Changes in the Relative Economic Performance of Immigrants to Great Britain and the United States, 1980–2000
John Schmitt and Jonathan Wadsworth
Abstract
We compare the relative labour market performance of immigrants in the USAand in Britain over the period 1980–2000, when the stocks of immigrants wererising in both countries alongside differential shifts in demand and changes tolabour market institutions. We find that the average relative employment prospectsof immigrants are generally better in the USA than in Britain, particularlyfor non-white英國dissertation網immigrants, while the average relative wage prospects for immigrantsare generally better in Britain, particularly for men. Over time, relativewage and employment prospects for immigrants to the USA appear to havedeteriorated, particularly among women, in a way that is not as apparent inBritain.
1. Introduction
Over the past 20 years, one of the most notable labour market developments,common to both Britain and the USA, has been a rise in the number ofimmigrants coming to live and work. The share of immigrants increased fromabout 8 per cent of the US working age population in 1980 to about14 per cent in 2000 and from 8 per cent to around 11 per cent in Britain,a boost to the respective domestic populations of some 16 million and1.8 million, respectively.1 Over the same period, the labour markets of Britainand the USA have been subject to two recessions and latterly a prolongedperiod of growth. Both economies have also experienced increased trade withthe developed world, technological innovation and changes in labour marketinstitutions, notably declining union presence and tightening of welfare,alongside periodic falls in the real value of the minimum wage, or its closestJohn Schmitt is at the Centre for Economic and Policy Research. Jonathan Wadsworth is atRoyal Holloway College, University of London, the Centre for Economic Performance at theLondon School of Economics and IZA Bonn.
British Journal of Industrial Relations doi: 10.1111/j.1467-8543.2007.00646.x45:4 December 2007 0007–1080 pp. 659–686
© Blackwell Publishing Ltd/London School of Economics 2007. Published by Blackwell Publishing Ltd,9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.British equivalents the wages set by wage councils. All these factors appear to
have contributed towards a deterioration of the relative labour market performanceof less-skilled workers manifested in widening inequalities in wageand employment levels. However, these two countries do differ significantly
along other dimensions, notably with regard to patterns of demand andsectoral composition, welfare systems, employment protection legislation,
the extent of union coverage, working hours and workplace regulation.
Moreover, despite similar aggregate trends in immigration, the compositionof immigrants in the two countries is somewhat different.#p#分頁標題#e#
Against this background, the aim of this article is to give an overview of the
impact that the combined effect of these institutions and other changes mayhave had on the wage and employment prospects of immigrants relative to
the native-born populations in these two countries. Many recent studies havefocused on the impact of rising immigration on native-born workers, (e.g.Borjas 2003; Card 2001, 2005).2 Relatively, little http://www.400190.com/dissertation_writing/Ecommerce/current attention has beengiven to the performance of immigrants in the host labour market. For policypurposes, it is important to know whether immigrants benefit from economicconditions in the host country and, if so, how quickly they adapt and howthis adaptation process changes with time or across policy regimes. Comparative
studies of the experience of immigrants in different countries arerarer still. Antecol et al. (2006) is a notable recent exception, studying immigrantsin the USA, Australia and Canada in 1980 and 1990, and looking atthe influence of each country’s institutional structures on relative immigrantperformance. They argue that assimilation of immigrant wages towards thatof native-born workers is faster in countries such as the USA, that make useof fewer institutions in wage determination.
Our article builds on this comparative approach by first extending theperiod of analysis for the USA, adding more recent micro data from the
2000 US Census and by comparing the experience of immigrants to anotherEnglish-speaking country, Great Britain, over the period 1980–2000 using a
time series of cross-sections taken from the General Household Survey(GHS). We are interested in measuring any changes over time in relativeimmigrant performance. To do this, we look at three different, but related,concepts. First, we examine the average gap between immigrant wages andemployment relative to the native-born. Second, we look at the years in thehost country needed before 留學生dissertation定制immigrants’ earnings (or employment probabilities)reach parity with those of native-born workers with comparable
characteristics (the assimilation profile), and how this varies over time andwithin each country. Third, we track changes in the pace of immigrants’longer-term earnings growth or employment prospects relative to the
native-born.Earlier research has, of course, addressed some of the issues we examinehere. Chiswick (1978) used data from the 1970 US Decennial Census todemonstrate that the earnings of male immigrants to the USA converged on
the earnings of US-born males after 10 to 15 years in the country. Borjas(1985) concluded that use of a single cross-section of data on immigrants
660 British Journal of Industrial Relations© Blackwell Publishing Ltd/London School of Economics 2007.
could bias the estimate of the speed of convergence because a singlecross-section could not control adequately for changes in the ‘quality’ of#p#分頁標題#e#
immigrants over time. Borjas also noted, but did not emphasize, that deteriorationsin the (absolute or relative) demand for workers with the same
skills as those brought by immigrants could also lead to a slow-down inconvergence. Antecol et al. (2003) is a rare recent example of an examinationof the determinants of performance among female immigrants. They stress
the role of country of origin in explaining relative differences in educationalattainment and income between female immigrants to the USA, Australiaand Canada, and native-born women.
For Britain, Bell’s (1997) study of the relative wage performance of maleimmigrants, using the same data source as this study, for a 20-year periodending in 1992, concluded that the wage assimilation process was much
quicker for white immigrants. Dustmann and Fabbri (2005) and Clark andLindley (2006) use data from the Labour Force Survey covering the latter
third of our sample period to study male immigrant assimilation in wage andemployment by ethnicity. Both studies highlight the considerable diversity inthe patterns of earnings and employment assimilation.
Our research extends these earlier efforts in several dimensions. First, weexamine data across three recent decades, centred around 1980, 1990 and 2000.Second, unlike most of the earlier research, we analyse the experience of men
and women and differences by ethnicity within gender. While studies of wagesfor women (and men) may of course be influenced by selective labour forceparticipation based on unobservables, examination and discussion of any
trends in the relative labour market performance of immigrant women is toimportant a topic to neglect. Third, unlike most of the earlier research, weestimate relative performance with respect to earnings and employment.
Finally, by comparing our results over time and across two countries undergoingsubstantial restructurings of their wage distributions, but with differentialinstitutions and immigration histories, it should be possible to get a
broader insight into the nature of immigrant labour market adjustment.The article is structured as follows. Section 2 provides a brief summary of
immigration policy in both countries while Section 3 outlines the conceptsused to measure the extent of convergence of immigrants’ labour market
performance towards that of the native-born. Section 4 discusses the datasets
used in this study. Section 5 outlines the main results of the study while
Section 6 concludes.
2. Immigration policy
US immigration numbers are influenced by both documented and undocumentedimmigration. Formal immigration policy has placed country-specificlimits on the numbers of immigrants since the 1960s, and the system ofpreference categories embedded in visa applications is skewed towards skilledworkers or workers in areas in which labour is in short supply, along withChanges in Relative Economic Performance of Immigrants 661
© Blackwell Publishing Ltd/London School of Economics 2007.protection for refugees and reunification of families (Wasem 2004a). The#p#分頁標題#e#
1990 Immigration Act introduced a system of visas allowing entry of hightechworkers with a sponsoring employer for up to six years (with the subsequentchance of applying for US citizenship) and also raised the share ofemployment-based visas in the total immigrant flow. In practice, this hasmeant that since the beginning of the 1990s, documented immigration hasbeen oriented towards highly qualified http://www.400190.com/dissertation_writing/Ecommerce/immigrants, although Baker (2006)notes that substantial barriers to higher-skilled immigration remain.
However, because dependants of immigrants also qualify for entry, familyties account for around 60 per cent of immigrant entries. Undocumented
immigration to the USA is at least as important a source of immigrationflows as documented immigration. The Congressional Research Service estimated
that ‘unauthorized aliens’ rose from about 1.9 million in 1988 tobetween 7.0 and 8.5 million by 2000, the latest year of data analysed below(see Wasem 2004b). Undocumented immigrants tend to have lower educationalqualifications than documented immigrants.
As regards Britain, policy has, at times, sought to encourage immigrants
from both ends of the skill distribution. In part, this is because previous
immigration policy here has been greatly influenced by its links with former
colonies. Before 1962, any Commonwealth or Irish citizen had the right of
entry. A system of work permits was introduced after that point and the
principle of right of entry to Commonwealth citizens was abolished in 1973,
replaced by a system of work permits (again skewed towards skilled workers
in short supply) and entry rights for dependants that has continued, subject
to periodic modifications, to this day. In addition, there is an unrestricted
right of entry to any citizen of the European Union (EU) member states.3 The
only official estimate of illegal immigration based on similar methods as in
the USA (Woodbridge 2005), puts the number of illegal immigrants, adults
and children combined, in 2001 at 430,000.4 Any undocumented immigrants
in the population are probably underreported in the US census (Heer and
Passel 1987), and, while we know of no estimates relating to this, the same
probably also applies to the British dataset that we employ below. This could
make our estimates of immigrant performance unrepresentative of all
immigrants. If, however, documented and undocumented immigrants differ
systematically by observable characteristics, then the results we present
below which control for observables should net out any such selection
effects.
3. Measuring the labour market performance of immigrants
Ideally, we would like to measure immigrant performance using data from a
large panel survey that followed several waves of workers, both native and#p#分頁標題#e#
foreign born, over a long enough period of time to give a clear picture of
immigrant workers’ performance, both in real terms and relative to other
workers. Unfortunately, no such large, multi-wave, panel surveys exist,
662 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
forcing us to rely instead on measures afforded by large, repeated, crosssectional
surveys of the labour force.
Average Gap
The first measure captures whether, on average, the wages or employment of
immigrant workers equal native-born levels. One simple way to do this is to
add a dummy variable for immigrant status into a wage or employment
equation. The coefficient on the immigrant dummy gives the average difference
in pay or employment between immigrant and native-born workers. By
comparing this coefficient over time or across different sub-groups of immigrants,
we can assess the overall change in the relative labour market performance
of immigrants.
Returns to Experience in the Host Country
Much of the debate over immigration effects concerns the speed of adaptation
to a host country’s labour market institutions. Since the Census and
GHS record year of arrival,5 we can regress hourly earnings (or employment
status) against dummy variables that group immigrants by the number of
years in the host country (0–5, 6–10, 11–15, etc.), with each of these yearssince-
migration variables taking the value zero for native-born workers.
These dummy variables effectively partition the binary immigrant status
dummy variable into mutually exclusive sub-categories. Recent immigrants,
on average, might be expected to earn less than the average worker if the
skills and experience that they bring are not rewarded as much as those
of native-born workers. This may be attributed, for example, to language
barriers, unfamiliarity with the host countries’ job market institutions,
institutional impediments to practice or discrimination.6 In these cases, the
coefficient on the 0–5 years in the country dummy variable will be negative.7
Over time, if there is assimilation, then the wage and employment prospects
of the average immigrant worker may converge on that of the average nativeborn
worker. At that point, the coefficient on that particular years-in-thecountry
dummy variable would equal zero. For immigrants beyond that
point, the coefficient on the other years in the country variables should
remain close to zero.
Estimating such equations at different periods of time can reveal changes
in immigrant performance in at least two ways. The first is through changes
in the estimated point of convergence. If performance relative to native-born
workers increased, for example, the point of convergence might occur earlier,#p#分頁標題#e#
say, after 11–15 years, instead of after 16–20 years. Conversely, if performance
declined, we might expect to see the estimated point of convergence
recede. Second, if relative immigrant performance improved, we might expect
to see lower penalties for immigrants at shorter durations or higher premiums
for longer-duration immigrants, independent of any changes in the point of
convergence with native-born wages.
Changes in Relative Economic Performance of Immigrants 663
© Blackwell Publishing Ltd/London School of Economics 2007.
One obvious drawback is that immigrants in the host country for a long
time are almost mechanically much older than the average native-born
worker and these extra years of (potential) work experience act to raise the
wages of immigrants, for reasons that have little to do with assimilation.
Immigrant and native-born workers differ with respect to educational attainment,
and the composition of immigrant population by country of origin
may also vary greatly from one sample point to the next. We therefore add a
basic set of controls to try to account for differences in these factors between
both immigrants and native-born and between samples over time.
A second well-known limitation of this approach is that, in practice, we
cannot distinguish between years-since-migration and cohort effects (Borjas
1985, 1987, 1994), without imposing restrictions on the data.8 The coefficients
on the years-since-migration variables, rather than representing the returns
to time in the host country, could also reflect year-of-arrival cohort effects—
systematic differences in the unobserved characteristics (and/or prevailing
macroeconomic conditions) of the cohorts that arrived in different periods.
The separation of cohort from time-in-the-country effects may contribute to
an understanding of any changes we observe, but, because the main aim of
the article is to measure whether performance has changed over time, the first
task is to determine whether the relative position of immigrant workers has
changed.
Long-Term Relative Growth
In an attempt to address the issue of unobserved cohort effects, our final
measure compares 10-year gains (or losses) experienced by immigrants in
recent periods with 10-year gains (or losses) experienced by immigrants in an
earlier period. This approach is the closest in spirit to the ideal based on a
multi-wave panel survey. In the absence of true panel data, we can only
follow changes in the average wages or employment levels of cohorts (by
year-of-arrival), not the actual changes experienced by an individual. Moreover,
because we do not have retrospective wage and employment data, in
order to include a particular cohort in our analysis, we must observe the
cohort in at least two of the three time periods we use.#p#分頁標題#e#
With three data points 10 years apart, we can compare real wage and
employment growth for different groups of workers over two time periods,
1980–1990 and 1990–2000. We can, for example, calculate the average wage
change between 1980 and 1990 for recently arrived immigrants by looking at
the average wage for the cohort that arrived in 1975–1979 in both the 1980
and 1990 data. We can then measure the average change in wages or employment
between 1990 and 2000 for recently arrived immigrants in 1990 by
calculating the average for the cohort that arrived in 1985–1989 in the 1990
and 2000 data. By choosing cohorts with the same years-since-arrival in the
1980, 1990 and 2000 data, we can construct a set of matched cohorts that can
add to the understanding of mobility for immigrants with different amounts
of time in the host country. If immigrant performance increased between the
664 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
first period and the second period, we would expect real wage or employment
growth to be faster for the cohort that arrived in 1985–1989, say, than it was
for the cohort that arrived in 1975–1979. Note that this does not necessarily
imply that the rate of convergence towards native-born wage or employment
levels has increased because convergence is the product of wage growth and
the initial entry wage (or employment) level. We return to this issue below.
A simple difference-in-differences strategy will compare the difference in
10-year growth rates in real wages across matched cohorts of immigrants
with the difference in 10-year growth rates in real wages for native-born
workers over the same periods. To do this9 we pool the data for two consecutive
periods for each cohort and then fit a regression of the form:
Wit = a + bGG+ bMM+ bGMG*M+ bXXit + uit (1)
where Wit is the wage (or employment) of individual i at time t; G is a binary
variable that takes the value 1 for observations in the second of the two
periods and 0 for observations from the first period; M is a dummy variable
indicating immigrant status; Xit is a set of control variables for individual i
observed at time t and uit is a residual term. The coefficient on the interaction
term of G with M, bGM, the difference-in-difference estimate, is an estimate of
the growth rate for any immigrant cohort over and above that of native-born
workers across the two periods, controlling for differences in the age, educational
attainment, region of residence and area of origin embodied in the
control vector X. This difference-in-difference estimator nets out any cohort
fixed effects, such as the size of the entry cohort that might otherwise influence
any estimates based on single cross-sections, such as the calculation of#p#分頁標題#e#
the assimilation profiles outlined above.10 However, the technique does
assume that any factors not present in the observed set of controls X, and
which also change over the period, have similar effects on immigrants and
native-born workers and so can be differenced away. If not, then the
difference-in-difference estimates will incorporate some of this effect. Given
these caveats, if immigrants were doing better we might expect to see the
difference-in-difference estimates become larger over time, indicating larger
wage growth for immigrants relative to native-born. However, faster relative
wage growth in itself does not necessarily imply convergence because convergence
depends on both wage growth and the initial wage gap. If the latter
falls over time, then faster relative wage growth may not imply wage convergence.
For this reason, we integrate the evidence from the different
approaches in Section 5.
4. The data
The main source of US data is a consistent extract of labour market variables
from the 1 per cent version of the Public Use Micro data Sample (PUMS) for
the 1980, 1990 and 2000 decennial censuses. In addition to an extensive set of
Changes in Relative Economic Performance of Immigrants 665
© Blackwell Publishing Ltd/London School of Economics 2007.
standard labour market variables, PUMS includes information on an individual’s
country of birth — on which we define an immigrant as someone
born outside the USA—and year of arrival. The dependent variable for pay
is the natural log of the average gross hourly wage in the calendar year before
each census for those currently in work, deflated by the relevant consumer
price index. The dependent variable measuring presence in employment is
derived from self-assessed labour market status questions in the US census,
which are based on the International Labour Organisation (ILO) definition
of employment in 2000 and are close to the ILO definition in earlier years.
The data are coded 1 if the individual is in work, employee or self-employed,
and 0 otherwise. This latter category includes both the unemployed and
economically inactive.
The data for Britain are taken from the GHS from 1977 to 2003. As the
British datasets are much smaller than the US data, we pool seven years of
data centred on each of the three years covered by the US census so that the
British data coded as 1980, for example, contain individual responses from
the 1977 to 1983 GHS. This gives a total of around 100,000 observations in
each — ‘decade’, of which about 50,000 have wage data. The pay variable is
the natural log of the average gross hourly wage in the survey reference week
for all current employees, deflated by the appropriate monthly value of the
retail price index.11 The binary employment status variable is derived from#p#分頁標題#e#
self-assessed labour market status questions, which, from 1998 onwards,
correspond to the ILO definition of employment and are close to the ILO
definition before that. As with the US data, we include both unemployed and
inactive in the definition of those not in work. The definitions of immigrant
and native-born are similar to those used for the US data. An immigrant to
Britain is an individual who was born outside the host country.12 To net out
any effects caused by differences in education, we include dummy variables
that capture the level of education of each individual in either dataset
(primary only, incomplete secondary, secondary, some college, degree and
higher degree). These are based on the highest level of educational attainment
recorded in both datasets. As the British data contain more information on
qualifications, the British education data are grouped so as to correspond as
closely as possible to the six categories allowed for in the US census.13 The
British data on immigrant qualifications are also likely to underestimate the
level of education. Manacorda et al. (2005) show that many immigrants who
have qualifications obtained overseas are wrongly classified as having ‘other’
qualifications when their true level is often higher. We are, however, obliged
to use qualifications rather than age left education in order to facilitate
comparisons across the two country datasets. While the education systems
and qualifications awarded in the two countries differ significantly, this is the
best we can do with the available data. The sample is restricted to men and
women aged 18 or more but below pension age in both countries. To net out
age differences, we add five age-range dummy variables to the set of controls,
(18–24, 25–34, 35–44, 45–54 and 54+). We then interact the age and education
dummies to pick up any differential within-group effects.
666 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
Different cohorts of immigrants could consist of different sets of (unobservable)
skills, such as English language ability, quality of education in
addition to observable differences in age, education or region of residence.
Changes in these unobserved characteristics over time could help explain any
changes in the relative labour market position of immigrants. To proxy these
effects, we introduce a set of area of origin dummy variables, which reflect the
effect in each period of originating from these different areas. As the British
data are quite small, we are obliged to restrict area of origin to eight broad
geographic areas reflecting the share of the area in the total stock of immigrants
and possible English language fluency. For Britain, we use Ireland, the
rest of Europe, India and Pakistan, Africa, Central and Southern America,#p#分頁標題#e#
the Caribbean, the Old Commonwealth and a residual ‘Other’ category. For
the USA, the area groupings are: Mexico, Canada, the Americas, the
Caribbean, Africa, the English-speaking countries of Australia, New
Zealand, Ireland and the UK, the rest of Europe and a residual ‘Other’
category. All regressions also control for either 51 US states of residence or
11 British regions to net out any regional effects on wages or employment
combined with the uneven geographic distribution of immigrants in both
countries. It is also possible that return migration will influence these estimates.
If immigrants leave the host country, then comparing years of arrival
estimates at a point in time may be biased by differential return rates across
the year of arrival groupings.14 We make no distinction between immigrants
who arrived as children and those who arrived as adults, even though the
existing evidence suggests that the experience of these two groups may be
different (Kossoudji 1989) because our intention is to focus on the average
trends in immigrant performance.
5. Results
Table 1 outlines the changes in immigrant shares, area of origin and country
of origin for the USA and Britain over the sample period. Net immigration
has risen appreciably in both countries, but more so in the USA where the
immigrant share has more than doubled, from 7.2 to 14.8 per cent between
1980 and 2000. The immigrant share grew from 7.6 to 9.7 per cent in Britain
over the same period. Rising immigration in both countries has been associated
with a change in the country of origin mix. In the USA, the trend is for
more immigration from the Americas at the expense of Europe. In particular,
the share of immigrants from Mexico was even larger in 2000 than in 1980.
Nearly one in three immigrants to the USA is now from Mexico. In contrast,
immigration by country of origin is becoming much more heterogeneous in
Britain. While the share of immigrants from Europe has also fallen in Britain
over this period and the share of immigrants from Africa has risen, by 2000,
no single country dominates the country of origin mix to the same extent as
in the USA. India has overtaken Ireland as the source country for the largest
number of immigrants, but India accounts for less than 10 per cent of the
Changes in Relative Economic Performance of Immigrants 667
© Blackwell Publishing Ltd/London School of Economics 2007.
British immigrant population. Given these notable changes over the sample
period, we control for area of origin effects in the regression estimates that
follow.15
Relative Labour Market Performance
Table 2 outlines the change in the relative employment position of immigrants,
separately for men and women and then by ethnicity for each group.
TABLE 1
Immigrant Shares, Area of Origin and Years in Host Country, 1980–2000#p#分頁標題#e#
1980 1990 2000
Britain
Immigrant share % from: 7.6 8.3 9.7
Europe 41.6 35.9 32.2
Asia 28.0 32.9 32.4
Africa 11.8 15.0 19.7
Americas 16.3 13.2 11.8
Oceania 2.3 3.0 3.9
Largest senders (% of all immigrants)
1 Ireland (19.3) Ireland (15.8) India (9.6)
2 India (13.3) India (12.3) Ireland (8.9)
3 Jamaica (6.2) Pakistan (7.3) Pakistan (6.8)
4 Pakistan (5.8) Germany (4.4) Germany (5.3)
5 Italy (4.3) Jamaica (4.0) Bangladesh (4.4)
Years in Britain
0–5 9.3 10.2 19.8
5–10 13.7 8.2 12.9
11–15 15.9 10.9 10.1
16–20 19.0 13.9 8.4
21–30 24.0 31.1 18.8
31+ 18.0 25.9 30.0
USA
Immigrant share % from: 7.2 10.4 14.8
Europe 33.1 21.7 15.4
Asia 21.6 27.1 26.4
Africa 1.9 2.0 2.8
Americas 42.8 48.6 54.8
Oceania 0.7 0.6 0.6
Largest senders (% of all immigrants)
1 Mexico (16.3) Mexico (23.0) Mexico (30.7)
2 Germany (6.5) Germany (4.9) Philippines (4.7)
3 Canada (5.9) Philippines (4.9) Vietnam (3.3)
4 Cuba (4.9) Cuba (3.7) Germany (3.2)
5 UK (4.6) Canada (3.5) India (3.0)
Years in the USA
0–5 24.1 22.1 21.3
5–10 18.4 19.1 17.2
11–15 15.8 15.1 16.0
16–20 12.6 13.3 14.3
21–30 17.7 17.9 18.0
31+ 11.4 12.5 13.2
668 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
TABLE 2
Effect of Immigrant Status on Employment Probability (Marginal Effects)
Great Britain USA
1980 1990 2000 1980 1990 2000
Men
Immigrant -0.087 (0.007)* -0.078 (0.008)* -0.089 (0.010)* 0.014 (0.002)* 0.028 (0.002)* 0.004 (0.001)*
Of which:
White immigrant -0.058 (0.009)* -0.028 (0.010)* -0.057 (0.015)* 0.017 (0.003)* 0.011 (0.003)* 0.013 (0.003)*
Other immigrant -0.119 (0.011)* -0.126 (0.012)* -0.138 (0.016)* 0.013 (0.002)* 0.037 (0.002)* 0.001 (0.002)
Women
Immigrant -0.079 (0.008)* -0.073 (0.009)* -0.113 (0.010)* -0.007 (0.002)* -0.031 (0.002)* -0.074 (0.002)*
Of which:
White immigrant -0.045 (0.011)* -0.017 (0.012) -0.078 (0.015)* -0.020 (0.003)* -0.048 (0.003)* -0.065 (0.003)*
Other immigrant -0.128 (0.013)* -0.141 (0.013)* -0.155 (0.016)* 0.008 (0.003)* -0.020 (0.003)* -0.075 (0.002)*
Notes: Sample sizes for Great Britain are 116,960, 100,045 and 63,972, in 1980, 1990 and 2000, respectively; 60,326, 51,280 and 32,281 for men and 56,634, 48,765
and 31,691 for women. For the USA, sample sizes are 1,369,240, 1,518,093 and 1,708,990 in 1980, 1990, 2000, respectively; 668,154, 744,085 and 843,639 for men,
and 701,086 774,008 and 865,351 for women. Coefficients are percentage point differences (/100) relative to native-born workers evaluated at the sample means
of the control variables (four dummy variables for age, five dummies for education (based on the categories in Table 1) and their interactions along with 50 (11)#p#分頁標題#e#
state (region) dummies for the USA and Britain, respectively). Full results available from the authors upon request.
* Significant at 5 per cent; robust standard errors in brackets.
Changes in Relative Economic Performance of Immigrants 669
© Blackwell Publishing Ltd/London School of Economics 2007.
Once we control for differences in the age, educational composition and
region of residence between immigrants and native-born along with area of
immigrant origin, then there is a significant employment disadvantage for
immigrants to Britain, both male and female. This relative disadvantage has
remained unchanged for men over time, but has increased somewhat among
immigrant women. In the USA, male immigrants experience a small relative
employment premium relative to native-born men, which appears to have
narrowed recently. Immigrant women in the USA have experienced a
widening employment gap relative to native-born women over the sample
period. The size of this gap is now similar to that observed among immigrant
women in Britain during the 1980s. In Britain, the aggregate immigrant
employment gaps disguise larger gaps for non-white immigrants. In the USA,
there is little difference between non-white and white immigrants.16
Conditional on being in work, Table 3 presents estimates of the log hourly
wage gap between immigrants and native-born workers in 1980, 1990 and
2000. In Britain, immigrant men, on average, receive lower wages relative to
native-born workers and this penalty appears to have risen somewhat, from
4.5 to 7.5 per cent over time. There is also an average pay penalty for women
of between 3 and 5 per cent, but this does not appear to have changed
significantly over the sample period. The average immigrant to the USA
receives lower wages than US-born workers and this pay gap has increased
over time, from around 7 to 9.5 per cent for men and from 1 to 5 per cent for
women. In both countries, the pay gap is larger for non-white immigrants,
both male and female. Indeed, among white immigrants to both Britain and
the USA, men and women are paid, on average, similar amounts to nativeborn
workers. Some of these widening gaps are consistent with rising wage
inequality (Butcher and DiNardo 2002), which has risen appreciably in both
countries over the sample period. If immigrants tend to be located towards
the bottom of the wage distribution, this could help explain some of the rising
average gap net of characteristics that we observe here.
Returns to Time in the Host Country
Table 4 and Figure 1 summarize the average relative employment position of
immigrants over time conditional on years spent in the host country. The
table presents the employment gap between native-born workers adjusted for
differences in age, education, region of residence and area of origin. Reading#p#分頁標題#e#
down each column gives an indication of the speed of convergence of immigrants’
wages to those of native-born workers. For men in 1980, (column 1),
for example, recent male adult immigrants (0–5 years) to Britain were about
32 percentage points less likely to be in employment than the average Britishborn
worker. Thereafter, the immigrant employment gaps fall with time
spent in the country but remain negative. Therefore, male immigrant employment
rates never converged to those of the British-born population. Over
time, the employment gap on entry (0–5 years) has fallen for male immigrants
to Britain, but there has been no fall in the gap for those in the country for
670 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
TABLE 3
Effect of Immigrant Status on Log Real Hourly Wage, 1980–2000
Great Britain USA
1980 1990 2000 1980 1990 2000
Men
Immigrant -0.046 (0.007)* -0.093 (0.013)* -0.073 (0.014)* -0.069 (0.004)* -0.095 (0.003)* -0.094 (0.003)*
Of which:
White immigrant -0.003 (0.010) 0.008 (0.017) 0.030 (0.021) 0.025 (0.006)* 0.033 (0.005)* 0.027 (0.005)*
Other immigrant -0.095 (0.010)* -0.218 (0.018)* -0.181 (0.021)* -0.146 (0.006)* -0.159 (0.004)* -0.134 (0.003)*
Women
Immigrant -0.031 (0.009)* -0.006 (0.011) -0.048 (0.013)* -0.011 (0.004)* -0.041 (0.004)* -0.048 (0.003)*
Of which:
White immigrant -0.025 (0.011)* 0.035 (0.014)* -0.017 (0.018) 0.010 (0.006) 0.003 (0.005) -0.008 (0.005)
Other immigrant -0.041 (0.013)* -0.071 (0.017)* -0.110 (0.020)* -0.026 (0.006)* -0.065 (0.004)* -0.063 (0.003)*
Notes: Sample sizes for Great Britain are 69,467, 54,794 and 36,784 in 1980, 1990 and 2000, respectively; 39,666, 28,230 and 18,324 for men; and 29,801, 26,564
and 18,460 for women. For the USA, sample sizes are 998,010 1,156,840 and 1,303,470 in 1980, 1990 and 2000, respectively; 554,640, 613,991 and 680,697 for men;
and 443,370, 542,849 and 622,773 for women. PUMS1% and pooled GHS files. Coefficients are log point differential (/100) relative to native-born workers.
* Significant at 5 per cent; robust standard errors in brackets.
Changes in Relative Economic Performance of Immigrants 671
© Blackwell Publishing Ltd/London School of Economics 2007.
longer spells of time. If anything the employment penalties at longer durations
have risen, so that employment convergence is further away than ever.
On this basis it would seem that mobility among male immigrants to Britain
has fallen over time. Among immigrant women to Britain, the entry employment
gap has remained constant. The gap has risen at intermediate durations
over the sample period and fallen at longer durations. Unlike for immigrant
men, on this basis, employment rates for longer-term female immigrants to
Britain do appear to have drawn closer to those of the native-born over the#p#分頁標題#e#
latter part of the sample period.
Figure 1 shows that for Britain, this obscures a widening of the entry
employment gap among non-white male immigrants (column 1, row 3) and
the emergence of an employment premium among recently arrived white
male immigrants over time (column 1, row 2). Moreover, the relative employment
gap among immigrants who have been in the country for more than
30 years appears to have risen for white males and fallen among non-white
males and females over time.17 Whether these patterns are influenced by
TABLE 4
Effect of Time in Host Country on Relative Employment Probability of Immigrants
Great Britain USA
1980 1990 2000 1980 1990 2000
Men
0–5 years –0.320 –0.310 –0.239 0.027 0.042 0.018
(0.034)* (0.035)* (0.025)* (0.005)* (0.003)* (0.003)*
6–10 years –0.135 –0.163 –0.131 0.094 0.084 0.038
(0.025)* (0.040)* (0.031)* (0.003)* (0.003)* (0.003)*
11–15 years –0.104 –0.132 –0.114 0.101 0.084 0.031
(0.023)* (0.032)* (0.035)* (0.003)* (0.003)* (0.003)*
16–20 years –0.078 –0.089 –0.123 0.091 0.083 0.037
(0.019)* (0.026)* (0.036)* (0.004)* (0.003)* (0.003)*
21–30 years –0.068 –0.058 –0.081 0.099 0.082 0.051
(0.017)* (0.019)* (0.024)* (0.004)* (0.003)* (0.003)*
31+ years –0.054 –0.083 –0.078 0.077 0.073 0.047
(0.016)* (0.020)* (0.019)* (0.005)* (0.003)* (0.003)*
Women
0–5 years –0.233 –0.241 –0.238 –0.124 –0.176 –0.227
(0.029)* (0.031)* (0.024)* (0.008)* (0.006)* (0.004)*
6–10 years –0.119 –0.226 –0.184 –0.006 –0.057 –0.137
(0.025)* (0.033)* (0.028)* (0.008) (0.006)* (0.004)*
11–15 years –0.108 –0.191 –0.179 0.023 –0.025 –0.093
(0.025)* (0.031)* (0.031)* (0.008)* (0.006)* (0.005)*
16–20 years –0.087 –0.067 –0.111 –0.026 –0.005 –0.048
(0.025)* (0.026)* (0.033)* (0.009)* (0.006) (0.005)*
21–30 years –0.074 –0.047 –0.037 –0.009 –0.005 –0.027
(0.022)* (0.020)* (0.022) (0.008) (0.005) (0.004)*
31+ years –0.063 –0.053 –0.020 –0.029 –0.025 –0.040
(0.025)* (0.024)* (0.021) (0.009)* (0.006)* (0.005)*
Notes: See Table 2. Estimates are net of controls for age, education, host country region and
area of origin as in Table 3. All coefficients measured relative to reference category of
native-born workers.
672 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
FIGURE 1
Employment Rate Differential Relative to Native-Born by Years in Host Country, 1980–2000.
-.35 -.25 -.15 -.050 .05.1#p#分頁標題#e#
0-5 6-10 11-15 16-20 21-30 31+
GB Men
1980 2000
-.35 -.25 -.15 -.050 .05 .1
0-5 6-10 11-15 16-20 21-30 31+
US Men
1980 2000
-.35 -.25 -.15 -.050 .05.1
0-5 6-10 11-15 16-20 21-30 31+
GB Women
1980 2000
-.35 -.25 -.15 -.050 .05.1
0-5 6-10 11-15 16-20 21-30 31+
US Women
1980 2000
-.35 -.25 -.15 -.050.05.1
0-5 6-10 11-15 16-20 21-30 31+
GB Men: White
1980 2000
-.35 -.25 -.15 -.050 .05.1
0-5 6-10 11-15 16-20 21-30 31+
US Men: White
1980 2000
-.35 -.25 -.15 -.050 .05.1
0-5 6-10 11-15 16-20 21-30 31+
GB Women: White
1980 2000
-.35 -.25 -.15 -.050 .05.1
0-5 6-10 11-15 16-20 21-30 31+
US Women: White
1980 2000
-.35 -.25 -.15 -.050 .05.1
0-5 6-10 11-15 16-20 21-30 31+
GB Men: Non-white
1980 2000
-.35 -.25 -.15 -.050 .05 .1
0-5 6-10 11-15 16-20 21-30 31+
US Men: Non-white
1980 2000
-.35 -.25 -.15 -.050 .05.1
0-5 6-10 11-15 16-20 21-30 31+
GB Women: Non-white
1980 2000
-.35 -.25 -.15 -.050 .05.1
0-5 6-10 11-15 16-20 21-30 31+
US Women: Non-white
1980 2000
Changes in Relative Economic Performance of Immigrants 673
© Blackwell Publishing Ltd/London School of Economics 2007.
differential return migration rates by ethnicity, differences in retirement
rates, or, for women, by differential fertility rates, seems to be important
avenues for future research.18
In contrast to Britain, Table 4 and Figure 1 indicate that male immigrants
to the USA have an employment advantage that rises with time spent in the
country. Over the sample period, this relative employment advantage has
shrunk, but not disappeared, for each of the years-in-the-country bands used.
Among female immigrants to the USA, there is an employment gap relative
to native-born on entry that has grown over the sample period (row 1,
column 4 of Figure 1). This gap now appears to dissipate more slowly with
time spent in the country than in the past. Figure 1 disaggregates these trends
by ethnicity19 and shows that this deterioration in the employment assimilation
profile of female immigrants to the USA is primarily attributed to a
decline in the relative employment rates of non-white immigrants (panel 4,
rows 2 and 3). The employment advantage for male immigrants to the USA
at any years-in-the-country segment stems from the employment outcomes
for non-white male immigrants (panel 2, row 3). White male immigrants have
an employment penalty that does, however, appear to fall with time spent in
the USA and over the sample period at any given duration (panel 2, row 2).
Nevertheless, on this basis, white male immigrants to the USA never achieve
employment convergence, but non-white males achieve instant parity in#p#分頁標題#e#
employment rates and then improve relative to their starting position. This
differential pattern by ethnicity does not fit easily with the hypothesis of
Antecol et al. (2006) that a flexible economy, such as the USA, should
provide employment opportunities for immigrants. The fact that employment
of immigrants varies by ethnicity and gender suggests that further
thinking is needed here.
Table 5 and Figure 2 give the results of a similar analysis for relative wage
levels conditional on years in the host country. While Table 3 shows that
average wages of immigrants relative to native-born do not differ much
between the two countries, Table 5 shows that the wage gaps conditional on
years-since-arrival are larger in the USA than in Britain, particularly for men.
Indeed for Britain, once we net out the effects of age, education, region and
area of origin, there is no evidence of ‘convergence’ for either men or women,
nor does there appear to have been much change over time. The estimated
profiles for non-white immigrants are generally more negative than those for
white immigrants. However, the standard errors around these estimates
(available from the authors on request), are quite large, making it harder to
determine the overall shape of the assimilation profile for this group. While
the smaller sample of the British data may affect the precision of our estimates,
this general lack of a ‘typical’ assimilation profile is consistent with
some results contained in an earlier UK study by Bell (1997). Age, region and
education controls take away most of the ‘typical’ positive association
between wages and years-since-arrival observed in the raw British data,
leaving an immigrant wage penalty based around the central estimate
observed in Table 2, that does not appear to vary much with time spent in the
674 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
host country. This finding again sits somewhat at odds with the conclusions
of Antecol et al. (2006) that more flexible labour markets should experience
similar immigrant wage-assimilation profiles to that of the USA. On this
reasoning, Britain is more flexible than the USA, such that wage opportunities
for immigrants in Britain are equalized immediately. Alternatively, it
may be that the institutions of the British labour market act to equalize wages
for immigrants much more rapidly than in the USA. More research is needed
to untangle these competing explanations.
For the USA, neither men nor women reach a point of convergence with
native wages. The entry wage penalty for male immigrants appears to have
fallen over time (row 1, column 2 in Figure 2), but the entry wage gap has
risen for female immigrants over time (row 1, column 4). At intermediate#p#分頁標題#e#
durations, the wage gap appears to have risen over time for both men and
women. At longer durations in the host country, the wage gap has narrowed.
For both men and women, the pattern of differentials across time-in-thecountry
suggests that recent immigrants to the USA spent more time in 2000
at wages farther below their native-born counterparts than they had in 1980.
TABLE 5
Effect of Time in Host Country on Log Hourly Wage of Immigrants with Respect to
Native-Born Workers
Great Britain USA
1980 1990 2000 1980 1990 2000
Men
0–5 years 0.011 –0.003 0.042 –0.289 –0.321 –0.250
(0.042) (0.065) (0.037) (0.013)* (0.007)* (0.006)*
6–10 years –0.087 –0.066 –0.029 –0.179 –0.249 –0.226
(0.030)* (0.051) (0.040) (0.014)* (0.007)* (0.007)*
11–15 years 0.002 –0.076 –0.005 –0.135 –0.163 –0.187
(0.027) (0.049) (0.044) (0.014)* (0.008)* (0.007)*
16–20 years –0.003 –0.030 0.005 –0.074 –0.130 –0.157
(0.024) (0.042) (0.049) (0.015)* (0.009)* (0.007)*
21–30 years –0.029 0.001 0.024 –0.084 –0.080 –0.103
(0.020) (0.027) (0.034) (0.014)* (0.008)* (0.007)*
31+ years 0.069 0.077 0.038 –0.106 –0.075 –0.066
(0.021)* (0.026)* (0.028) (0.018)* (0.009)* (0.008)*
Women
0–5 years –0.044 0.117 –0.066 –0.193 –0.279 –0.232
(0.053) (0.054)* (0.037) (0.018)* (0.010)* (0.009)*
6–10 years –0.038 0.032 0.003 –0.095 –0.181 –0.188
(0.032) (0.049) (0.038) (0.017)* (0.008)* (0.008)*
11–15 years –0.001 –0.015 –0.007 –0.077 –0.081 –0.142
(0.030) (0.051) (0.041) (0.018) (0.010)* (0.008)*
16–20 years 0.013 0.025 –0.041 –0.057 –0.066 –0.116
(0.026) (0.033) (0.048) (0.018)* (0.009)* (0.008)*
21–30 years 0.028 0.046 –0.023 –0.060 –0.053 –0.066
(0.019) (0.024) (0.030) (0.018)* (0.009)* (0.007)*
31+ years 0.009 0.063 0.069 –0.071 –0.036 –0.042
(0.025) (0.024)* (0.027)* (0.019)* (0.011)* (0.008)*
Notes: See Table 3. Coefficients measure effect relative to native-born workers.
Changes in Relative Economic Performance of Immigrants 675
© Blackwell Publishing Ltd/London School of Economics 2007.
FIGURE 2
Log Hourly Wage Differential Relative to Native-Born Workers by Years in Country, 1980–2000.
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
GB Men
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
US Men
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
GB Women
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+#p#分頁標題#e#
US Women
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
GB Men: White
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
US Men: White
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
GB Women: White
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
US Women: White
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
GB Men: Non-white
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
US Men: Non-white
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
GB Women: Non-white
1980 2000
-.3 -.2 -.1 0.05 .15
0-5 6-10 11-15 16-20 21-30 31+
US Women: Non-white
1980 2000
676 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
Most of any deterioration in wage convergence in the USA at a given duration
over the sample period is among non-white immigrants. The relative
wage gap for white male immigrants to the USA has improved somewhat. By
2000, convergence in wages for this group relative to native-born appears to
be a possibility after around 21–30 years spent in the USA. Non-white male
immigrants never converge (row 3, column 2). Non-white female immigrants
converged to US-born wages after 16 to 20 years in the USA in 1980, but not
until after 31 years in the USA in 2000 (row 3, column 4).
Changing area of origin mix is unlikely to underlie any deterioration in the
relative employment and wages of immigrant women to the USA observed
above, or explain why wage prospects for male immigrants to Britain seem to
have improved but worsened for female immigrants, because the area of
origin effects are included among the set of controls. Nor can the differences
in average educational attainment of immigrants in the two countries explain
these differential assimilation profiles since these factors are also netted out
from the estimates. One explanation is that there are other changes in the
composition of these immigrant cohorts that have gone on over time, an issue
we address next.20
10-Year Changes by Cohort
Tables 6 and 7 and Figures 3 and 4 present the results of the third test of
immigrant mobility. The tables give the difference-in-difference estimates of
employment (Table 6) and real wage growth (Table 7) for four arrival
cohorts, matched across the data net of cohort fixed effects and the same
controls used above. The first row in the tables shows the 10-year growth
relative to native-born workers for the group of immigrants that had an
average of 2.5 years in the country in the initial sample year and an average
of 12.5 years in the second sample year. The subsequent rows give relative#p#分頁標題#e#
growth rates for different year-of-arrival cohorts.
For native-born British men (Table 6, row 5), the employment rate fell by
around 8 percentage points between 1980 and 1990 and by a further 1.6 percentage
points between 1990 and 2000. In contrast, the employment rate of
new arrivals (row 1) grew by 0.8 percentage points between 1980 and 1990
(–0.079 + 0.087), and by 4.7 percentage points between 1990 and 2000
(–0.016 + 0.053). Therefore, while the relative growth rate for new arrivals
was higher in the first decade of the sample period, absolute growth was
higher in the second decade of the sample. For the British sample of women,
the employment rate for native-born women grew faster in the first half of the
sample period. However, the estimates, net of controls, are not well determined
for many of the British cohorts so some caution is needed here.
The estimates for the USA are better determined. While the employment
rate of native-born US men fell by 2.7 points between 1980 and 1990, the
employment rate of those who were new arrivals in 1980 grew by 5.8 percentage
points (-0.027 + 0.085). Similarly, the employment rate of female
new arrivals rose by 15 percentage points (0.065 + 0.085). By the second
Changes in Relative Economic Performance of Immigrants 677
© Blackwell Publishing Ltd/London School of Economics 2007.
TABLE 6
Regression-Controlled 10-Year Employment Probability Change by Year-of-Arrival
Men Women
1980–1990 1990–2000 Net change in
convergence
rate towards
native-born
1980–1990 1990–2000 Net change in
convergence
rate towards
native-born
Great Britain
2.5–12.5 years 0.087 (0.011)* 0.053 (0.026)* -ve 0.038 (0.034) -0.026 (0.035) -ve
7.5–17.5 years 0.046 (0.017)* 0.045 (0.030) -ve 0.018 (0.031) 0.106 (0.033)* +ve
12.5–22.5 years 0.039 (0.016)* 0.027 (0.028) -ve 0.052 (0.028) 0.054 (0.032) -ve
17.5–27.5 years -0.007 (0.019) 0.026 (0.028) +ve 0.042 (0.029) 0.019 (0.034) 0
10-year change -0.079 (0.002)* -0.016 (0.003)* — 0.025 (0.003)* 0.003 (0.004) —
UK-born — —
USA
2.5–12.5 years 0.085 (0.004)* -0.014 (0.005) -ve 0.085 (0.007)* 0.065 (0.005)* -ve
7.5–17.5 years -0.005 (0.006) -0.082 (0.006)* -ve -0.011 (0.008) -0.007 (0.006) -ve
12.5–22.5 years -0.022 (0.008)* -0.060 (0.007)* -ve -0.025 (0.009)* -0.016 (0.007)* -ve
17.5–27.5 years 0.003 (0.009) -0.060 (0.007)* -ve 0.009 (0.009) -0.023 (0.007)* -ve
10-year change -0.027 (0.001)* -0.047 (0.001)* — 0.065 (0.001)* -0.001 (0.001) —
US-born — —
Notes: See Table 2. Years-in-country coefficients show 10-year change in employment rate relative to 10-year change of native-born population. Net change
columns show change in the combined effect of 10-year growth rate and cross-section estimates of time in host country for each cohort in Table 5 between 1990#p#分頁標題#e#
and 2000.
678 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
TABLE 7
Regression-Controlled 10-Year Log Real Hourly Wage Growth by Year-of-Arrival
Men Women
1980–1990 1990–2000 Net change in
convergence
rate towards
native-born
1980–1990 1990–2000 Net change in
convergence
rate towards
native-born
Great Britain
2.5–12.5 years -0.089 (0.054) -0.008 (0.064) +ve 0.035 (0.054) -0.164 (0.066)* -ve
7.5–17.5 years -0.006 (0.041) 0.004 (0.076) +ve 0.022 (0.036) 0.013 (0.061) +ve
12.5–22.5 years -0.059 (0.031) 0.132 (0.055)* +ve 0.037 (0.037) -0.016 (0.052) -ve
17.5–27.5 years -0.026 (0.031) 0.055 (0.046) +ve 0.074 (0.030)* -0.012 (0.036) -ve
10-year change 0.139 (0.003)* -0.042 (0.004)* — 0.181 (0.003)* 0.066 (0.004)* —
UK-born — —
USA
2.5–12.5 years 0.175 (0.011)* 0.141 (0.009)* -ve 0.158 (0.013)* 0.140 (0.010)* -ve
7.5–17.5 years 0.097 (0.012)* 0.089 (0.009)* -ve 0.076 (0.012)* 0.063 (0.010)* -ve
12.5–22.5 years 0.112 (0.013)* 0.043 (0.010)* -ve 0.070 (0.013)* 0.007 (0.011) -ve
17.5–27.5 years 0.085 (0.015)* 0.015 (0.011) -ve 0.044 (0.015)* 0.005 (0.011) -ve
10-year change -0.024 (0.001)* 0.043 (0.001)* — 0.058 (0.001)* 0.108 (0.001)* —
US-born — —
Notes: See Table 6.
Changes in Relative Economic Performance of Immigrants 679
© Blackwell Publishing Ltd/London School of Economics 2007.
decade of the sample, employment growth was similar to that of native-born
workers for new male arrivals, but had declined, relatively and absolutely, for
other male arrival cohorts. Among women, strong relative and absolute
growth was maintained for new arrivals, but, as with the earlier sample
period, not for other arrival cohorts. On this basis, the relative employment
advantages observed by some immigrants in the 1980s were not as apparent
in the 1990s. Further analysis of the data, available on request, suggests that
relative employment growth among non-white immigrants to the USA was
worse than among white immigrants in the second period.
As it is possible that changes in the 10-year relative growth rates could be
offset by changes in the cohort year of arrival effects,21 columns 3 and 6 in
Table 6 show the direction of the combined effect of the changes in these
10-year cohort growth rates and the changes in the years-since-arrival effects
outlined in Table 4. The results for the USA indicate that net effect was
negative so that employment rates of immigrants diverged rather than converged
in the latter half of the period compared to the first half of the sample
period. This is attributed to a combination of declining year of arrival effects#p#分頁標題#e#
between 1980 and 1990 and a decline in the 10-year growth rate. The results
for Britain, while less well-determined, also hint at divergence, particularly
for men.
Table 7 and Figure 4 show that relative wage growth of immigrants to
Britain was not generally significantly different from wage growth among
FIGURE 3
10-Year Employment Growth Rate of Immigrants by Initial Years in Country Relative to
Native-Born, 1980–2000.
-.15 -.1 -.05 0 .05 .1 .15
2.5 7.5 12.5 17.5
GB Men
1980/90 1990/2000
-.15 -.1 -.05 0 .05 .1 .15
2.5 7.5 12.5 17.5
US Men
1980/90 1990/2000
-.15 -.1 -.05 0 .05 .1 .15
2.5 7.5 12.5 17.5
GB Women
1980/90 1990/2000
-.15 -.1 -.05 0 .05 .1 .15
2.5 7.5 12.5 17.5
US Women
1980/90 1990/2000
680 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
native-born workers in either period. In the USA, however, wages of immigrants
in all four cohorts grew faster than wages of native-born, but there
was a marked slow-down in this relative advantage in the last decade. As
the average immigrant tends to start work in the USA at substantially
lower wages than US-born workers do and because immigrants tend to
converge eventually on the earnings of US-born workers with comparable
education and work experience, it is not surprising that the wages of immigrants
tend to grow more rapidly than those of US-born workers over
time.22 The lack of a clear pattern of wage growth by arrival cohort of the
British wage data is also consistent with the lack of an assimilation profile
net of controls. Figure 4 suggests that relative wage growth for a given
arrival cohort was stronger for non-white immigrants in the second half of
the sample period.
Again, the net effect of changes in these 10-year growth rates and year-ofarrival
effects points to deterioration in relative wage opportunities over time
in the USA. For Britain, rising 10-year cohort effects have tended to offset
any declines in the year-of-arrival effects so that the net effect for men hints
at greater convergence towards native-born wages over time although the
effects are not well determined. For female immigrants to Britain, the results
are suggestive of divergence.
FIGURE 4
10-Year Wage Growth Rate of Immigrants by Initial Years in Country Relative to
Native-Born, 1980–2000.
-.15 -.1 -.05 0 .05 .1 .15
2.5 7.5 12.5 17.5
GB Men
1980/90 1990/2000
-.15 -.1 -.05 0 .05 .1 .15
2.5 7.5 12.5 17.5
US Men
1980/90 1990/2000
-.15 -.1 -.05 0 .05 .1 .15
2.5 7.5 12.5 17.5
GB Women
1980/90 1990/2000
-.15 -.1 -.05 0 .05 .1 .15
2.5 7.5 12.5 17.5
US Women
1980/90 1990/2000
Changes in Relative Economic Performance of Immigrants 681#p#分頁標題#e#
© Blackwell Publishing Ltd/London School of Economics 2007.
6. Conclusions
This article uses repeated cross-section data from 1980, 1990 and 2000 to
measure changes in the relative labour market performance of immigrants to
Britain and the USA against a background of rising immigration in both
countries, shifting demand and changes to labour market institutions. While
the two countries are often grouped together, perhaps too readily, as
examples of economies with minimal regulations and limited institutions, so
the experience for immigrants in the two countries might be expected to be
similar. The reality is somewhat different. We find that the average relative
employment prospects of immigrants are generally better in the USA than in
Britain, particularly for non-white immigrants, while the average relative
wage prospects for immigrants are generally better in Britain, particularly for
men. Over time, relative wage and employment prospects for immigrants to
the USA appear to have deteriorated, particularly among women, in a way
that is not as apparent in Britain. As these trends are net of standard human
capital and area of origin controls, this suggests the likely importance of
broader changes in the wage structure and related labour-market institutions,
although more work also needs to be done on the effects of students, selective
return migration and unobserved heterogeneity on these results. However, it
is unlikely that the idea that different degrees of labour market flexibility
across countries lead to specific patterns of immigrant performance is sufficient
to explain the variation across countries and within-countries over time
that we observe here.
With immigration levels continuing to rise in both countries, our findings
suggest several important areas for future research. First, most comparative
analyses of labour market performance and institutions tend to emphasize
the similarities between both institutions and outcomes in the USA and
Britain. Our findings suggest important differences in the pattern of immigrant
economic assimilation both in 2000 and over time between 1980 and
2000. Further research should identify what aspects of the immigrant experience
in the two countries might account for these differences. Second, some
of the within-country differences in the immigrant experience across gender
are striking. In the USA, male immigrants are more likely to work than their
native-born counterparts, while female immigrants are less likely to do so. In
Britain, differences across gender in employment gaps are much smaller than
in the USA. Immigrant experience in both countries differs sharply between
white and non-white immigrants, even after controlling for standard measures
of human capital, including area of origin. Exploring these and other#p#分頁標題#e#
issues are beyond the scope of this overview, but further analysis of these
issues promises to shed light not only on the experience of immigrants in the
two countries, but also on broader forces that have worked to change the
wage distributions and employment opportunities so dramatically in both
countries over the last 30 years.
Final version accepted on 15 June 2007.
682 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
Notes
1. Sources: US Census and for Britain the Labour Force Survey. The nativeborn
working age populations grew by 20 million and 1.8 million, respectively,
over the same period as a result of the baby boom generation reaching
maturity.
2. For other analyses of the impact of immigrants on the US wage structure, see,
among others: Antecol et al. (2006), Borjas (2000), DeFreitas (1998), Duleep
et al. (1997), Funkhouser (2000), LaLonde and Topel (1991, 1992) and Schoeni
et al. (1996). See Dustmann et al. (2005) for the UK.
3. Although this right of entry does not entitle right to work for citizens of the latest
EU member countries, Bulgaria and Romania.
4. If true, this would amount to an additional 10 per cent of the immigrant stock.
5. The former dataset records years of arrival in five-year bands, the latter by
individual year. To facilitate comparison, we group the British data into similar
five-year bands as in the US data.
6. DeCoulon and Wadsworth (2006) document many of the institutional barriers to
practise certain occupations that exist for immigrants to the USA and the UK.
7. There may, of course, be a positive wage premium if immigrants have skills that
are lacking in the native-born population.
8. With the (debatable) assumption that aggregate wage growth is identical for
natives and immigrants, then it is possible to estimate cohort and years-sincemigration
effects separately. See Bell (1997) for an example of this.
9. While it assumes homogeneity of parameter effects in the control variables
between immigrant and native-born, pooling is often practical given the relatively
small sample sizes for some of the immigrant cohorts in the British data, which do
not allow us to estimate separate control effects with precision.
10. Unless there is any selective return migration, an issue we are unable to address
because of data limitations.
11. The US hourly wage is derived by dividing annual gross pay by 52 (weeks) by
usual weekly hours, while the British hourly wage is derived by dividing gross
weekly wage by usual weekly hours. Differential non-response between immigrant
and native-born workers is a potential issue for the analysis of wages.
However, in our data, the discrepancies are not that large. Eighty-eight per cent
of UK native-born employees gave valid hourly wage data compared to#p#分頁標題#e#
84 per cent of immigrants. For the (2001) US census, the wage response rates are
91 per cent native-born and 89 per cent immigrants.
12. This may, of course, include children born to British/US citizens living abroad,
although the data are unable to distinguish these individuals. There is no way of
determining how many of the immigrants in the surveys are undocumented.
13. As such, the British primary category includes all qualifications typically obtained
before the age of 18, the secondary category is all those obtained, typically, at 18
and the some college category contains all higher qualifications above this latter
threshold but below degree standard.
14. It is difficult to estimate return rates because GHS does not contain population
weights that would allow the comparison of appropriate cell group numbers over
time. The US census does have population weights, but they appear to be inconsistent
across time, making inter-temporal comparisons difficult.
15. Table 1 also shows that the share of new entrants in the immigrant stock has risen
alongside the rise in the immigrant share over time in Britain but not in the USA.
Changes in Relative Economic Performance of Immigrants 683
© Blackwell Publishing Ltd/London School of Economics 2007.
In the absence of reporting error, this suggests that there may be falling return
migration rates in the USA or differential responses among ‘undocumented’
immigrants in the USA over time.
16. The differences without controls are generally larger but display similar trends
over time. Results are available from the authors upon request. Some of the
differences in employment patterns may be attributed to the relatively high share
of British immigrants in continuing education. If we exclude students from the
UK data, the employment gaps fall by around one-third. Unfortunately, we
cannot identify students in the US census data.
17. These are statistically significant effects. The table of results that underlie the
ethnic breakdowns in Figures 1 to 4 are available from the authors upon request.
18. Some of these negative employment effects are again probably caused by higher
enrolment rates of immigrants in further education. The datasets for both countries
are not rich enough to identify the early retired consistently over time.
19. Regression results available from the authors upon request.
20. Another is that the assimilation profile has changed for given characteristics of
immigrants.
21. For example, a rise in the 10-year relative growth rate from 5 to 10 per cent would
not indicate convergence towards native-born levels if the year of arrival effect fell
from –10 to –20 per cent over the same period.
22. The trends by ethnicity (not shown) are similar in the US data and are available
from the authors upon request. There are no significant differences when the#p#分頁標題#e#
British sample is split by ethnicity.
Acknowledgement
The authors would like to thank the editor and two anonymous referees for
many helpful comments.
References
Antecol, H., Cobb-Clark D. and Trejo, S. (2003). ‘Human capital and earnings of
female immigrants to Australia, Canada, and the United States’. In J. Reitz (ed.),
Host Societies and the Reception of Immigrants. San Diego, CA: Center for Comparative
Immigration Studies, University of California, pp. 327–59.
——, Kuhn, P. and Trejo, S. (2006). ‘Assimilation via prices or quantities? Labor
market institutions and immigrant earnings growth in Australia, Canada and the
United States’. Journal of Human Resources, 41 (4): 821–40.
Baker, D. (2006). The Conservative Nanny State. Washington, DC: Center for Economic
and Policy Research.
Bell, B. (1997). ‘The performance of immigrants in the United Kingdom: evidence
from the GHS’. Economic Journal, 107 (441): 333–45.
Borjas, G. (1985). ‘Assimilation, changes in cohort quality, and the earnings of
immigrants’. Journal of Labor Economics, 3 (4): 463–89.
—— (1987). ‘Self-selection and the earnings of immigrants’. American Economic
Review, 77 (4): 531–53.
—— (1994). ‘The economics of immigration’. Journal of Economic Literature, 32 (4):
1667–717.
684 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.
—— (2000). ‘The economic progress of immigrants’. In G. J. Borjas (ed.), Issues in the
Economics of Immigration. Chicago: University of Chicago Press, pp. 15–49.
—— (2003). ‘The labour demand curve is downward sloping: re-examining the
impact of immigration on the labour market’. Quarterly Journal of Economics, 118:
1335–74.
Butcher, K. and DiNardo, J. (2002). ‘The immigrant and native–born wage distributions:
evidence from the United States census’. Industrial and Labor Relations
Review, 56 (1): 97–121.
Card, D. (2001). ‘Immigrant inflows, native outflows, and the local labor market
impacts of higher immigration’. Journal of Labor Economics, 19 (1): 22–63.
—— (2005). ‘Is the new immigration really so bad?’. Economic Journal, 115: F300–
23.
Chiswick, B. (1978). ‘The effect of Americanization on the earnings of foreign born
men’. Journal of Political Economy, 86 (5): 897–921.
Clark, K. and Lindley, J. (2006). ‘Immigrant Labour Market Assimilation and
Arrival Effects’. IZA Discussion Paper No. 2228, IZA, Bonn.
De Coulon, A. and Wadsworth, J. (2006). ‘Do Relative Gains to Migration Vary? A
Comparison of the Labour Market Performance of Indians in the UK, the USA#p#分頁標題#e#
and India’. Centre for Economic Performance Working Paper No. 1432, London
School of Economics.
DeFreitas, G. (1998). ‘Immigration, inequality and policy alternatives’. In G. Epstein
and R. Pollin (eds.), Globalization and Progressive Economic Policy. New York:
Cambridge University Press, pp. 337–56.
Duleep, S., Orcutt, H. and Regets, M. (1997). ‘Immigrant entry earnings and human
capital growth: evidence from the 1960–1980 censuses’. Research in Labour Economics,
16: 297–317.
Dustmann, C. and Fabbri, F. (2005). ‘Immigrants in the British labour market’. Fiscal
Studies, 26 (4): 423–70.
——, —— and Preston, I. (2005). ‘The impact of immigration on the UK labour
market’. Economic Journal, 115 (November): F324–41.
Funkhouser, E. (2000). ‘Convergence in employment rates of immigrants’. In G.
Borjas (ed.), Issues in the Economics of Immigration. Chicago: University of
Chicago Press, pp. 143–84.
Heer, D. and Passel, J. (1987). ‘Comparison of two different methods for computing
the number of undocumented Mexican adults in Los Angeles County’. International
Migration Review, 21: 1446–73.
Kossoudji, S. (1989). ‘Immigrant worker assimilation: is it a labor market phenomenon?’.
Journal of Human Resources, 24 (3): 494–527.
LaLonde, R. and Topel, R. (1991). ‘Immigrants in the American labour market:
quality, assimilation, and distributional effects’. American Economic Review Papers
and Proceedings, 81 (2): 297–302.
—— and —— (1992). ‘The assimilation of immigrants in the U.S. labour market’. In
G. Borjas and R. Freeman (eds.), Immigration and the Work Force: Economic
Consequences for the United States and Source Areas. Chicago: University of
Chicago Press, pp. 67–92.
Manacorda, M.,Manning, A. and Wadsworth, J. (2005). ‘The Impact of Immigration
on the Structure of Wages in Britain’. Centre for Economic Performance Discussion
Paper 754, London School of Economics.
Schoeni, R., McCarthy, K. and Vernez, G. (1996). The Mixed Economic Progress of
Immigrants. Santa Monica, CA: The RAND Corporation.
Changes in Relative Economic Performance of Immigrants 685
© Blackwell Publishing Ltd/London School of Economics 2007.
Wasem, R. (2004a). ‘US Immigration Policy on Permanent Admissions’. Congressional
Research Service Report RL32235. Available at http://fpc.state.gov/
documents/organization/31352.pdf
—— (2004b). ‘Unauthorized Aliens in the United States: Estimates Since 1986’.
英國dissertation網Congressional Research Service Report RL21938. Available at http://fpc.state.gov/#p#分頁標題#e#
documents/organization/31352.pdf
Woodbridge, J. (2005). ‘Sizing the Unauthorised (Illegal) Migrant Population in the
United Kingdom in 2001’. Home Office Online Report 29/05, London, Home
Office. Available at http://www.homeoffice.gov.uk/rds/pdfs05/rdsolr2905.pdf
686 British Journal of Industrial Relations
© Blackwell Publishing Ltd/London School of Economics 2007.