Beyond Stocks and Surges: The Demographic Impact of the Unauthorized Immigrant Population in the United States

Abstract

Stock estimates of the US unauthorized foreign-born population are routinely published, but less is known about this population’s dynamics. Using a series of residual estimates based on 2000 Census and 2001–2022 American Community Survey (ACS), I estimate the components of change for the unauthorized immigrant population from 2000 to 2022 by region and country of origin. Further, I develop and present novel measures of expected duration in unauthorized status and demographic impact of unauthorized entries (i.e., person-years lived in unauthorized status). Results reveal dramatic changes over the last two decades. In the early 2000s, the unauthorized immigrant population was dominated by Mexicans who tended to remain in the United States for extended periods of time and whose demographic impact on the US population was substantial. After the 2007–2008 Great Recession, a new pattern emerged. Unauthorized migrants now arrive from across the globe, including Central America and Asia (up through 2018), and most recently from Europe, Africa, Canada, Venezuela, and other parts of South America. These new unauthorized immigrants are more likely to arrive on temporary nonimmigrant visas (which typically allow a foreigner to live and work in the United States for six years) and, with the exception of Venezuelans, spend less time in unauthorized status. Overall, the demographic impact of this new type of unauthorized migration is lower than it was two decades ago.

Stock estimates of the total unauthorized foreign-born population in the United States have hovered around 11 million since 2008, give or take a few hundred thousand, and only recently ticked up to 11.3 million in 2023. Beneath this apparent stability, however, is a dynamic and evolving population. Even before the surge in unauthorized border crossings started in 2021 (Ruiz Soto 2022), there were 459,000 migrant encounters with US border patrol on average each year between 2008 and 2020 (U.S. Customs and Border Protection 2024). Such statistics suggest a steady stream of newly arrived unauthorized immigrants. Yet of course, entries can be offset by exits as unauthorized immigrants leave the country voluntarily, are deported, legalize, or die. In this article, I present new estimates of the components of population change for the unauthorized immigrant population from 2000 to 2022, and I use these estimates to gauge the “demographic impact” of unauthorized immigration on the US population.

Estimates of the components of change are important because they can be used to estimate how long, on average, unauthorized immigrants remain living in the United States, which in turn is related to the demographic impact of unauthorized immigration on the US population. All things equal, groups with higher out-migration rates stay for shorter periods of time. They may fulfill short-term labor needs but are less likely to be integrated socially and economically. In contrast, groups with low out-migration rates tend to stay longer. These longer term settlers are more likely to work, purchase homes, raise children, and age for decades in the United States. They are likely to contribute to and have a long-lasting impact on host communities. Overall, the demographic impact of unauthorized immigration on the US population depends on the group’s size (i.e., the number of entrants into unauthorized status) and their average duration in unauthorized status, with large groups with long expected durations making the greatest and longest lasting contributions to the size, composition, and growth of the US population.

Prior research has explored the migration dynamics of Mexican and, to a lesser degree, Central American immigrants (Massey and Pren 2012; J. Passel et al. 2012; J. S. Passel and Cohn 2009; Villarreal 2014), but little is known about the migration dynamics of other national origin groups. This research gap has grown in importance since 2008 as the share of unauthorized immigrants from Mexico and Central America declined (J. Passel et al. 2012; Van Hook, Gelatt, and Ruiz 2023; Villarreal 2014). Here, I produce estimates of the components of change, expected duration, and demographic impact for the unauthorized immigrant population from 2000 to 2022 for 27 different country and region-of-origin groups.

Background

Media and political discussions about unauthorized immigration pay outsized attention to unauthorized immigrant entries, for example, calling attention to “surges” or “chaos” at the US–Mexico border. Although such inflows create enormous logistical and humanitarian challenges (Ruiz Soto, Putzel-Kavanaugh, and Meissner 2024), they exclude the large numbers of people who become unauthorized because they overstay their nonimmigrant visas (U.S. Customs and Border Protection 2023), which allow foreigners to live and work in the United States for a limited period of time, typically six years. Additionally, a focus on entries—whether as border crossings or visa overstays—overlooks the overall impact unauthorized immigration is likely to have on the US population because it does not account for how long unauthorized immigrants spend in the United States in an unauthorized status. I conceptualize “demographic impact” in simple population terms as the number of person-years expected to be lived in unauthorized status by unauthorized entrants. In this conceptualization, “demographic impact” is a measure of the demographic footprint of a group. Although it would be valuable to account for unauthorized immigrants’ economic or social contributions and costs in an assessment of impact, such an undertaking extends well beyond the scope of this paper.

Assessments of demographic impact require a deep understanding of the population dynamics of unauthorized immigrant populations. Some of the most influential research on the population dynamics of the unauthorized immigrant population has been based on analyses of Mexican Migration Project (MMP) and Latin American Migration Project (LAMP) data. These surveys collected detailed information on all international trips to the United States from household heads in selected migrant-sending communities in Mexico and other parts of Latin America¹ (Massey, Durand, and Malone 2002). Analyses of the MMP data, the longer running migration survey of these sources, have shown that Mexican immigrants were engaged in temporary circular migration during the 1970s and 1980s. Making repeated trips to the United States for short work stints brought income and wealth to poor Mexican families, and allowed employers to hire and then dispense of labor as needed, as frequently occurs in the case of seasonal agricultural and construction work and during economic downturns (Hanson 2009). Because this form of migration tended not to result in permanent settlement, the unauthorized immigrant population from Mexico grew slowly during this period. In other words, the inflow of unauthorized immigrants from Mexico did not have a large demographic impact on the US population.

This pattern of circular labor migration was disrupted during the 1990s and early 2000s by the build-up of enforcement along the US–Mexico border. Ironically, rather than slowing new migration, it disincentivized return migration (Massey, Durand, and Malone 2002; Massey, Durand, and Pren 2015). The consequence was longer durations of stay and the rapid build-up of the population during the late 1990s and early 2000s (Rosenblum et al. 2012; J. S. Passel and Cohn 2010; Reyes 2001, 2004). Then, when both in- and out-migration slowed following the Great Recession, the Mexican unauthorized population started to age in place. Recent projections indicate that the number of unauthorized Hispanic immigrants entering old age without the support of public programs like Social Security and Medicare is likely to grow dramatically in the next two decades (Ro, Van Hook, and Walsemann 2022). Because of their longer duration of stay, the demographic impact of Mexican unauthorized immigrants increased dramatically between the 1980s and late 1990s.

The MMP data revealed important insights about the evolving dynamics of the migration flows between Mexico and the United States. However, over the last 10 years, the national origin composition of unauthorized immigrants has diversified (Van Hook, Gelatt, and Ruiz, 2023). Between its peak in 2007 and 2022, the number of Mexican-born unauthorized immigrants dropped from 7.7 to 5.1 million, and the Mexican share of the total unauthorized population dropped from 63 percent to 45 percent. At the same time, the share from all other regions increased. The percentage from Central America rose from 12 to 20 percent; the Caribbean from 2 to 4 percent; South America from 7.5 to 11 percent; and Asia from 7.4 to 10 percent (Ruiz Soto, Gelatt, and Van Hook 2024). Less is known about the population dynamics and potential demographic impact of these growing migration streams, although they likely differ dramatically from the patterns seen among Mexicans (Massey 2020a).

Some of the “new” unauthorized immigrants may have a large demographic impact on the US population. In particular, asylum seekers from Latin America, primarily Venezuela and Central America, increased dramatically in the last decade. Many are fleeing violence, political oppression, and devastating inflation and unemployment in their countries of origin, and they are less likely to return voluntarily to their home country or engage in circular migration (Massey 2020b). If they are not granted asylum or their court hearings are delayed yet remain living in the United States, they may remain in the United States as unauthorized immigrants for many years.

In contrast, others of the new unauthorized immigrants may have smaller demographic impact, akin to the early decades of Mexican migration. In particular, unauthorized immigrants from India or other parts of Asia and Africa, often arrive on a temporary nonimmigrant visa, such as a student, high skilled worker, or visitor, but then stay beyond the terms of their visa. They may spend less time in an unauthorized status than other groups. For one, they may not have incentives to remain in the country for long (Warren 2019; Warren and Kerwin 2017). Many are highly educated and could have employment opportunities outside the United States. Second, they may spend more of their time in the United States in a legal status. During their first few years of US resident, many will have lived and worked as a legal nonimmigrant. Then, after overstaying their visa, they may have more opportunity to legalize their status given their greater educational attainment and potential access to high-skilled jobs whereby they can be sponsored by a US employer.

The current analysis

I explore the population dynamics and gauge the demographic impact of unauthorized immigration to the United States. First, I update prior research on population dynamics for Mexico and extend it to other countries and regional groups. My estimates enable systematic comparisons of inflows and outflows for the multiple unauthorized immigrant groups now living in the United States. Second, I use the estimated inflow and outflow rates to produce new period lifetable measures of the expected duration of stay, accounting for both the total duration and time spent in unauthorized status. These lifetable measures have advantages over direct observations of the average duration of US residence because they are unaffected by year-to-year fluctuations in the size of entry cohorts. Finally, I use the estimates of inflows and expected duration in unauthorized status to gauge the demographic impact of unauthorized immigrant entrants.

Like the dynamics seen in the MMP data, I expect to find lower rates of out-migration, longer durations, and greater demographic impact of unauthorized immigrants from Mexico and among groups with a lot of humanitarian migrants (e.g., Central Americans and Venezuelans) than other groups. That said, the demographic impact for these groups could change as unauthorized entries continue to decline among Mexicans but increase among Central Americans and Venezuelans. Conversely, I expect to find higher rates of out-migration, shorter durations, and lower demographic impact among other groups, particularly those with a large share of visa overstayers. Finally, establishing the social and economic causes of the trends in the components of change is not a goal of this paper. Nevertheless, the results, if plausible, should be consistent with prior research and theory. For example, migration flows should be responsive to major economic shocks, such as the Great Recession of 2007–08 (Durand 2016).

Methods

Data

I estimated the components of change, expected duration in unauthorized status, and demographic impact by analyzing the 2000–2022 residual estimates of the unauthorized immigrant population that were published by the Migration Policy Institute (MPI) (Van Hook et al. 2023). These estimates are not individual-level measures, but rather are the estimated number of unauthorized immigrants for birth and entry cohorts for 27 different countries or regions for each year from 2000 to 2022. MPI’s estimates were made by subtracting the number of immigrants who were legally admitted since 1982 from the total number of foreign-born individuals who arrived since 1982, as represented in 2000 Census and 2001–2022 American Community Survey (ACS) data. The estimates include adjustments for coverage error of the unauthorized population in the ACS, and for heaping in year-of-arrival reporting. I describe the MPI methodology and its key assumptions in greater detail in Appendix 1 of the Supporting Information.

The unauthorized immigrants in these estimates include those who entered the country without inspection and those who arrived legally with temporary visas (e.g., for agricultural workers H-2A, nonagricultural workers, H-2B, and students, F1) but overstayed or otherwise violated the terms of their visas. Also included among the unauthorized are people who have received an official, temporary reprieve from deportation but lack a durable legal status, for instance: Temporary Protected Status (TPS) recipients, Deferred Action for Childhood Arrival (DACA) participants, and asylum applicants with work authorization. For estimation purposes, MPI limits the unauthorized population to those who arrived in the country in 1982 or later, with the rationale that most immigrants who arrived before 1982 would have legalized because they were eligible for amnesty under the Immigration Reform and Control Act of 1986. Note also that the unauthorized population does not include the US-born children of unauthorized immigrants. According to the 14th Amendment of the US Constitution, citizenship is automatically granted to all persons born within US borders.

The quality of the ACS data has been inconsistent over time. This creates special challenges for my analysis of population dynamics, which depends on year-over-year changes in cohort size. The ACS was not implemented at full scale until 2005, and the 2020 ACS was found to be of low quality due to difficulties in data collection during the first year of the COVID-19 pandemic, leading the U.S. Census Bureau to issue warnings about its use (United States Bureau of the Census 2021). I therefore dropped all ACS estimates based on 2001–2004 and 2020 ACS data, and linearly interpolated the size of the size each cohort between 2000 and 2005 and between 2019 and 2021. In other words, all estimated trends provided in this paper for 2000–2005 and 2019–2021 represent average annual changes within these time frames.

Components of change

To estimate the components of change for the unauthorized foreign-born population, I reshaped the MPI data into country/region birth- and entry-cohorts. Each cohort can be followed over time from 2000 (or the year of entry of the cohort if it arrived after 2000) to 2022. There are distinctive patterns in the build-up and decline of cohorts. For groups with few visa overstayers, entry cohorts typically grow quickly in their first year in the country, signifying the entry of new arrivals, as shown in Figure 1a for Mexicans who arrived in 2005. The cohorts tend to decline starting around the second year of residence. The pattern is different for groups with large numbers of visa overstayers. Visa overstayers are people who originally entered the country legally on a nonimmigrant visas but then stayed longer than permitted. Temporary work visas (e.g., H-1B) allow people to live and work in the United States for up to six years, although extensions are possible under certain circumstances.² As a consequence, visa overstayers typically become unauthorized after having lived in the country for six to eight years. This is depicted in the case of Indians who arrived in 2005 in Figure 1b. From the first to the ninth year, the cohort tends to grow slowly as visa overstayers add to the unauthorized immigrant population. After the ninth year, the cohort declines in size as members die, legalize, or move away.

FIGURE 1.

Cohort change among unauthorized immigrant arrivals for selected groups

I analyzed changes in the size of each birth- and year-of-entry-cohort to infer the components of change for the cohort, which I then summed across cohorts to produce components of change estimates by country/region and year. Entries into the unauthorized population include new arrivals and likely visa overstays. I estimated the number of new arrivals as the growth in the size of the cohort during its first full year in the country. For example, for the 2016 entry cohort, new entries are those in the 2017 ACS who arrived during the 2016 calendar year. I estimated the number of likely visa overstays as the additional positive growth in the cohort (after accounting for deaths) between its second and tenth year in the country. If there was no additional growth in the cohort (i.e., the cohort declined), the number of likely visa overstays was set to zero. Conversely, if the additional growth was greater than MPI’s estimate of visa overstays for the group³, I adjusted the estimate downward to match MPI’s estimate. In spite of these calibrations, the number of likely visa overstayers is probably underestimated because some unauthorized immigrants arrive on tourist visas. These visas expire after six months, so they would have become unauthorized within their first year of residence and would therefore be classified as new arrivals in my estimates.

Exits from the population included deaths, legalizations and return migration. I did not have enough information to separate legalization from return migration, so I combined them into a single indicator that I refer to as “net outflows”. I estimated the number of expected deaths by multiplying the number of person-years lived in each cohort by the gender-, age-, and year-specific death rate (i.e., matching on gender, age, and year). I relied on the foreign-born death rates that MPI uses to create its estimates (see Appendix 1 of the Supporting Information). Note that deaths make up only a small share (6 percent) of the total outflows from 2000 to 2022.

Relying on standard residual methods for estimating net migration (Ahmed and Robinson 1994; Leach and Jensen 2013), I estimated net outflows as a residual of the other components of change: Net outflows = total net growth + expected deaths - new arrivals - visa overstays. Most of the estimates of net outflows were negative, with negative values signifying out-migration or legalizations. However, the estimates were positive in a few instances, likely signifying returns of emigrants (i.e., when people return once again to the United States after originally arriving and leaving in earlier years) or estimation error. It is important to recognize that these components of change—both entries and net outflow—do not capture the rapid entries and exits that occur every day at the US–Mexico border. If an unauthorized immigrant crosses the border but is deported quickly—within hours, days, or even weeks—they are unlikely to have participated in the ACS and will not appear in my estimates.

Lifetables

Next, I combined the death and net outflow rates to develop lifetable measures of the expected duration in unauthorized status and total expected duration for three time periods: 2000–2006, 2007–2014, and 2015–2022. The lifetables model the decline in the size of a synthetic cohort from entry into unauthorized status until death, legalization, or return migration. I produced the lifetable estimates separately by gender and age-at-arrival (five-year groupings), and then averaged the estimates (while weighting on cohort size) to create summary statistics for each country/region grouping.

As an illustration, Table 1 shows lifetables for two groups of male Mexican immigrants in 2015–2022: those who arrived at age 20–24 and those who arrived at age 40–44. As shown in the first and second columns, each cohort enters the lifetable at their age at arrival. For simplicity, all are assumed to enter the country at the beginning of the age-at-entry category, and the cohorts are then followed over time as they age (column 1) and spend more time in the United States (column 2). New arrivals are added to the cohort in the first year of US residence (column 3), while visa overstays are added in subsequent years (column 4).

TABLE 1.

Examples of life tables of time lived in unauthorized status by Mexican male immigrants who arrived at age 20 and 40, 2015–2022

		Entries into unauthorized status (thousands)				Outflows
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Age x	Duration, tx	New arrivals, an,x	Overstay temp. visa, at,x	All entries, ax	Number in unauthorized status, lx	Deaths, dx	Other outflows, fx	Person-years lived in interval, Lx	Person-years lived age x+, Tx	Average person-years in unauthorized status, eu	Average person-years lived on a temporary visa, et
a. Mexican male immigrants, arrived age 20
20	0	55.8	—	55.8	55.8	0.1	6.2	58	1584	23.1	0.7
21	1	—	10.8	10.8	60.4	0.3	1.1	241	1526
25	5	—	1.3	1.3	60.2	0.4	13.8	268	1285
30	10	—	0.7	0.7	46.8	0.4	6.3	217	1017
35	15	—	—	—	40.1	0.4	6.2	184	800
40	20	—	—	—	33.5	0.5	7.7	147	616
45	25	—	—	—	25.3	0.6	3.1	118	468
50	30	—	—	—	21.7	0.6	6.5	91	351
55	35	—	—	—	14.6	0.7	0.9	69	260
60	40	—	—	—	13.0	0.9	1.0	60	191
65	45	—	—	—	11.1	1.0	1.6	49	131
70	50	—	—	—	8.4	1.2	1.2	36	82
75	55	—	—	—	6.1	1.3	0.8	25	46
80	60	—	—	—	3.9	1.6	0.5	15	21
85	65	—	—	—	1.9	1.9	0.0	6	6
b. Mexican male immigrants, arrived age 40
40	0	23.3	—	23.1	23.1	0.1	3.1	22	398	14.2	1.5
41	1	—	1.1	1.1	21.0	0.3	4.0	80	376
45	5	—	2.2	2.2	18.9	0.4	3.8	87	296
50	10	—	1.1	1.1	15.9	0.4	6.4	63	209
55	15	—	0.4	0.4	9.5	0.4	1.0	44	146
60	20	—	—	—	8.0	0.5	1.4	35	102
65	25	—	—	—	6.0	0.5	1.2	26	67
70	30	—	—	—	4.3	0.6	0.6	18	42
75	35	—	—	—	3.1	0.7	0.4	13	23
80	40	—	—	—	2.0	0.8	0.2	7	10
85	45	—	—	—	1.0	1.0	0.0	3	3

Each cohort (column 6) initially grows as new arrivals and visa overstays are added, but then declines as some members die (column 7) and or leave the unauthorized population due to legalization or out-migration (column 8). Deaths are the number expected if the cohort experienced the average gender-, age-specific death rates of the foreign-born population. Other outflows are the number expected if the cohort experienced average net outflow rates by duration and age, gender, and country/region. As in multidecrement lifetables, the synthetic cohort experiences death and other outflow rates simultaneously throughout each age interval. Thus, for each age interval x to x + n, the numbers of deaths, d_x, and other outflows, f_x, account for competing risks of mortality and other outflows, and are calculated as:

$$d_x=\frac{l_x{m}_x}{-m_x-g_x}{e}^{\left(\left(-m_x-g_x\right)n-1\right)}$$

$$f_x=\frac{l_x{g}_x}{-m_x-g_x}{e}^{\left(\left(-m_x-g_x\right)n-1\right)},$$

where l_x = size of synthetic cohort at age x; m_x = mortality rate (positive-signed); g_x = net outflow rate (with positive-signed rates indicating flows out of the unauthorized population); and n = width of age interval (years).

Columns 9–12 concern the duration lived by the cohort in various statuses:

L_x = person-years in unauthorized status within each age interval ($L_x=\frac{n}{2}\left(l_x+l_{x+n}\right)$; column 9);

T_x = person-years in unauthorized status from age x to the last age interval ($T_x={\sum }_x^{85}{L}_x$; column 10);

e_u = expected duration spent in unauthorized status, averaged across all entries ($e_u=\frac{T_0}{\sum a_x}$; column 11); and

e_t = duration spent on a temporary visa prior to becoming unauthorized, averaged across all entries ($e_t=\frac{\sum a_{t,x}\frac{2t_x+n}{2}}{\sum a_x}$; column 12).

The expected duration in unauthorized status (e_u) is a period measure that indicates the expected number of years an unauthorized immigrant would live in unauthorized status in the United States if he/she experienced all current death and outflow rates for their group. For visa overstayers, the expected duration in unauthorized status excludes time spent in the United States on a temporary visa prior to becoming unauthorized (e_t). However, I also present estimates of the total expected duration (e = e_u + e_t), which includes this time.

Demographic impact

I multiplied the estimates of unauthorized entries (new arrivals + visa overstays) by the expected duration in unauthorized status to create estimates of demographic impact by year and country/region of birth. As noted earlier, demographic impact is defined as the total number of person-years expected to be lived in unauthorized status in the United States by unauthorized entrants. It is a simple population measure. For example, it does not account for the impact unauthorized immigrants might have on the US economy, public health, or fertility levels, and does not account for immigrants’ age at entry, educational attainment, or skills. Additionally, it indicates the entry cohort’s potential population footprint but is not predictive of the future. For example, the demographic impact of an entry cohort could decrease in the future if rates of return migration, deportations, or legalizations increased. This is similar to the limitation of period life expectancy measures, which are not predictions of how long people will live but instead are indicators of how long they would live if they experienced all current age-specific mortality rates. To place the estimates in context, I compare them to the demographic impact of the US-born population, which is calculated as the product of the number of US births and US life expectancy.

Uncertainty and sensitivity analyses

My estimates of the components of change, expected duration in unauthorized status, and demographic impact build upon MPI’s residual estimates of the unauthorized population, which themselves are sensitive to assumptions about coverage and emigration (Van Hook et al. 2021), and are subject to sampling error. I conducted sensitivity analyses to gauge the likely impact of these factors.

First, the residual estimates assume, based on empirical evidence, that a large share of the unauthorized foreign-born population are represented in the ACS (see Appendix 1 of the Supporting Information). MPI adjusts its estimates upward to account for modest levels of coverage error—an average of 13.1 percent for its 2000–2022 series—with higher coverage error assumed for the early 2000s (about 22 percent before 2007) and recent arrivals (20 percent for those with less than 10 years of US residence) (see Appendix 1 of the Supporting Information for more information on MPI’s coverage assumptions). However, such adjustments may be insufficient. For example, Van Hook et al.’s (2021) review found coverage error estimates for some groups as high as 30 percent to 40 percent (e.g., Mexican men in the early 2000s). If coverage error were higher than what MPI accounts for, this would mean that the size of the unauthorized population is higher than estimated. More importantly, if MPI underestimates coverage error especially for new arrivals, this would downwardly bias estimates of the number of entries and upwardly bias estimates of their duration of stay (because the size of each cohort would appear to decline less over time). Since demographic impact is calculated as the product of expected duration and the number of entries, and these two factors would be biased in opposite directions, it is unclear how coverage error assumptions would affect estimates of the demographic impact. To gauge the size and direction of these effects, I re-estimated the outcomes after reducing/increasing coverage error assumptions for each demographic group by 50 percent (i.e., with coverage error ranging from a low of 6.5 to a high of 20 percent on average, and from 10 percent to 30 percent for new arrivals).

Second, the residual estimates depend on accurate estimates of the size of the legally-resident foreign-born population, which is calculated as the number admitted as legal permanent residents and humanitarian entrants minus the estimated number of deaths and emigrants. Deaths are a small component and known with relative precision, but the numbers of emigrants are more uncertain (Van Hook et al. 2021). MPI relies on emigration rates that were estimated by observing variations in migration behavior among (mostly legal) immigrants in the Social Security earnings file by country of birth, duration of residence, age, and sex, and adjusted to account for annual levels and trends in emigration observed in the ACS (see Appendix 1 of the Supporting Information). On average, the assumed emigration rate is 0.9 percent (1.2 percent for those with less than 10 years of US residence and 0.7 percent for longer term residents). Yet Van Hook et al. (2021) reviewed the literature on foreign-born emigration and found that estimates of the emigration rate for legally-resident immigrants typically vary from one another with a standard deviation of about .4 percent. If emigration rates of the legal foreign-born are higher than MPI assumes, this would mean that their estimates of the unauthorized population are too low, particularly those who have lived in the country for longer periods of time. Accordingly, my estimates of expected duration in unauthorized status and demographic impact would be too low. The opposite would be true if emigration were lower than MPI assumes. To assess the size of these effects, I re-estimated the key outcomes after reducing/raising assumptions about emigration by 50 percent, or a range 0.4 percent to 1.3 percent coverage error on average (0.6 to 1.8 percent for new arrivals).

Finally, the estimates rely on a sample, the ACS; involve the analysis of numerically small groups (e.g., birth and entry cohorts for twenty-seven different country/region of birth groups); and therefore are subject to sampling variation. To gauge the size of sampling error, I bootstrapped all estimates (i.e., sampling with replacement) and used the standard deviation across 500 iterations as an estimate of the standard error. In charts and tables, I present 95 percent confidence intervals of the estimates (±1.96 × SE).

Results

Components of change

Estimates of the components of change for the unauthorized immigrant population are shown in Figure 2. Bars dipping below the horizontal axis represent exits from the population; deaths are shown in solid black and other outflows in gray. Bars extending above the axis represent entries, with new arrivals shown in dark green and likely visa overstays in light green. The red line indicates net growth (entries minus exits). The background (the gray silhouette) depicts the estimate of the population size; it is measured on a separate scale on the right. Figure 2 also includes error bars indicating the 95 percent confidence intervals for the estimates of inflows, outflows, and net growth, but the standard errors are so small that they are difficult to see on the chart.

FIGURE 2.

Components of change in the unauthorized foreign-born population, 2000–2022

The components of change reveal distinctive patterns over the last 22 years that correspond as expected with periods of economic growth, shocks, and recovery. From 2000 to 2007, entries exceeded exits and net growth was large and positive, exceeding half a million per year. This growth was part of a much longer trend (not seen in my results) in which the unauthorized immigrant population grew nearly continuously during the period of economic growth from the mid-1990s until the mid-2000s (J. S. Passel and Krogstad 2023). Consistent with research on Mexican unauthorized immigrants (Villarreal 2014), this growth spell was abruptly disrupted by the Great Recession in 2007 when large numbers of unauthorized immigrants left the country, new entries dropped, and the total population fell by over 1.5 million. As shown in Figure 3 and Appendix 2 of the Supporting Information, unauthorized migrant populations from nearly all regions and countries declined due to a sudden increase in exits and declines in entries in 2007 and 2008. The population decline was especially abrupt for unauthorized immigrants from Mexico, Central America (El Salvador, Guatemala, Honduras), Brazil, the Philippines, and South Korea. The 95 percent confidence intervals for the different regions and countries in Figure 3 and Appendix 2 of the Supporting Information tend to be larger than for the overall unauthorized population (Figure 2), particularly for small sized populations (e.g., “Rest of Asia”), but nevertheless indicate that the trends, particularly the sudden declines in net flow in 2007–2008, are robust to sampling error.

FIGURE 3.

Components of change for the unauthorized foreign-born population from Mexico and other world regions, 2000–2022 (error bars represent 95 percent confidence intervals)

After the Great Recession, the unauthorized foreign-born population levelled off as entries were balanced by exits throughout the 2010 decade. We do not see any abrupt changes during this period, even as the country transitioned from the Obama (2009–2016), to the Trump (2017–2020), and to the Biden presidencies (2021–). It was not until 2021 that the population grew beyond 11 million, reaching 11.3 million in 2022, but not because of a surge in entries (as is a common narrative), but because fewer people departed the country. Although it is unclear what accounts for this change, it may be related to the recovering economy following the COVID-19 pandemic. It is important to note that these trends do not extend past mid-year 2022 and therefore do not account for the surge in border encounters between 2022 and 2024 (Ruiz Soto 2022; U.S. Customs and Border Protection 2024).

Despite the slow growth from 2007 to 2022, there have been large shifts in the national origin composition of unauthorized immigrants. As documented in prior research, unauthorized migration flows were once dominated by migration from Mexico, but this is no longer the case. After the Great Recession, the unauthorized populations from most regions and countries recovered quickly and experienced positive net growth throughout the 2010 decade. However, the Mexican unauthorized population did not. Its population declined every year since 2007 (Figure 3). Mexican unauthorized immigrants continue to be the largest single national origin group, but this is the result of earlier growth during the 1990s and 2000s and not its current growth.

Less well-known is that the composition of unauthorized migration has continued to shift since 2015, with some country-of-origin groups joining those from Mexico in experiencing population stability or decline, and other origin groups growing quickly. To summarize recent growth patterns, I plotted inflows (new arrival plus visa overstay rates) against outflows (death plus net outflow rates) for each country/region of origin for the 2015–2022 time period in Figure 4. The marker for each country/region of origin group is scaled according to the 2015–2022 average annual size of its unauthorized immigrant population. Origin groups above the diagonal are increasing in size and those below the diagonal are shrinking. Additionally, those clustered close to the diagonal in the lower left corner are relatively static and are likely to age in place. Those in the upper right corner tend to experience a lot of “churning,” whereby exits are continually replaced by new arrivals.

FIGURE 4.

Entry and exit rates from the unauthorized foreign-born population, 2015–2022

NOTE: The size of the markers are scaled to population size.

Outflows exceed inflows for only a handful of origin groups: Mexico, South Korea, Peru, Vietnam, and “rest of Asia”. The unauthorized populations from these countries and regions have been declining in size in recent years. Notably, the Mexican unauthorized population has become static; it has some of the lowest in- and outflow rates of all countries/regions of origin, meaning that its unauthorized population is likely to age in place in the United States (Ro, Van Hook, and Walsemann 2022). Many groups from Asia also have very slow growth, including those from China, the Philippines, and India; the unauthorized population from Asia as a whole has declined or remained steady every year since 2017. Also notable are unauthorized immigrants from El Salvador, Guatemala and Honduras. Inmigration recovered for these groups following the Great Recession, showing sustained growth from 2008 until about 2018 (Appendix 2 of the Supporting Information). Since 2018, however, inflows declined and outflows increased, resulting in much slower growth. By 2021–2022, net growth for the unauthorized populations from Guatemala and Honduras was close to zero, and net growth for unauthorized immigrants from El Salvador was negative (Appendix 2 of the Supporting Information).

In contrast, inflows exceed outflows for unauthorized populations from all the countries and regions of origin shown above the diagonal in Figure 5, including Europe, Canada, Africa, and many parts of South America. The Venezuelan unauthorized immigrant population stands out as experiencing the most rapid growth, with very high inflow and low outflow rates (see detailed trends in Appendix 2 of the Supporting Information, which depict a rapid increase in the unauthorized Venezuelan population starting around 2015). In the last seven years, many people fled Venezuela following massive human rights violations of the Nicolás Maduro regime. Venezuelans were granted TPS by the U.S. Department of Homeland Security in 2022 but are still classified as unauthorized immigrants by organizations that produce estimates of the unauthorized immigrant population because they continue to lack a durable residency status.

FIGURE 5.

Expected duration (total and in unauthorized status) and likely visa overstay rate among unauthorized immigrants, by country/region of birth, 2015–2022 (error bars = 95 percent confidence intervals)

Expected duration in unauthorized status

The size and growth of an unauthorized immigrant population is only one factor determining a group’s demographic impact. It is also important to account for expected duration of stay. Figure 5 plots each origin group by its expected duration for the 2015–2022 period. The figure displays two measures: the expected duration in unauthorized status is shown in the bars shaded in a different color for each region of the world, and the total duration in the United States (i.e., time spent prior to overstaying a visa and time as unauthorized), is shown in the gray bars. The error bars depict 95 percent confidence intervals. To illustrate the correspondence between visa overstays and duration, I overlaid the chart with estimates of the percentage of entries that are likely visa overstays, marked by white circles.

Unauthorized immigrants from Mexico, Guatemala, Honduras, El Salvador, and Venezuela have long expected durations. If they were to experience current net outflow and death rates throughout their lives, they could expect to live in the United States for a total of 23 years (ranging from 21 to 24 years) before leaving, dying or adjusting to Legal Permanent Residence (LPR) status. Of these 23 years, they could expect to spend 21 of them (ranging from 19 to 23 years) in unauthorized status, or about 94 percent of their time in the country. This is because relatively few in this group (23 percent on average) appear to be visa overstayers and therefore did not spend much time on nonimmigrant visas.

Groups with shorter expected durations come from a variety of regions: the Caribbean, South America (aside from Venezuela), Asia, Europe, Canada, Oceania, and Africa. If they were to experience current net outflow and death rates throughout their lives, they could expect to spend an average of 11 years in unauthorized status (ranging from 3 to 17 years), and 14 years total (ranging from 6 and 20 years) before leaving, dying or adjusting to LPR status. A large share is likely to have overstayed a nonimmigrant visa (41 percent on average) and they are expected to have spent about 25 percent of their time in the country on a temporary nonimmigrant visa prior to overstaying.

Demographic impact

How large is the impact of unauthorized immigration on the US population? As discussed earlier, the answer depends on a combination of the number of entries and expected duration in unauthorized status. I plot estimates of demographic impact (i.e., expected person-years lived by unauthorized entries) by major region of birth from 2000 to 2022 in Figure 6⁴. I also present estimates for all 27 countries/regions in Table 2 for three time periods, 2000–2006, 2007–2014, and 2015–2022.

FIGURE 6.

Demographic impact of unauthorized immigrant entries, by region of origin, 2000–2022

NOTE: Demographic impact is the expected person-years lived by new unauthorized immigrants. It is calculated as the number of entries (new arrivals + visa overstays) times the expected duration in unauthorized status. Error bars represent 95 percent confidence intervals.

TABLE 2.

Average annual demographic impact of unauthorized immigrant entries, by country/region of origin

	Demographic impact (expected person-years lived by unauthorized entries in thousands)			Margin of error (1.96 × standard error)
	2000–2006	2007–2014	2015–2022	2000–2006	2007–2014	2015–2022
Mexico	15,391	3,049	2,463	815	100	82
Central America	2,012	1,635	2510	142	65	84
El Salvador	615	602	565	74	46	36
Honduras	463	446	823	54	28	46
Guatemala	832	520	980	105	36	58
Nicaragua	28	41	96	4	3	20
Rest of Central America	74	27	46	10	2	4
Caribbean	334	321	478	41	22	34
Haiti	83	80	106	13	8	12
Rest of Caribbean	251	241	372	39	20	31
South America	1,266	625	1,859	75	17	49
Colombia	147	136	353	13	9	22
Venezuela	68	99	722	7	5	34
Brazil	342	80	370	28	5	17
Ecuador	229	113	187	28	8	17
Peru	219	89	76	32	6	8
Rest of South America	260	108	151	53	7	12
Asia	1,221	1,611	1,252	75	76	67
China	135	351	374	29	43	41
Philippines	252	346	235	43	22	20
S. Korea	384	129	63	32	7	7
India	257	588	525	42	59	44
Vietnam	34	108	36	8	9	5
Turkey	40	11	11	7	6	5
Rest of Asia	118	77	8	15	14	5
Europe, Canada, Oceania	932	566	792	57	21	46
United Kingdom	106	74	81	14	6	11
Germany	77	58	97	9	4	9
Rest of Europe	457	219	416	44	16	43
Canada	233	128	98	30	10	10
Oceania	59	86	99	11	5	6
Africa	380	220	661	38	22	51
Total	21,535	8,027	10,015	836	147	844
Demographic impact of U.S. births	316,997	319,480	297,264

NOTE: Demographic impact is calculated as the average annual number of entries (new arrivals + visa overstays) times the expected duration in unauthorized status. The demographic impact of U.S. births is shown for comparison; it is the average annual number of births times the life expectancy at birth.

In the early 2000s, the demographic impact of unauthorized entries from Mexico was high, dwarfing other regions of origin. Prior to the Great Recession between 2000 and 2006, nearly half a million Mexican immigrants were added to the unauthorized population each year (522,000, or about 51 percent of all unauthorized entries), and each had an expected duration of 29 years in unauthorized status, meaning that each years’ entry cohort was likely to contribute 15.4 million (95 percent CI ±0.8 million) person-years to the US population. This was 71.5 percent of the total for all unauthorized immigrants, and 4.9 percent of the demographic impact of all US births.

However, new unauthorized migration from Mexico plummeted after 2007, leading to a much lower demographic impact: 3.0 million (95 percent CI ±0.1 million) person-years per year during the 2007–2014 period (38 percent of the total), and 2.5 million (95 percent CI ±0.08 million) person-years per year during the 2015–2022 period (25 percent of the total). In supplemental analyses, I decomposed the 2000–2006 to 2015–2022 change and found that 81 percent of the reduction in Mexican’s demographic impact was attributable to declines in entries. Only 19 percent was due to declines in the duration in unauthorized status⁵. At the same time, the demographic impact of unauthorized entrants from Central and South America increased in absolute and percentage terms largely due to growth in entries among Guatemalan, Honduran, and Venezuelan unauthorized immigrants. For these groups, growth in entries accounted for 79 percent of the increases in demographic impact. Currently, the demographic impact of unauthorized immigrants is much more evenly distributed across regions of origin, with Mexicans contributing 24.6 percent; Central Americans 25.1 percent; Caribbeans 4.8 percent; South Americans 18.6 percent; Asians 12.5 percent; Europeans, Canadians, and Oceanians 7.9 percent; and Africans 6.6 percent.

Finally, the demographic impact for unauthorized immigration from all regions altogether for the 2015–2022 period is about half (47 percent) that of the 2000–2006 period, a decline that is partially attributable to declines in entries (62 percent) and partially due to declines in duration in unauthorized status (38 percent). Overall, this change can be understood as a gradual replacement of a longer duration group (Mexicans) with smaller, shorter duration groups (immigrants from regions other than Mexican and Central America). Unauthorized Venezuelans are an exception to this trend. They are a fast-growing group with long expected durations, but they currently make up only 2 percent of the unauthorized foreign-born population and therefore do not exert much influence on the overall patterns yet.

Sensitivity analyses

The estimates presented here depend on residual estimates of the unauthorized population, which in turn are sensitive to underlying assumptions about coverage and emigration. Table 3 shows how the 2015–2022 estimates of expected duration in unauthorized status vary when assumptions about coverage and emigration are decreased and increased by 50 percent, and Table 4 shows the same for the 2015–2022 estimates of demographic impact.

TABLE 3.

Sensitivity of expected duration in unauthorized status to assumptions about coverage error and emigration, 2015–2022

		Coverage error of unauthorized foreign born		Emigration of legal foreign born
	Point estimate	Low (−50%)	High (+50%)	Low (−50%)	High (+50%)
Mexico	22.0	22.7	20.7	19.9	22.9
Central America	20.7	20.5	20.2	20.0	21.3
El Salvador	21.2	20.5	20.3	20.4	22.0
Honduras	20.9	21.4	20.7	20.7	21.6
Guatemala	23.0	22.5	22.4	21.5	23.6
Nicaragua	10.8	11.1	10.8	10.4	11.0
Rest of Central America	8.6	8.3	8.2	7.9	9.0
Caribbean	13.1	13.7	13.8	9.8	15.3
Haiti	9.3	9.5	9.2	8.1	10.1
Rest of Caribbean	14.8	15.5	15.7	10.9	17.2
South America	15.0	14.8	14.8	14.2	15.9
Colombia	14.1	14.3	14.1	13.7	15.1
Venezuela	19.2	18.8	18.9	18.3	19.8
Brazil	13.5	13.7	13.2	12.8	15.1
Ecuador	13.8	12.9	13.8	12.3	14.7
Peru	9.0	9.2	9.6	8.6	9.7
Rest of South America	12.2	11.7	11.5	10.5	13.1
Asia	12.5	12.2	11.7	10.5	14.2
China	12.8	13.4	12.4	11.5	14.0
Philippines	11.3	11.3	11.3	7.6	14.4
S. Korea	7.6	7.9	7.8	5.3	10.2
India	16.7	15.3	14.7	13.6	19.5
Vietnam	7.2	7.5	7.1	4.8	9.6
Turkey	4.3	4.1	4.2	4.3	4.5
Rest of Asia	3.2	3.0	2.9	1.0	4.9
Europe, Canada, Oceania	10.8	10.9	10.5	8.7	12.5
United Kingdom	10.3	10.9	10.2	7.8	13.4
Germany	11.6	11.4	10.9	9.8	13.4
Rest of Europe	11.7	12.1	11.4	9.5	13.0
Canada	8.5	8.2	8.2	6.3	10.7
Oceania	10.3	10.2	10.2	8.7	11.3
Africa	12.7	12.8	12.7	12.2	15.0
Total	16.3	16.3	15.9	15.4	17.3

TABLE 4.

Sensitivity of demographic impact of unauthorized entries to assumptions about coverage error and emigration, 2015–2022

		Coverage error of unauthorized foreign born		Emigration of legal foreign born
	Point estimate	Low (−50%)	High (+50%)	Low (−50%)	High (+50%)
Mexico	2,463	2,364	2,520	2,169	2,630
Central America	2,510	2,316	2,649	2,392	2,600
El Salvador	565	508	581	536	590
Honduras	823	781	884	815	848
Guatemala	980	893	1,034	915	1,011
Nicaragua	96	92	104	89	100
Rest of Central America	46	42	47	38	50
Caribbean	478	470	534	242	661
Haiti	106	103	112	80	126
Rest of Caribbean	372	368	422	161	535
South America	1,859	1,711	1,978	1670	2,036
Colombia	353	331	378	320	396
Venezuela	722	655	769	669	752
Brazil	370	354	390	338	424
Ecuador	187	162	199	159	203
Peru	76	73	89	67	88
Rest of South America	151	136	154	116	174
Asia	1,252	1,177	1,243	785	1,739
China	374	374	380	271	465
Philippines	235	224	250	107	362
S. Korea	63	63	68	28	104
India	525	462	490	359	687
Vietnam	36	36	37	11	67
Turkey	11	10	12	8	14
Rest of Asia	8	8	7	0	40
Europe, Canada, Oceania	792	759	814	463	1,053
United Kingdom	81	81	85	33	137
Germany	97	91	96	69	123
Rest of Europe	416	406	430	234	533
Canada	98	89	99	52	146
Oceania	99	92	104	75	114
Africa	661	629	699	553	871
Total	10,015	9,425	10,438	8,273	11,589

In general, higher assumptions about coverage error reduce estimates of expected duration, and higher assumptions about emigration increase them. Additionally, higher assumptions about both coverage error and emigration increase estimates of demographic impact. Nevertheless, the size of the effects of coverage error are relatively small. For example, comparing the low and high coverage error scenarios, the expected duration for Mexicans would decline two years from 22.7 to 20.7 years, and for all unauthorized immigrants, it would decline from 16.3 to 15.9 years (Table 3). Emigration assumptions have a somewhat larger impact than coverage error. For example, among Mexicans, low versus high emigration assumptions are associated with three-year change in expected duration (19.9 to 22.9 years), and an increase of 461,000 in demographic impact (2,169,000 to 2,630,000). Nevertheless, the regional patterns shown in Figures 5 and 6 would remain the same regardless of which assumptions were made. For example, the demographic impact of Mexican unauthorized immigration would remain low—ranging from 23 percent to 26 percent—regardless of the coverage or emigration assumption.

Discussion

In this paper, I presented new estimates of the dynamics of the unauthorized immigrant population in the United States for 27 different country and region of origin groups. Further, I developed novel measures of the unauthorized immigrants’ expected duration in unauthorized status and their demographic impact on the US population. The results illuminate the growing diversity in the unauthorized foreign-born population. As already known from prior research, unauthorized inflows are no longer dominated by Mexican labor migrants. Unauthorized immigrants now arrive to the United States from across the globe under a wide variety of circumstances. In terms of the indicators presented here (population dynamics, expected durations, and demographic impact), two major patterns stand out.

The first pattern is seen among unauthorized immigrants from Mexico, Central America, and Venezuela. They are unlikely to arrive on a nonimmigrant visa, have low outflow rates, and are expected to stay for prolonged periods of time. For Mexicans, this settlement pattern has been attributed to the build-up of border enforcement in the 1990s and 2000s, which disrupted earlier patterns of circular migration and encouraged unauthorized immigrants to remain in the United States for longer periods of time (Kandel, Seelke, and Wasem 2012; Massey, Durand, and Malone 2002; Reyes 2001, 2004). The long durations seen among unauthorized immigrants from Central America and Venezuela may owe to a combination of conditions, including the costs and difficulties of crossing the US–Mexico border without a temporary visa, and the fact that many are humanitarian migrants fleeing violence, economic collapse and other adverse conditions and are therefore unlikely to return to their origin countries (Massey 2020b).

This “long duration” group is very large, making up over two-thirds of unauthorized immigrants. Unless immigration laws change, these unauthorized immigrants are likely to remain unauthorized, thus entering middle and older adulthood in some type of unauthorized or liminal status (e.g., DACA or TPS). Because this group is so large and they remain in unauthorized status for so long, they continue to have a large demographic impact on the US population. That said, their impact (as well as the impact of all unauthorized immigration) is much less that it was in the early 2000s because entries have slowed considerably, particularly among Mexicans since 2007.

The second pattern is seen among the “new” unauthorized immigrants arriving from places other than Mexico and Central America. This group is made up of mostly temporary migrants who are more likely to have arrived on a nonimmigrant visa and do not remain unauthorized or stay for long. Unfortunately, I am unable to disentangle legalization from out-migration, so it remains unclear what share of these groups are finding ways to regularize their immigration statuses versus returning home. This is an important area for future research because this group is still numerically small but is growing. They come from across the globe and could be a mixture of asylum seekers awaiting decisions on their cases, visa overstayers who have yet to regularize their status, and labor migrants engaged in circular migration. Despite their growth over the last decade, their demographic impact remains small in comparison to the first group because of their smaller population size and shorter durations in unauthorized status.

As the numbers of Mexican and Central American unauthorized immigrants fall and the share of other groups increases, the impact and needs of the unauthorized population are likely to shift in complex ways. For example, the growth of groups engaged in short-term migration may signal a robust labor demand for temporary work visa programs and other opportunities for high-skilled foreign students and other STEM workers. It may also signal demand for revisions of policies for long-term visitors, such as older parents and retirees who do not seek to work but only want to spend extended time with their adult children in the United States. In contrast to short-term migrants, the growth of groups with longer expected durations of stay (e.g., Venezuelans and Central Americans) may require policies targeted toward childhood arrivals and humanitarian entrants who are in “legal limbo” (e.g., policies like DACA and TPS), to ensure their access to basic health care, education, and services that protect public safety (e.g., workplace safety and police protection).

The estimates and analyses presented here have important limitations. First, the duration and demographic impact measures do not foretell the future. Rather they describe the hypothetical experiences of synthetic cohorts. For example, Venezuelan unauthorized immigrants may not actually spend as much time in unauthorized status as my estimates suggest if they are granted pathways to legal status or if conditions improve in Venezuela that enable them to return home in the future. Second, it is important to recognize that the data presented here do not extend beyond 2022 and therefore do not fully capture the recent increase of entries at the US–Mexico border. Encounters at the US–Mexico border surged more than fourfold from 401,000 in 2020 to 1.7 million in 2021, and then further increased to 2.4 million in 2022 and 2.5 million in 2023 (U.S. Customs and Border Protection 2024). It is possible that we will see sharp increases in inflows and demographic impact of unauthorized immigration within the next few years. More generally, the estimates shown here do not capture the large numbers of rapid entries and departures often seen along the US–Mexico border, but instead are representative of the “settled” unauthorized population (i.e., those who are likely to be Census and ACS respondents). Third, the measure of demographic impact presented here focuses on the population footprint (i.e., person-years). It is silent about the social and economic contributions and costs of unauthorized immigration, which likely depend on immigrants’ characteristics (e.g., age, education, skills), labor needs in the US economy, and the support available within immigrant communities and families. Finally, the measures of expected duration and demographic impact are based on the residual estimates of the unauthorized population, which could be inaccurate due to low coverage of the unauthorized population in the ACS, uncertainty about the level of emigration among legal immigrants, sampling error, and reporting error on year of arrival and other immigration-related questionnaire items.

Regardless of these shortcomings, the estimates presented here are robust to sensitivity tests that vary assumptions about coverage, emigration and sampling error, and they shed light on the growing diversity and evolving dynamics of the unauthorized immigration. Two decades ago, the population was dominated by Mexican labor migrants who, in response to the difficulties and costs of crossing the US–Mexico border, tended to remain in the United States for extended periods of time. Their demographic impact on the US population was substantial. After the 2007–2008 Great Recession, however, new patterns emerged. Mexicans no longer dominate; unauthorized migrants now arrive from across the globe. At first, they came from Central America and Asia (up through 2018), and now we are seeing increasing numbers from Europe, Africa, Canada, and South America (especially Venezuela). The “new” unauthorized immigrants are more likely to arrive on nonimmigrant visas under a variety of circumstances, and, aside from Venezuelans, tend to spend less time in unauthorized status. Overall, the demographic impact of this new type of unauthorized migration appears to be lower than it was two decades ago, yet its diverse origins and varied circumstances may require a more tailored set of policies responses.