Beruflich Dokumente
Kultur Dokumente
Matthew Hall
Institute of Government and Public Affairs and Department of Sociology
University of Illinois at Chicago
Deborah Graefe
Population Research Institute
Penn State University
Gordon F. De Jong
Department of Sociology and Population Research Institute
Penn State University
For presentation at the 2010 Fall Meeting of the Association of Public Policy Analysis and
Management, Boston (November). Please do not cite without permission from the authors.
*Direct correspondence to Matthew Hall, Institute of Government and Public Affairs, 815 W.
Van Buren St., Chicago IL 60707; Email: mshall@uic.edu.
1
We use individual-level data from the 2006-2008 American Community Survey on high-skilled
immigrants and natives linked to metropolitan-level data on trends in high-skill immigration and
characteristics of labor markets to explore differences in labor market outcomes for those living
in industrial and immigrant-gateway cities. Our results indicate that high-skilled immigrants in
industrial cities are more likely to be employed and less likely to be underemployed than high-
skilled immigrants in gateway cities; patterns that are robust to micro- and metro-level controls.
In our analysis of the consequences of skilled immigration for natives, we find that, on balance,
the settlement of high-skill immigrants in industrial cities improves economic outcomes for high-
skill natives living there. Specifically, we find that in industrial cities where the immigrant share
of the high-skill population is growing, native men and women are more likely to be employed
and have higher earnings than natives in industrial cities where the skilled-immigrant population
is not growing. We also find no evidence that high-skill immigration affects the
underemployment of natives or geographically displaces natives in industrial cities.
2
million. The impact these new arrivals are having on the labor market opportunities of native-
born Americans is topic of great scholarly discussion and is at the center of debates concerning
immigration reform. At the risk of over generalizing, extant research regarding the consequences
of immigration for native outcomes has centered on the effects of total immigration (i.e., of all
Mexican immigrants. On balance, this work tends to ignore the potentially unique impacts of a
Recent work by demographers and economists has drawn attention to the distinctive
qualities of high-skill immigrant workers (e.g., Batalova and Fix 2008; Borjas 2005; Kaestner
and Kaushal 2010; Sana 2010; Stephan and Levin 2001). Overlooked, however, are the unique
majority of high-skilled immigrants work and live in a handful of major metropolitan gateways,
the growth in the high-skill immigrant population has been exceedingly high in old, industrial
cities; predominantly those making up America’s Rust Belt. Yet, little is known about the factors
that attract high-skill immigrants to these areas, and just as little known about the consequences
of their arrival for their high-skill native counterparts that also reside there. The goal of this
paper is to offer some initial clues into these questions. As the title suggests, our overarching
objective is to examine who the beneficiaries are of high-skill immigration to America’s Rust
Belt.
3
large industrial cities and 7 traditional immigrant gateway cities1 from the three-year 2006-2008
American Community Survey public use microdata file, linked to metropolitan-level data on the
recent immigration trends, industrial mix, and housing costs of the areas these workers live and
work. Our analytic approach examines first how high-skill immigrants fare in industrial versus
traditional gateway metropolitan areas, considering three critical labor market outcomes –
employment, underemployment, and earnings. We next explore differential impacts of the in-
flow of high-skill immigrants to industrial and gateway cities for high-skill natives on the same
three economic outcomes, plus metropolitan out-migration. We analyze each of these models
separately for male and female, prime-age workers. The multistep step process in which we first
consider the economic well-being of immigrants and then investigate how these foreign workers’
arrivals affect native labor market outcomes allows us to simultaneously offer insights into the
intriguing growth of the high-skill immigrant population in industrial cities and provide needed
Background
Despite low-skill Mexican immigrants being popularly perceived as the face of “new” American
come to America from nearly all reaches of the globe; in 2009 alone, 129 of the 194 officially
recognized by the State Department sent at least 1,000 migrants to America.2 Moreover, despite
regular portrayals of immigrants as lacking educational credentials, recent growth in the high-
1
Throughout this paper, “cities” and “metros” are used interchangeably to refer to Primary Metropolitan Statistical
Areas.
2
Based on analysis of data from the 2009 Yearbook of Immigration Statistics (U.S. Department of Homeland
Security 2010), including persons obtaining legal permanent resident status, refugee arrivals (but not those already
in the U.S. who were granted asylum), and temporary workers and their dependents.
4
skill (i.e., college-educated) immigrant population has outpaced that of the low-skill (i.e., high
school dropout) immigrant population by a factor of 3, and among prime-age workers, high-skill
immigrants now outnumber low-skill immigrants. This trend in immigration by skill-level and
the corresponding convergence of the low- and high-skill immigrant populations is shown in
Figure 1.
Despite the rapid increase in the number of high-skilled immigrants, these workers are
not settling uniformly across the American landscape. A burgeoning body of social and policy
research has focused on the rise of new immigrant destinations: areas outside of the traditional
immigrant ports-of-entry (New York, Los Angeles, Chicago, Houston, and Miami) that
experienced phenomenal growth in their foreign-born populations during the 1980s and 1990s.
The dispersion of immigrants to communities with little to no prior histories of immigration has
had transformative effects on social, economic, and political institutions in these areas. This
literature has, however, almost exclusively focused on the redistribution of comparatively low-
skilled Latino immigrants (e.g., Lichter and Johnson 2006; Massey 2008; McConnell 2008;
Zúñiga and Hernández-León 2005). In other work, we have described the unevenness in the
geography of high-skilled immigrants (Hall et al. 2011). Not only have high-skilled immigrants,
like other immigrants, been resettling (or settling directly) in non-traditional areas, they have
been doing so in areas very different than those that have received the greatest amount of recent
attention – the new destinations of Mexican migrants. As the focus of this paper suggests, the
areas that have been witness to the most rapid increase in skilled immigration are old, industrial
cities. Many of these areas, such as Detroit and St. Louis, are what Singer (2009) refers to as
“Former” Gateways – cities that were once home to large Western and Eastern European
immigrant population around the turn of the last century. Others, such as Akron and Syracuse,
5
have never been major recipients of immigrants, but are now home to growing high-skill
immigrant populations. For example, between 1990 and 2006, the number of high-skilled in
Baltimore increased by 200 percent (from about 26,714 to 80,482); a rate of growth nearly twice
as high as that observed nationally and in traditional immigrant gateways like Chicago and Los
Angeles.
That high- and low-skill immigrants are settling in different cities is no surprise. Different
industries demand different sets of skills from workers and the economic and industrial structure
of labor markets vary quite substantially across metropolitan areas. Cities also differ in their need
for foreign labor. Some metro areas draw on immigrants to fill vacancies in low-skill sectors,
where they work as builders, groundskeepers, farm hands, and cooks. Others attract immigrants
What is, however, surprising is that the skilled immigrant population is growing so rapidly in
places Americans traditionally associate with steel mills, auto plants, shipyards, and factories;
the very antithesis of high-tech America. Indeed, these areas have been credited with population
loses by their inability to attract modern industry and native talent (Frey 2004; Florida 2005).
While far from complete, the revitalization of America’s Rust Belt has is underway. The
collapse of the steel industry in Pittsburgh, for example, has been counterbalanced by growth in
engineering, and health care. Pittsburgh is home to three of the largest pharmaceutical companies
in the world (Bayer USA, GlaxoSmithKline, and Mylan, Inc.), a number of large chemical
engineering firms (Nova Chemicals, PPG Industries) and to the research arm of Thermo Fischer
Scientific. Likewise, Northeastern Ohio, historically known as the tire capital of the country is
today referred to as “Polymer Valley” for its emergence as a major player in plastics research
6
and nanotechnology. The transformation of (some) industrial cities from predominately factory-
based economies to more diverse economies with growing knowledge-based sectors, coupled
with the fact that native workers have been abandoning these cities for decades offers some
insight into why high-skill immigrants are settling there. Our goal is to attempt to address this
question a bit more systematically, examining whether immigrants living in industrial cities
really do reap labor market benefits above and beyond those available in traditional gateway
cities, and whether their presence presents any unique challenges for native workers already
there.
Do Immigrants Benefit?
One simple possibility for why immigrant outcomes may differ in industrial and gateway cities is
compositional: that is, differences in the labor market success of high-skill immigrants living in
industrial versus gateway cities are due to compositional differences between the immigrant
populations living in these areas. While growth in skilled immigration is especially high in
industrial cities, far more immigrants (both unskilled and skilled) live in traditional gateway
cities. Extending the spatial assimilation model to a broader scale, if high-skill immigrants
possess personal characteristics – e.g., more schooling, longer stays in the U.S., or being
naturalized – that are also the conducive to employment and higher earnings (see Hall 2009),
than we may observe differences in these outcomes between immigrants in the two types of
in labor market outcomes between those living in industrial and gateway cities.
7
in immigrant economic well-being are due to the different structures of local labor markets. As
previously noted, while industrial cities have, by definition, historically had manufacturing- and
factory-based economies, a revival from the rusting-out of these cities has taken place in many of
them. This recovery has been stimulated at least partially by the growth of highly technical
industries in the sciences, health, and information technology (Hall et al. 2011) – sectors that
employ sizeable shares of the high-skill population generally, and especially large shares of high-
skill immigrants (Batalova and Fix 2008). Following this demand-side explanation, high-skill
immigrants are attracted to industrial cities in the first place because job opportunities are more
plentiful, or at least because there are more openings for highly-technical jobs that compensate
well in industrial areas. Thus, any observed difference in labor markets outcomes between high-
skill immigrants in industrial and gateway cities should be explained by features of the
Do Natives Benefit?
Any interest in the benefits of high-skill immigration in industrial cities cannot stop with the
immigrants. Indeed, the consequences of sustained levels of high immigration for natives’
economic well-being are of great concern and the focus of major political debates. There has
been a great deal of work dedicated to understanding the labor market consequences of
immigration, writ large (see Longhi, Nijkamp, and Poot 2005; Smith and Edmonston 1997). Two
classic microeconomic model of labor market competition predicts that any increase in the
supply of workers (e.g., via in-migration) will result in greater competition between workers over
8
a fixed set of jobs, depressing wages (at least over the short-run). Studies testing the impact of
immigration on native job loss and wage decline have produced varied results (c.f., Borjas 2003;
Card 2001). Generally, they indicate small, but negative impacts of immigration on the
employment and wages of natives (Lohghi et al. 2008; Friedberg and Hunt 1995).
The potential for job competition with natives depends on the substitutability of native
and immigrant workers. While a consensus on this topic is far from being reached (c.f., Borjas,
Grogger, and Hanson 2008; Ottaviano and Peri 2006), the hour-glass shape of the immigrant
population means that natives at the tail ends of the educational distribution are most sensitive to
foreign labor. Borjas (2003), for example, finds that immigration during the 1980s resulted in
reductions in wages of 8.9% and 4.9% for high-school dropout and college graduates,
respectively; considerably higher than the immigration effects for the middle-education groups.
Among the relatively few studies focused explicitly on skill workers, Borjas (2005) finds that a
10% increase in the supply of foreign-born doctorates is associated with a 3% reduction in the
Native displacement on behalf of immigrant workers may not just operate via job loss,
but also by inducing natives to flee labor markets (Borjas 2006; Card 2001; Borjas, Freeman, and
Katz 1997; Frey 1995; Filer 1992). Indeed, Borjas and colleagues (e.g., Borjas 2003; Borjas,
Freeman, and Katz 1996; Borjas and Katz 2007) argue that area-based approaches (i.e., studies
that consider the effect of immigration within labor markets) are inherently flawed because
natives may respond to the in-migration of foreign workers by all together leaving. While the
implied assumption of labor market integration has been challenged (see Card and DiNardo
2000), there is some evidence of a positive correlation between immigration and native out-
migration (Frey 1995; Kritz and Gurak 2001; but see Card and Dinardo 2000; Wright, et al.
9
1997). Available research on this topic suggests that low-skill native workers (those with a high
school education or less) are most likely to be geographically displaced (Altonji and Card 1991;
Borjas 2006; Frey 1995; Walker, Ellis, and Barff 1992; White and Liang 1998). In short, because
labor is mobile, it is important to consider not just the impact of immigration on native
employment and earnings, but also if immigration influences the likelihood of native out-
migration.
A related, but conceptually unique perspective holds that immigrants merely replace
native workers who previously left geographies for better opportunities elsewhere. This is
especially salient in the context of Rustbelt and other industrial cities which have experienced
sustained population loss for at least two decades. Even more relevant is the fact that out-
migration from these areas has tended to be selective of young college graduates (Franklin
2003), prompting some to refer to these areas as “Brain Drain Cities” (Harden 2003). This
possibility – that immigrants may not be displacing, but merely replacing workers who had left
(or otherwise would) – increases the need for appropriately lagged variables in research seeking
innovativeness may create new markets, goods, and services which could spur enhanced
opportunities for native workers (Hunt and Gauthier-Loiselle 2010; Lofstrom 2000). It is widely
offering services below market value (e.g., in landscaping and housekeeping) and that high-skill
immigrants complement low-skill workers by creating labor demands that increase employment
(e.g., in housing construction) (see Borjas 1995). A complementing process is also possible if
10
immigrants and natives are not competing for the same jobs. In the low-skill sector, the anecdote
is typically that low-skill natives will not do the work that low-skill immigrants will (e.g.,
picking lettuce in the blistering sun). In the high-skill sector, however, the argument is typically
that the local pool of high-skill natives does not possess the skills that immigrants do.
Consequently, if immigrants are not “taking” jobs from natives, but are increasing the overall
economic vitality of labor markets (through knowledge flows, ties to foreign institutions, and
increased enrollment in graduate programs), then employment opportunities and wages for high-
skill natives should increase (Herman and Smith 2010). Consistent with this line of reasoning,
Orrenius and Zavodny (2007) find no effects of immigration on the wages of skilled workers.
Similarly, Kaestner and Kaushal (2010) explore the impact of immigration on native registered
nurses’ economic outcomes, and find no effects on either employment or wages. Peri and
Sparber (2008) argue that these null or positive findings are generated because foreign- and
native-born skilled workers specialize differently (i.e., immigrants in jobs requiring analytic
There is reason to believe that natives in labor markets with different histories of
immigrant reception will respond differently to immigrants. Sociologists have long been
interested in the structural dimensions of labor market process (see Berg 1981; Farkas, England
and Barton 1988). Meta-analysis of economic studies largely confirm that context matters – i.e.,
the impact of immigration on employment and wages depends on the flexibility of the market
and the ease with which native workers are able to relocate (Longhi et al. 2008; also see Card
and Lewis 2007). Certainly, any observed effects of immigration – in our case, of high-skill
immigration – may vary according to the composition of the local native population; i.e.,
may be due to differences in the composition of the latter between industrial and immigrant
gateway cities.
To explore the differential outcomes for high-skilled workers in industrial and gateway cities, we
draw on individual-level data from the 2006-2008 three-year American Community Survey
Public-Use Microdata Sample. The ACS is an annual survey of 3 million households conducted
by the U.S. Census that collects detailed information formerly available in the long form of the
decennial census. Because our main group of interest – high skill immigrants – represents a
relatively narrow slice of the total U.S. population, we use the three-year ACS file which
effectively appends together the annual PUMS files from 2006 to 2008 into a single database.
We restrict our sample to high-skilled men and women of prime working-age: college-
educated persons between the ages of 25 and 45.3 For the native-born, we further restrict the
analysis to those who identify as non-Latino white or black. We limit our analysis to individuals
who reside in an “industrial” or “immigrant gateway” primary metropolitan statistical area, based
on boundaries set by the Office of Management and Budget in 2008. In our classification,
“industrial” cities refer to the 37 metropolitan areas that the Brookings Institution’s State of
Metropolitan America (2010) defines as either a “Skilled Anchor” (n=19) or “Industrial Core”
(n=18), which are distinguished from five other categories (e.g., Diverse Giant, Border Growth,
New Heartland) of the 100 largest metro areas.4 “Immigrant gateway” cities refer to 7 large
3
Widening the age interval to include older workers (e.g., 25-55 or 25-64) produces substantively similar, but
generally weaker findings. Results from these models are available on request.
4
The results are very similar if respondents in the 19 “Skill Anchors” are excluding from the sample. For the
analysis on native outcomes, the employment models (for both men and women) produce slightly weaker results
when these metro areas are excluded, but the substantive interpretation remains the same. For all other outcomes,
this alternative definition of industrial status produces results nearly identical to those shown here.
12
metropolitan areas that serve as traditional destinations for high-skilled immigrants. A list of
metro areas included in the analysis along with the number of natives and immigrants in each is
Our analysis of high-skill immigrants and natives focuses on four outcomes (three for
immigrants). Employment is a binary indicator of whether individuals in the labor force are
currently working. Those coded “0” are not working, but seeking employment.
occupation-specific educational means and standard deviations were calculated based on all
American workers using the three-year 2006-2008 ACS PUMS. Underemployment is defined as
a mismatch between one’s schooling and the average educational level of those with the same
occupation. Workers are considered underemployed if their level of schooling is one or more
standard deviations above the occupation-specific mean.5 Earnings refers to annual earned
income from all jobs for employed persons. Because of the right skew of the distribution,
out of one of the 44 metropolitan areas under study within the last year (t-1). For the purposes of
this paper, we focus our attention on the “departure” dimension of migration (in contrast to the
“destination” dimension) which models the likelihood that an individual will leave a
metropolitan area. Thus, we do not distinguish migrants based on the area they move to and
persons are considered migrants even if they move to another one of our 44 metro areas.
5
While this operationaliation of underemployment is preferable to more crude measures of overqualification (e.g.,
based on poverty status or hours worked), it is not without limitations. By definition, a certain percentage of workers
must lie above one standard deviation from the mean, and in leptokurtic distributions, this operationalization may
not necessarily correspond to an objective mismatch in skills (see Burris 1983; but see Clogg and Shockey 1984)
13
economic well-being of high-skill immigrants in industrial versus gateway cities, the key
“industrial” or not. The goal in the native analyses is different: to test whether the impact of
high-skill immigration varies by metro type. In these models, the key variable is the product term
of the industrial dummy and changes in the immigrant share of the high-skill population (i.e., the
percent of the total high-skill population that is foreign-born) between 2000 and 2006.6 Micro-
level control variables incorporated in both the native and immigrant analyses include age (in
linear years7) and educational attainment (in number of years of schooling). For models specific
identifies immigrants who have become naturalized; English, which refers to foreign-born
persons who speak English “only,” “very well,” or “well”; Years in US represents the number of
years individual immigrants have lived in the U.S.; and region of birth which is a vector of
dummy variables representing where in nine world regions (Western Europe, Eastern
Europe/Former USSR, Asia, Latin America, Caribbean, South America, Africa, and Oceania)
one was born, with Canada serving as the referent. In models specific to the native-born, we also
include a binary indicator of whether the respondent is non-Latino black. All of these
unemployment rate and real gross domestic product per capita, both measured in 2006 and based
6
In supplemental models, numerous alternative specification of high-skill immigration were investigated, including
relative change in the high-skill immigrant share, absolute change in the number of high-skill immigrants, relative
change in the high-skill immigrant population, and corresponding changes in the percent of the immigrant
population that is high-skill. Results from these models are generally consistent with the presented ones, based on
absolute changes in the high-skill immigrant share, but tend to produce statistically weaker results.
7
Because of the relatively short age range which defines our sample, the relationship between age and labor market
outcomes is approximately linear (i.e., there is no evidence of curvilinearity).
14
on BLS data from the Regional Economic Information Systems. Metropolitan housing costs are
determined by the median rent-to-income ratio for all residents in 2006, and come from the one-
year 2006 ACS PUMS. We also include measures of sector-specific job growth between 2000
and 2006 from REIS for four industrial groupings: Professional, Scientific, and Technical
Services (Education), and Health Care and Social Assistance (Health). In out-migration models,
metropolitan-level covariates refer to the metropolitan area of residence in the prior year. Thus,
for movers, the metro variables represent characteristics of the metropolitan areas in which they
migrated from.
We analyze these data using a series of regression models that take the following general
the immigrant share of the high-skill population between 2000 and 2006; imm is the immigrant
a random error term. All descriptive and multivariate analyses are weighted using the person
weights provided by the ACS, and because there are multiple respondents in any given metro
area, we adjust for this nested structure of the data using the cluster procedure in Stata to
Results
Descriptives
Descriptive mean differences in employment, underemployment and earnings are shown for
high-skill immigrants in gateway and industrial cities in Table 1, separately for men and women.
For men, high-skill immigrants living in industrial cities are significantly more likely to be
employed and less likely to be underemployed than their counterparts living in gateway metros.
While the mean earnings of high-skill immigrant men are noticeably higher for those in gateway
cities (by $2,883), the difference is not statistically significant (t = -.56). Among immigrant
women, differences in employment are small but indicate slightly higher rates of employment in
gateway cities. Like with men, however, differences in underemployment are more visible with
rates of underemployment being about 7.6% lower in industrial than gateway cities. A similar to
their male counterparts, high-skill immigrant women’s earnings are higher in gateway cities, but
Corresponding descriptive patterns are shown for high-skill immigrant natives in Table 2.
For both high-skill male and female natives, levels of employment and underemployment are
relatively comparable in industrial and gateway cities. However, the industrial-gateway disparity
in earnings between these workers is large; much larger than the equivalent gap for immigrants.
For men, high-skill natives in gateway cities earn, on average, $22,576 more than high-skill
natives in industrial cities. Among women, the gap in earnings is $13,767. Also shown in Table 2
are mean differences in rates of out-migration for high-skill natives. Consistent with the long-
term trend of skilled population loss from industrial cities, migration rates are significantly
higher for high-skill male and female natives in industrial than gateway cities.
16
While these descriptive patterns are interesting and informative, they cannot reveal to what
extent these discrepancies are due to compositional or structural differences between industrial
and gateway cities. The series of regression models in Tables 3 and 4 seek to offer some useful
insights into these possibilities for high-skill immigrant men and women, respectively. In each
table, three models are shown for each of the three labor market outcomes (employment,
underemployment, and log earnings) – a baseline, unadjusted model (1); one that adds in micro-
Starting with men, the regression coefficients Table 3 indicate that high-skill immigrant
men in industrial cities are significantly more likely to be employment and significantly less
Adding micro-level characteristics we see that labor market outcomes are better for older
immigrants, those with more years of schooling, those who are proficient in English, and for
those that have been in the U.S. longer. Caribbean and African immigrants have significantly
lower levels of employment than Canadian immigrants, and immigrants from essentially all non-
Canadian regions of the globe have higher levels of underemployment and lower earnings than
their Canadian counterparts. Most importantly, the addition of the controls for individual-level
characteristics of high-skill immigrants does not explain differences in these outcomes between
immigrants in industrial and gateway cities. The industrial coefficient for employment is
partially mediated –reduced to about three-fourths (72.4%) of its original size – but the
17
these variables included, a net negative industrial effect is revealed.8 Thus, differences in
markets. For all three outcomes, the local unemployment rate registers a significant association,
With a few exceptions, none of the other metropolitan variables are significantly associated with
high-skill immigrants’ labor market outcomes. For our purposes, however, the most crucial
observation is that including controls for overall metropolitan economic growth, housing costs,
and industrial-sector job growth, does not explain away the significant, positive industrial effect
characteristics do, in contrast, account for the negative industrial coefficient in the earnings
equation indicating that once metropolitan factors are held constant, there is no earnings
Corresponding regression models for high-skill immigrant women are shown in Table 4.
The unadjusted results for high-skill immigrant women parallel the descriptive patterns:
immigrants in industrial cities are less likely to be employed and underemployed and earn less
than immigrants in gateways, although the difference is only significant for underemployment.
8
The suppression of both underemployment and earnings is due almost entirely to high-skill immigrants in
industrial cities being more highly educated (Pearson’s r = .1186), and since education is positively related to both
underemployment and earnings.
9
Especially important in the attenuation of the industrial coefficient in the earnings model is GDP per capita,
possibly suggesting that differences in cost-of-living can account for the lower earnings of high-skill immigrants in
industrial cities.
18
Micro-level controls added in column 2 for each outcome indicate that economic well-being is
generally better for high-skill immigrant women who are older, better educated, naturalized,
English speakers, and those who have been in the country longer. Immigrants from non-
Canadian countries tend, on average, to have poorer outcomes than Canadian immigrants.
Crucially, with these qualities of immigrants held at their means, the absolute size of the
industrial coefficients on underemployment and log earnings increase for the original (Model 1)
values.10
Some mediation of the industrial coefficients does occur when metropolitan variables are
added to the equation. The industrial effect on both underemployment and earnings is cut by
about 43% with the addition of structural features of labor markets. Despite these reductions in
industrial and gateway metro areas, the industrial coefficients on both of these important labor
market outcomes remain relatively large in magnitude and statistically significant. The final
models indicate that with micro- and metro-level characteristics controlled, high-skill immigrant
women are about 9.8% less likely to be underemployed, but report an earnings deficit of about
6.9%.
Our immigrant-specific models can help us understanding if high-skill immigrants are benefiting
from their settlement in industrial cities, but an equally as – if not more – important question
concerns the impact high-skill immigration is having on their native-born counterparts in these
metropolitan areas. The next series of regression models, shown in Tables 5 (men) and 6
10
Like with men, this suppression is due mostly to the higher levels of education among high-skill immigrant
women in industrial cities.
19
(women), seek to offer some evidence on this topic. As with immigrants, three models are shown
for each of the main outcomes, which include not just the same three examined for immigrants,
but also the likelihood of metropolitan out-migration: the first model includes four key terms, a
lagged variable of change in the percent of high-skill person who are foreign-born between 2000
and 2006, a dummy indicator of industrial status, the product of the two terms, and an anchoring
effect of the change variable (i.e., percent immigrant of the high-skill population in 2000); the
second model includes micro-level characteristic of high-skill immigrants; and the third
incorporates the same metropolitan structural characteristics used in the immigrant models.
Models predicting employment, underemployment, and out-migration are estimated as logits; the
the immigrant share of the high-skill population are positively related to the log-odds of
employment for native high-skill men, but the share of high-skill immigrants in 2000 is
negatively associated with employment. Most importantly, the interaction between recent
changes in the high-skill immigrant share and industrial status is statistically insignificant,
indicating that the impact of recent changes in the high-skill immigrant share on high-skill native
immigration on the log-odds of high-skill native men being overqualified in their current jobs;
indeed, immigration appears to have no effect at all on native men’s underemployment. Both the
earnings and migration models, however, do suggest distinct immigration effects in industrial
areas. Specifically, while changes in the high-skill immigrant share have little to no effect on
high-skill natives’ earnings in gateway areas, the significant and positive interaction term
20
between changes in percent high-skill immigrant and industrial metro indicates that the earnings
of high-skill natives in industrial cities grow faster where the high-skill immigrant population is
model show some evidence of high-skill immigrants geographically displacing high-skill native
men in gateway cities, but not in industrial cities. Indeed, the net immigration effect in industrial
areas is negative and indicates that, while overall native men in industrial cities have heightened
odds of out-migration, they are less likely to leave metros if the high-skill immigrant population
is growing.
As with immigrants, the second model for each outcome adds in micro-level
characteristics, age, education, and race. The coefficients on these terms indicate that economic
well-being increases with age and education, and is lower for black than white men, even in the
high-skill population.11 Consistent with migration research (see Greenwood 1997), the likelihood
of out-migration from a metropolitan area decreases with age and is higher for those with more
migration. With these traits of individual workers controlled, high-skill natives are more likely to
be employed where the immigrant share of the high-skill population is increasing (regardless of
industrial status). For natives in industrial cities, their earnings also increase with growth in the
high-skill immigration population, and while there remains evidence of geographic displacement
of high-skill natives in metros experiencing growth in the high-skill immigrant population, this is
only true in gateway cities (i.e., industrial natives remain less likely to out-migrate if the high-
11
In supplemental models available upon request, separate models were run for non-Latino black and white natives.
The main effect of high-skill immigration and its interaction with industrial status does not vary by race for any of
the four outcomes, for either men or women.
21
industrial sector job growth are added in the final column for each outcome. Few coefficients
register statistically significant effects, but the value of overall job growth and growth in the
health sector is clear. With these variables held constant in the final models, significant effects of
immigration remain but, in some cases, are reduced in overall size. For employment, the main
effect of change in high-skill immigration is reduced by 46%; however, the coefficient remains
statistically significant and indicates that high-skill native men are more likely to be employed in
metro areas where the immigrant share of the high-skill population is growing.
In the earnings equation, with these same features of metro areas controlled, a significant,
negative main effect of immigration emerges suggesting that high-skill immigrants may be
depressing wages of high-skill native men; a one percentage point increase in the immigrant
share of the high-skill population reduces native earnings by 2.9%. However, the positive and
significant interaction between changes in high-skill immigration and industrial status indicates
that wage depression is only observed in gateway cities. The net earnings effect of high-skill
immigration on native men’s earnings is actually positive and indicates that a one percentage
point increase in the high-skill immigrant share increases the earnings of native men by about
3.3% [exp(-.029+.061)].
The full out-migration model also points to some positive effects of immigration in
industrial cities. Specifically, with the full slate of individual- and metropolitan-level variables
controlled, high-skill natives have higher log-odds of out-migration in areas where the high-skill
immigrant population is growing, offering some support to the geographical displacement thesis.
However, the negative interaction term between changes in the immigrant share of the high-skill
22
population and industrial status effectively offsets the main effect for high-skill native men in
industrial cities.
Corresponding results for high-skill native women are very similar to those for men. The
coefficients in Table 6 indicate that without micro- or metro-level controls, high-skill native
women have higher log-odds of employment when the immigrant percentage of the high-skill
population is increasing. Its interaction with industrial status is nonsignificant, indicating that the
industrial cities. Without controls, high-skill natives in industrial cities, but not gateway metros,
see earnings gains to increases in the immigrant share of the high-skill population. In addition,
while (like men) high-skill native women in gateway cities have higher log-odds of out-
migration when the high-skill immigrant population is growing, that evidence of geographic
displacement does not appear to play out in industrial cities, as the interaction term between
immigration and industrial status more than offsets the positive main effect.
does not mediate the effects of immigration on native women’s economic well-being in any
meaningful way. Indeed, with these variables in the model, the employment premium associated
with high-skill immigration remains moderately large and statistically significant and indicates
that a one-percentage point increase in the immigrant share of the high-skill population increases
the odds of employment by 8.8%, for high-skill native women in both gateway and industrial
cities. By contrast, both the earnings and out-migration models suggest some disadvantageous
effects of high-skill immigration for native women in gateway cities; namely that changes in the
high-skill immigrant population are associated with lower earnings and heightened log-odds of
migration. However, for women in industrial cities, the net wage effect is positive – a one
23
percentage point increase in the high-skill immigrant share is associated with an increase in
native women’s earnings of about 2.6% (e(-.026+.052)-1) – and the net migration effect is effectively
zero, suggesting the immigrants are not geographically displacing high-skill native women there.
Conclusion
Our general objective in this paper has been to complement the literature on immigration
dispersion and skill differentiation by describing who the beneficiaries are of high-skill
immigration to the new destinations of foreign-born skilled workers in Industrial America. Using
individual-level data from the 2006-2008 American Community Survey PUMS linked to metro-
level data on changes in immigration and labor market conditions in 44 industrial and immigrant-
underemployment, and earnings – for prime-age, high-skilled immigrant men and women and
the same outcomes, plus metropolitan out-migration for high-skilled natives. Results from these
analyses confirm that context matters: both immigrant outcomes and the effect of immigration on
So who does benefit from high-skill immigrant settlement in industrial cities? A graphical
summary of our results is shown in Table 7. Overall, both the immigrant and native portions of
the analysis suggest widespread benefits of the settlement of skilled immigrants in industrial
areas are more likely to be employed (men only) and, when employed, are significantly less
likely to be overqualified for their jobs. And, while there is an earnings disadvantage for skilled-
immigrants in industrial cities, it is largely explained by the differential costs of living between
Of at least as much interest, the results for natives also point to advantages associated
with the migration of skilled immigrants to industrial cities. Specifically, high-skill native men
and women in industrial (and gateway) cities in which the foreign-born share of the high-skill
population is expanding are more likely to be employed than their counterparts in areas where
the immigrant population is not growing. We also find no evidence that immigrants are pushing
natives into underemployment (i.e., into jobs that they are overqualified for). The earnings
equations indicate that while increases in immigration are associated with depressed earnings for
natives in gateway cities, the opposite is true of high-skill natives in industrial cities: as the
immigrant share of the high-skill population increases, the earnings of high-skill natives also
this pattern is only true for high-skill natives in gateway cities; for those in industrial cities, the
Our efforts to explain the advantages of industrial residence (for immigrants) and
industrial immigration (for natives) were unsuccessful. Very little of the industrial-gateway
living in these cities. In fact, adjusting for the compositional differences tends to suppress even
labor markets do explain a considerable portion of the premiums in economic well-being for
immigrants, but large and significant differences remain. For high-skill natives, neither
individual- nor metropolitan-level controls attenuate the beneficial effects of immigration for
those in industrial cities. What can we account the unexplained variability to? It is very likely
that we have been unable to capture all the dynamics of local labor markets with our limited set
of metro-level variables. As Peri and Sparber (2008) contend it is quite possible that immigrants
25
and natives specialize differently, and given the differences in the structure of industrial and
gateway cities’ labor markets, it is conceivable that differences in specialization may be more
pronounced in industrial than gateway cities. Thus, if the substitutability of immigrants is more
imperfect in industrial areas, there may be greater room for high-skill immigrants to complement
the work of high-skill natives. Alternatively, a more sociological explanation may focus on
differences in the context of reception between industrial and gateway cities. De Jong and Tran
(2001) report that native attitudes about immigrants are more favorable throughout the Midwest
region – where many of our industrial cities are located – than in other regions of the country.
The data they use – the 1994 GSS – however, predates much of the resettlement of immigrants in
these areas. Moreover, in subsequent work, De Jong and Steinmetz (2004) find that receptivity
climate affects the occupational attainment of low-skill, but not high-skill natives.
In sum, our results offer relatively little support for the traditional displacement
hypothesis that high-skill immigrants are taking jobs and depressing wages for high-skill natives,
or that they are prompting natives to search for new opportunities beyond the boundaries of their
metropolitan area. It is important to recognize here that we only find such positive outcomes for
high-skill workers in industrial cities. In contrast, the impact of immigration on the earnings and
disadvantageous, depressing earnings and driving natives out of gateway cities all together. We
also must stress here that the results we present here are largely descriptive in nature. We are
unable to state clear causal arguments about the impact of immigration, although our results are
generally inconsistent with at least the gloomiest accounts of the consequences of immigration.
Our initial investigations into addressing the endogeneity of changes in immigration have proved
26
unsuccessful; but future efforts will surely need to take this concern seriously.12 And, while we
have been careful to lag variables in an appropriate manner, related work would benefit by taking
advantage of longitudinal data sources to better isolate the impacts of skilled immigration, for
12
We experimented with 2SLS models instrumenting the endogenous change in the immigrant share of the high-
skill population (and its interaction with industrial status) with the national share of all high-skill immigrants living
in a metropolitan area. We have also explored models in which the earlier “anchor” point of the high skill immigrant
share in 2000 serves as an instrument. Diagnostic tests for both of these alternative specifications suggest weakly
identified models (Cragg-Donald Wald tests < 1.5).
27
References
Altonji, Joseph G., and David Card. 1991. “The Effects of Immigration on the Labor Market
Outcomes of Less-Skilled Natives,” Pp,. 201-234 in John M. Abowd and Richard B. Freeman
(eds.), Immigration, Trade, and the Labor Market. Chicago, IL: University of Chicago Press.
Batalova, Jeanne and Michael Fix. 2008. Uneven Progress: The Employment Pathways of Skilled
Immigrants in the United States. National Center on Immigrant Integration Policy. Washington,
DC: Migration Policy Institute.
Berg, Ivar E. 1981. Sociological Perspectives on Labor Markets. New York: Academic.
Borjas, George. 1995. “The Economic Benefits from Immigration.” Journal of Economic
Perspectives 9: 3-22.
Borjas, George J. 1999. “The Economic Analysis of Immigration,” Pp. 1697-760 in Orley C.
Ashenfelter and David Card (eds.), Handbook of Labor Economics. Amsterdam: North-Holland.
Borjas, George J. 2003. “The Labor Demand Curve Is Downward Sloping: Reexamining the
Impact of Immigration on the Labor Market,” Quarterly Journal of Economics 118: 1335-74.
Borjas, George J. 2006. “Native Internal Migration and the Labor Market Impact of
Immigration.” Journal of Human Resources 41: 221-58.
Borjas, George, Richard Freeman, and Lawrence Katz. 1996. “Searching for the Effect of
Immigration on the Labor Market.” American Economic Review 86: 246-51.
Borjas, George, Richard Freeman, and Lawrence Katz. 1997. “How much do Immigration and
Trade Affect Labor Market Outcomes?” Brookings Papers on Economic Activity 1: 1-90.
Borjas, George J., Jeffrey Grogger, and Gordon H. Hanson. 2008. “Imperfect Substitution
between Immigrants and Natives: A Reappraisal.” NBER Working Paper No. 13887.
Borjas, George J. and Lawrence F. Katz. 2007. “The Evolution of the Mexican-Born Workforce
in the United States” Pp. 13-55 George J. Borjas (ed.), Mexican Immigration to the United
States. Chicago: University of Chicago Press.
Burris, Val. 1983. “The Social and Political Consequences of Overeducation.” American
Sociological Review 48:454-67
Card, David. 2001. “Immigrant Inflows, Native Outflows, and the Local Labor Market
Impacts of Higher Immigration,” Journal of Labor Economics 19(1): 22-64.
28
Card, David and John DiNardo. 2000. “Do Immigrant Inflows Lead to Native Outflows?”
American Economic Review 90: 360-7.
Card, David and Ethan G. Lewis. 2007. “The Diffusion of Mexican Immigrants during the
1990s: Explanations and Impacts,” Pp. 193-227 in George J. Borjas (ed.), Mexican Immigration
to the United States. Chicago: University of Chicago Press.
Clogg, Clifford C., and James W. Shockey. 1984. “Mismatch between Occupation and
Schooling: A Prevalence Measure, Recent Trends, and Demographic Analysis.” Demography
21: 235-57.
De Jong, Gordon F., and Michelle Steinmetz. 2004. “Receptivity Attitudes and the Occupational
Attainment of Male and Female Immigrant Workers.” Population Research and Policy Review.
23: 91-115
De Jong, Gordon F., and Quynh-Giang Tran. 2002. “Warm Welcome, Cool Welcome: Mapping
Receptivity Toward Immigrants in the U.S.” Population Today 29: 1, 4-5.
Farkas, George, Paula England, and Margaret Barton. 1988. “Structural Effects on Wages:
Sociological and Economic Views.” Pp. 93-112 in George Farkas and Paula England (eds.),
Industries, Firms, and Jobs: Sociological and Economic Approaches. New York: Plenum.
Filer, Randall. 1992. “The Effect of Immigrant Arrivals on Migratory Patterns of Native
Workers,” Pp. 245-69 in George Borjas and Richard Freeman (ed.), Immigration and the Work
Force. Chicago: University of Chicago Press.
Florida, Richard. 2005. Cities and the Creative Class. New York: Routledge
Franklin, Rachel S. 2003. “Migration of the Young, Single, and College Educated: 1995 to
2000.” Census 2000 Special Report, CENSR-12. Washington D.C.: U.S. Census Bureau.
Frey, William. 1995. “Immigration and Internal Migration Flight from US Metropolitan Areas:
Toward a New Demographic Balkanization.” Urban Studies. 32: 733-757.
Frey, William H. 2004. “Brain Gains, Brain Drains.” American Demographics 26(5): 19-24.
Friedberg. R. M., and Hunt, J. 1995. “The Impact of Immigrants on Host Country Wages,
Employment and Growth.” Journal of Economic Perspectives 9: 23-44.
Greenwood, Michael J. 1997. “Internal Migration in Developed Countries,” Pp. 647-720 in M.R.
Rosenzweigh and O. Stark (eds.), Handbook of Population and Family Economics. Amsterdam:
Elsevier.
Hall, Matthew. 2009. “Interstate Migration, Spatial Assimilation, and the Incorporation of U.S.
Immigrants.” Population, Space, and Place 15: 57-77.
29
Hall, Matthew, Deborah Graefe, and Gordon F. De Jong. 2011. “The Geography of Immigrant
Skills: Educational Inequality across Metropolitan Destinations.” Washington, DC: Brookings
Institution.
Harden, Blaine. 2003. “Brain-Gain Cities Attract Educated Young,” Washington Post,
November 9, A01.
Herman, Richard T., and Robert L. Smith. 2010. Immigrant, Inc.: Why Immigrant Entrepreneurs
Are Driving the New Economy (and How They Will Save the American Worker). Hoboken, NJ:
Wiley.
Hunt, Jennifer and Marjolaine Gauthier-Loiselle. 2010. “How Much Does Immigration Boost
Innovation.” American Economic Journal: Macroeconomics. 2: 31-56.
Kaestner, Robert and Neeraj Kaushal. 2010. “Effect of Immigrant Nurses on Labor Market
Outcomes of US Nurses.” Presentation at the Population Association of America, 2010 (Dallas).
Kritz, Mary M. and Douglas T. Gurak. 2001. “The Impact of Immigration on the Internal
Migration of Natives and Immigrants.” Demography 38: 133-45.
Lichter, Daniel T., and Kenneth M. Johnson. 2009. “Immigrant Gateways and Hispanic
Migration to New Destinations.” International Migration Review 43: 496-518.
Longhi, Simonetta, Peter Nijkamp, Jacques Poot. 2005. “A Meta-Analytic Assessment of the
Effect of Immigration on Wages.” Journal of Economic Surveys 19: 451-77.
Longhi, Simonetta, Peter Nijkamp, Jacques Poot. 2008. “The Impact of Immigration on the
Employment of Natives in Regional Labour Markets: A Meta-analysis.” Pp. 173-193 in Poot,
Waldorf, and van Wissen (eds.), Migration and Human Capital. Northampton, MA: Edward
Elgar.
McConnell, Eileen Diaz. 2008. “The U.S. Destinations of Contemporary Mexican Immigrants.”
International Migration Review 42: 767-802.
Massey, Douglas S (ed.). 2008. New Faces in New Places: The Changing Geography of
American Immigration. New York, NY: Russell Sage Foundation.
Orrenius, Pia and Madeline Zavodny. 2007. “Does Immigration Affect Wages? A Look at
Occupation-Level Evidence.” Labour Economics 14: 757-73.
Ottaviano, Gianmarco, and Giovanni Peri. 2006. “Rethinking the Effects of Immigration
30
Peri, Giovanni and Chad Sparber. 2008. “Highly Education Immigrants and Native Occupational
Choice.” CReAM Working Paper #07/08.
Sana, Mariano. 2010. “Immigrants and Natives in U.S. Science and Engineering Occupations.”
Demography 47: 801-20.
Singer, Audrey. 2009. “The New Geography of United States Immigration.” Washington, D.C.:
Brookings Institution.
Smith, James P. and Barry Edmonston. 1997. The New Americans: Economic, Demographic,
and Fiscal Effects of Immigration. Washington, DC: National Academy Press.
StataCorp. 2009. Stata Statistical Software: Release 11.0. College Station, TX: Stata
Corporation.
Stephan, Paula E., and Levin, Sharon G. 2001. “Exceptional Contributions to US Science by the
Foreign-born and Foreign-Educated.” Population Research and Policy Review 20: 59-79.
Walker, Robert, Mark Ellis, and Richard Barff. 1992. “Linked Migration Systems: Immigration
and Internal Labor Flows in the United States.” Economic Geography 68: 234-48.
White, Michael J., and Zai Liang. 1998.”The Effect of Immigration on the Internal Migration of
the Native-born Population, 1981-90.” Population Research and Policy Review 17:141-166.
Wright, Richard, Mark Ellis and Michael Reibel. 1997. “The Linkage between Immigration and
Internal Migration in Large Metropolitan Areas in the United States.” Economic Geography
73:234-54.
Zúñiga, Víctor, and Rubén Hernández-León (eds.) 2005. New Destinations: Mexican
Immigration in the United States. New York, NY: Russell Sage Foundation.
31
36
Percent of working age immigrants
34
Low-Skill
32
30
28
26
High-Skill
24
22
20
Table 3: Regression Models for Labor Market Outcomes for High-Skill Immigrant Men
Employment Underemployment Log Earnings
(1) (2) (3) (1) (2) (3) (1) (2) (3)
Industrial MSA .199 * .144 * .156 ** -.153 ** -.209 *** -.190 ** -.010 -.075 * -.005
(.087) (.071) (.059) (.059) (.061) (.067) (.046) (.047) (.024)
Age .020 * .019 * .040 *** .040 *** .028 *** .028 ***
(.009) (.009) (.003) (.002) (.002) (.002)
Eduational Attainment .143 *** .139 *** .245 *** .247 *** .092 *** .091 ***
(.020) (.020) (.022) (.023) (.007) (.006)
Citizen -.123 -.119 .061 .061 .015 + .016 +
(.104) (.104) (.052) (.052) (.017) (.018)
Speaks English .464 *** .452 *** -1.141 *** -1.146 *** .554 *** .544 ***
(.057) (.062) (.047) (.047) (.031) (.035)
Years in US .006 .007 -.018 *** -.018 *** .012 *** .012 ***
(.008) (.008) (.002) (.002) (.001) (.001)
Region of Birth (Canada=ref.)
Western Europe .143 .153 .471 *** .470 *** -.053 -.064 *
(.253) (.256) (.076) (.076) (.034) (.034)
Eastern Europe/USSR -.214 -.194 .575 *** .563 *** -.307 *** -.316 ***
(.265) (.266) (.121) (.125) (.041) (.040)
Asia .013 .031 .329 *** .331 *** -.284 *** -.289 ***
(.276) (.275) (.093) (.095) (.041) (.039)
Latin American .066 .074 1.323 *** 1.338 *** -.616 *** -.592 ***
(.297) (.287) (.110) (.110) (.051) (.046)
Caribbean -.515 * -.566 * .808 *** .805 *** -.557 *** -.535 ***
(.239) (.241) (.094) (.092) (.064) (.072)
South America -.037 -.077 .875 *** .878 *** -.465 *** -.451 ***
(.287) (.287) (.089) (.089) (.067) (.072)
African -.468 -.505 * .832 *** .852 *** -.539 *** -.545 ***
(.246) (.256) (.053) (.054) (.063) (.059)
Oceania .457 .500 .276 * .271 * .100 .079
(.691) (.694) (.137) (.136) (.100) (.105)
Unemployment rate -.070 + .052 *** .024 *
(.042) (.015) (.009)
GDP per capita -.004 .006 + .012 ***
(.005) (.004) (.002)
Housing costs -.031 .015 .002
(.024) (.012) (.007)
Industrial Job Growth
Scientific -.004 -.001 -.001
(.007) (.002) (.001)
Managerial .001 .000 .001
(.001) (.001) (.000)
Education .004 .004 .002 **
(.003) (.003) (.001)
Health .013 -.006 -.001
(.010) (.005) (.002)
Constant 3.257 * .065 * 1.409 -.153 *** -4.970 *** -6.069 *** 11.088 *** 8.038 *** 7.214 ***
(.087) (.330) (1.127) (.043) (.408) (.822) (.044) (.054) (.299)
Notes: Robust standard errors in parentheses; * p < .05; ** p < .01; *** p < .001; Employment and underemployment columns show logistic regression
coefficients.
35
Table 4: Regression Models for Labor Market Outcomes for High-Skill Immigrant Women
Employment Underemployment Log Earnings
(1) (2) (3) (1) (2) (3) (1) (2) (3)
Industrial MSA -.080 -.059 -.039 -.135 ** -.181 *** -.103 *** -.088 -.128 ** -.072 **
(.088) (.089) (.074) (.042) (.042) (.027) (.051) (.046) (.023)
Age .011 *** .011 *** .025 *** .042 *** .009 *** .010 ***
(.008) (.008) (.003) (.005) (.001) (.001)
Eduational Attainment .092 *** .089 *** .159 *** .164 *** .111 *** .110 ***
(.019) (.018) (.019) (.019) (.008) (.007)
Citizen .431 *** .432 *** -.125 *** -.127 *** .110 *** .108 ***
(.106) (.106) (.036) (.035) (.013) (.012)
Speaks English .638 *** .639 *** -1.171 *** -1.166 *** .663 *** .653 ***
(.105) (.107) (.051) (.050) (.048) (.050)
Years in US .023 *** .023 *** -.023 *** -.023 *** .015 *** .015 ***
(.005) (.005) (.002) (.002) (.001) (.001)
Region of Birth (Canada=ref.)
Western Europe -.352 -.351 .279 *** .278 *** -.075 -.091
(.257) (.257) (.071) (.070) (.050) (.050)
Eastern Europe/USSR -.642 * -.623 * .584 *** .586 *** -.256 *** -.263 ***
(.260) (.261) (.070) (.071) (.041) (.043)
Asia -.586 * -.586 * .315 *** .320 *** -.134 *** -.143 ***
(.231) (.231) (.063) (.062) (.037) (.038)
Latin American -.607 * -.624 * .711 *** .731 *** -.437 *** -.422 ***
(.302) (.302) (.079) (.074) (.045) (.043)
Caribbean -.574 * -.570 * .579 *** .581 *** -.296 *** -.278 ***
(.246) (.251) (.058) (.063) (.055) (.046)
South America -.461 -.465 .631 *** .628 *** -.333 *** -.319 ***
(.267) (.264) (.071) (.075) (.035) (.039)
African -.782 *** -.810 *** .441 *** .447 *** -.250 *** -.258 ***
(.219) (.213) (.089) (.084) (.066) (.062)
Oceania .000 .000 .162 .161 .124 .108
(.000) (.000) (.134) (.129) (.126) (.127)
Unemployment rate -.069 ** -.021 .022
(.024) (.013) (.011)
GDP per capita -.005 .005 ** .011 ***
(.004) (.002) (.002)
Housing costs -.016 -.003 .006
(.016) (.008) (.005)
Industrial Job Growth
Scientific .003 -.003 * .001
(.004) (.001) (.002)
Managerial -.001 -.001 * .000
(.001) (.000) (.000)
Education -.006 * .008 *** -.001
(.003) (.002) (.001)
Health .011 .000 -.001
(.007) (.003) (.003)
Constant 3.088 *** .595 *** 1.710 ** -.274 *** -2.724 *** -2.921 *** 10.616 *** 7.657 *** 6.813 ***
(.021) (.400) (.817) (.041) (.383) (.532) (.039) (.174) (.298)
Notes: Robust standard errors in parentheses; * p < .05; ** p < .01; *** p < .001; Employment and underemployment columns show logistic regression
coefficients.
36
Table 5: Regression Models of the Impact of High-Skill Immigration on Labor Market Outcomes of High-Skill Native Men
Employment Underemployment Log Earnings Out-Migration
(1) (2) (3) (1) (2) (3) (1) (2) (3) (1) (2) (3)
Change in percent high-skill .179 *** .186 *** .100 *** .012 .004 .002 -.028 -.030 -.029 ** .138 * .151 * .181 ***
immigrant (.030) (.029) (.024) (.021) (.022) (.018) (.029) (.026) (.010) (.054) (.059) (.056)
Industrial Metro -.179 -.203 -.259 .033 .076 .035 -.340 *** -.356 *** -.185 *** .559 ** .631 *** .346 **
(.212) (.217) (.136) (.054) (.069) (.054) (.097) (.089) (.034) (.180) (.180) (.195)
Change in percent high-skill -.047 -.056 -.004 -.011 -.012 -.003 .100 ** .096 *** .061 *** -.186 *** -.193 *** -.177 **
immigrant X Industrial metro (.055) (.057) (.045) (.020) (.022) (.016) (.032) (.028) (.013) (.048) (.050) (.072)
Percent high-skill immigrant -.043 *** -.045 *** -.044 *** -.002 -.001 -.001 .003 .002 .002 -.002 -.001 -.020
(.008) (.008) (.006) (.003) (.003) (.003) (.005) (.004) (.002) (.012) (.012) (.012)
Age (in years) .030 *** .030 *** .002 .002 .046 *** .046 *** -.098 *** -.098 ***
(.005) (.005) (.002) (.002) (.001) (.001) (.004) (.004)
Educational attainment .125 *** .122 *** .227 *** .227 *** .083 *** .081 *** .138 *** .136 ***
(.017) (.017) (.016) (.016) (.003) (.003) (.015) (.015)
Non-Latino black -.684 *** -.692 *** .328 *** .329 *** -.314 *** -.317 *** .148 .149
(.100) (.104) (.043) (.043) (.027) (.025) (.091) (.093)
Metro unemployment rate -.092 *** -.005 .012 * -.067
(.020) (.009) (.005) (.034)
Metro GDP per capita .005 -.002 .010 * -.002
(.005) (.002) (.001) (.005)
Metro housing costs .026 .004 .000 .038
(.019) (.010) (.007) (.030)
Industrial Job Growth
Scientific .003 -.002 * .000 .001
(.002) (.001) (.000) (.001)
Managerial .002 * .000 .000 -.003 **
(.001) (.000) (.000) (.001)
Education .002 .001 .001 ** .001
(.002) (.001) (.000) (.002)
Health .007 * -.002 * .002 * -.007 *
(.004) (.001) (.001) (.003)
Constant 4.227 *** 1.184 ** .778 -.617 *** -4.565 *** -4.542 *** 11.254 *** 8.265 *** 7.636 *** -3.288 *** -2.337 *** -2.412 *
(.244) (.391) (.709) (.051) (.284) (.467) (.085) (.111) (.223) (.191) (.271) (.807)
N 99857 99857 99857 97632 97632 97632 97632 97632 97632 111162 111162 111162
Log likelihood -10592 -10482 -10461 -63466 -62315 -62306 -119148 -111730 -111335 -22756 -21816 -21771
Notes: All models control for ACS survey year and metro population size; Robust standard errors in parentheses; * p < .05; ** p < .01; *** p < .001; Employment, underemployment,
and out-migration columns show logistic regression coefficients.
37
Table 6: Regression Models of the Impact of High-Skill Immigration on Labor Market Outcomes of High-Skill Native Women
Employment Underemployment Log Earnings Out-Migration
(1) (2) (3) (1) (2) (3) (1) (2) (3) (1) (2) (3)
Change in percent high-skill .087 *** .109 *** .084 *** -.026 -.030 -.010 -.023 -.023 -.026 ** .148 * .167 * .209 **
immigrant (.025) (.023) (.025) (.026) (.026) (.018) (.026) (.023) (.009) (.061) (.064) (.071)
Industrial Metro -.088 -.109 -.021 .011 .049 .002 -.291 * -.286 * -.157 ** .523 * .589 ** .373 **
(.184) (.184) (.094) (.053) (.065) (.062) (.129) (.114) (.048) (.214) (.223) (.217)
Change in percent high-skill -.025 -.043 -.042 .032 .029 .021 .077 ** .074 ** .052 *** -.167 ** -.179 ** -.187 *
immigrant X Industrial metro (.039) (.039) (.034) (.023) (.024) (.020) (.027) (.024) (.012) (.053) (.054) (.086)
Percent high-skill immigrant -.023 ** -.027 *** -.020 *** .006 .007 * .002 .004 .004 .006 ** -.001 .000 -.022
(.007) (.006) (.004) (.004) (.003) (.003) (.005) (.004) (.002) (.013) (.014) (.014)
Age (in years) -.001 -.001 .004 .004 * .014 *** .014 *** -.102 *** -.102 ***
(.004) (.004) (.002) (.002) (.001) (.001) (.004) (.004)
Educational attainment .159 *** .157 *** .167 *** .167 *** .107 *** .106 *** .113 *** .110 ***
(.017) (.018) (.015) (.015) (.004) (.004) (.017) (.017)
Non-Latino black -.597 *** -.583 *** .267 *** .272 *** .006 .006 *** -.127 -.119
(.064) (.066) (.025) (.026) (.017) (.015) (.075) (.076)
Metro unemployment rate -.067 *** -.017 -.004 -.073
(.012) (.009) (.007) (.042)
Metro GDP per capita .000 .000 .006 *** .003
(.003) (.001) (.001) (.006)
Metro housing costs -.023 .009 -.008 .042
(.014) (.010) (.006) (.033)
Industrial Job Growth
Scientific .000 -.001 .001 * .002
(.001) (.001) (.000) (.002)
Managerial .002 * -.001 * .001 -.003 **
(.001) (.000) (.000) (.001)
Education .000 .000 .001 .000
(.001) (.001) (.000) (.003)
Health .001 -.003 ** .001 * -.005
(.004) (.001) (.000) (.008)
Constant 3.884 *** 1.371 *** 2.363 *** -.838 *** -3.858 *** -3.868 *** 10.694 *** 8.411 *** 8.311 *** -3.464 *** -2.009 *** -2.342 **
(.218) (.300) (.549) (.063) (.270) (.388) (.100) (.148) (.204) (.235) (.384) (.844)
N 105127 105127 105127 102469 102469 102469 102469 102469 102469 134579 134579 134579
Log likelihood -12374 -12266 -12252 -64583 -63954 -63935 -135038 -133295 -133135 -25301 -24225 -24162
Notes: All models control for ACS survey year and metro population size; Robust standard errors in parentheses; * p < .05; ** p < .01; *** p < .001; Employment, underemployment,
and out-migration columns show logistic regression coefficients.
38
Native Men B -- B B / --
Native Women B -- B B / --
Notes: "B" refers to effects that are significantly beneficial (at p < .05) to those in industrial cities;
"D" to effects that are significantly disadvantageous; "--" to null effects; and "n/a" to non-
applicable models (out-migration not an outcome for immigrants)
39
Appendix Table 1: High-Skill Immigrant and Native Sample Sizes, Total Number of High-Skill Immigrants, and Immigrant Share of
High-Skill Population, by Metropoltian Area
Notes: ACS sample includes prime-age, college-educated men and women; "Immigrant gateways" are italicized