits of (even undocumented) immigration with the broad public dissatisfaction. I read you and I'm like "duh, of course!" But obviously many people disagree.
I want to hear the Lauren Gilbert Steelman Argument For Why Immigration is Actually Bad for the US. In the interest of understanding the mix of economic and cultural beliefs that would lead someone to read this article but still say "nope, I think the US does not benefit from the presence of undocumented people."