Africans in the American Labor Market

Abstract

The number of migrants to the United States from Africa has grown exponentially since the 1930s. For the first time in America’s history, migrants born in Africa are growing at a faster rate than migrants from any other continent. The composition of African-origin migrants has also changed dramatically: in the mid-twentieth century, the majority were white and came from only three countries; but today, about one-fifth are white, and African-origin migrants hail from across the entire continent. Little is known about the implications of these changes for their labor market outcomes in the United States. Using the 2000–2011 waves of the American Community Survey, we present a picture of enormous heterogeneity in labor market participation, sectoral choice, and hourly earnings of male and female migrants by country of birth, race, age at arrival in the United States, and human capital. For example, controlling a rich set of human capital and demographic characteristics, some migrants—such as those from South Africa/Zimbabwe and Cape Verde, who typically enter on employment visas—earn substantial premiums relative to other African-origin migrants. These premiums are especially large among males who arrived after age 18. In contrast, other migrants—such as those from Sudan/Somalia, who arrived more recently, mostly as refugees—earn substantially less than migrants from other African countries. Understanding the mechanisms generating the heterogeneity in these outcomes—including levels of socioeconomic development, language, culture, and quality of education in countries of origin, as well as selectivity of those who migrate—remain important unresolved research questions.

Introduction

The number of migrants to the United States from Africa has grown exponentially since the 1930s. For the first time in America’s history, among all foreign-born migrants, those born in Africa are growing at the fastest rate. According to the 2011 American Community Survey (ACS), 1.7 million foreign-born migrants from Africa live in the United States, and they account for about 4 % of the foreign-born population.

Figure 1 displays the number of African-origin migrants who have obtained legal permanent status in the United States by decade since 1900 in thousands of people (upper panel) and in a logarithmic scale (lower panel). After declines before and during the Great Depression, the end of World War II marks the onset of the exponential growth that has characterized African-origin migration to the United States over the last 70 years. Since the 1950s, the number of foreign-born who have become legal permanent residents has quadrupled, but the number from Africa has increased nearly 60-fold—a rate of growth more than twice that for migrants from Asia, the next-fastest growing source of new Americans. According to World Bank estimates, the United States ranked third in 2010, behind France and Saudi Arabia, among destination countries of African-born migrants leaving the continent (Ratha et. al. 2011).

Fig. 1.

African-origin migrants who obtained lawful permanent resident status by decade, 1900–2009. Source: U.S. Department of Homeland Security, Yearbook of Immigration Statistics (2011)

This article explores labor market outcomes of African-origin migrants by country of origin, race, human capital, age, and year of arrival, drawing on the 2000 through 2011 waves of the ACS. By comparing African migrants with each other, we highlight the heterogeneity of this quickly growing group of new Americans—differences that are buried in comparisons of African-born with native-born Americans.

Over the past six decades, the composition of migrants from Africa has changed dramatically. In the 1950s, Morocco, South Africa, and Egypt accounted for about 60 % of migrants from Africa (see Fig. 2). The vast majority were white (Kollehlon and Eule 2003). Subsequently, migration of blacks has increased even more rapidly than migration of whites. According to the 2011 ACS, close to three-quarters of African-born migrants in the United States self-identify as black (or black and another race); only about one-fifth self-identify as white; about 60 % of the remaining 5% report themselves as Asian-origin; and the rest report themselves as another race(s). Initially, black migrants came mainly from English-speaking countries—particularly Ghana, Kenya, Liberia, and Nigeria—as well as from Ethiopia. More recently, the number of sending countries has increased, and today no single country dominates (Capps et al. 2012).

Fig. 2.

Percentage of all African-origin migrants who came from Egypt, South Africa, and Morocco and obtained permanent legal status by decade (1950–2009). Sources: U.S. Department of Justice, Statistical Yearbook of Immigration and Naturalization Service (1982–2001); U.S. Department of Homeland Security, Yearbook of Immigration Statistics (2002–2011)

Morocco, Egypt, and South Africa continue to be the primary source of white migrants from Africa. Today, whites from these countries account for 20 % of all African-origin migrants. Their number has risen 20-fold, from approximately 7,500 during the 1950s to more than 150,000 during 2000–2010. In the same period, the number of migrants from other African countries increased over 100-fold, more than doubling in just the last decade.

Changes in immigration policy, such as the 1980 Refugee Act, the 1990 Immigration Act, and the Diversity Visa Program have fueled some of the increase (Jasso 2011; Kollehlon and Eule 2003; Logan and Thomas 2012). Migrants enter the United States under an array of legal statutes. Figure 3 displays the number of African-origin migrants who obtained legal permanent residence in the United States between 1982 and 2011, distinguishing three primary visa types: family-based, employment-based (which includes diversity visas), and refugee-based visas.¹ In the 1980s, three-quarters of legal permanent residents from Africa entered the United States on family-based visas. The number of those visas grew fourfold over the next 30 years, but increases in employment-based and refugee-based visas were even larger. By 2010, employment-based visas accounted for approximately 25 % of all visas issued to African legal permanent residents, of which about three-quarters are diversity visas.

Fig. 3.

Visa types of African-origin migrants at time of entry as legal permanent residents in United States, 1982–2011. Family visas include all family-related visas (spouse, parent, sibling or child). Employment visas include all employment-based visas and diversity visas. Sources: U.S. Department of Justice, Statistical Yearbook of Immigration and Naturalization Service (1982–2001); and U.S. Department of Homeland Security, Yearbook of Immigration Statistics (2002–2011)

Temporal variation in visa composition across African countries is reflected in Fig. 4, which displays the number of visa types of legal permanent residents by year for 12 country groups. Growth in the number of family-based visas, particularly since the beginning of this century, is dominated by English-speaking countries (Ghana and Nigeria in West Africa, and the East African countries of Kenya, Tanzania, and Uganda) along with two non-English-speaking countries (Ethiopia/Eritrea). In part, these patterns reflect accumulating numbers of Africans from these countries who have sponsored family members to join them in the United States. In contrast, the number of family visas has grown relatively modestly in Egypt and South Africa, which are long-standing sources of migrants to the United States. Cape Verde has also sent migrants to the United States over many decades. Today, most migrants from Cape Verde enter the United States on family-based visas, a striking contrast with every other African country.

Fig. 4.

Visa types by country of origin at time of entry as legal permanent residents in United States, 1982–2011. Family visas include all family-related visas (spouse, parent, sibling or child). Employment visas include all employment-based visas and diversity visas. Sources: U.S. Department of Justice, Statistical Yearbook of Immigration and Naturalization Service (1982–2001); and U.S. Department of Homeland Security, Yearbook of Immigration Statistics (2002–2011)

The number of employment-based visas quadrupled in the mid-1990s. This category includes diversity visas that go to winners of lotteries conducted in countries underrepresented in recent migrant flows to the United States (all African countries are eligible). Diversity visas have become an important route for skilled migrants from a diverse set of African countries (Logan and Thomas 2012; Thomas 2011a). As shown in Fig. 4, employment visas account for relatively large numbers of migrants from some countries (South Africa/Zimbabwe and Egypt) but for relatively small numbers of new migrants from others (Liberia and Sudan/Somalia).

Refugees make up the third prominent visa type. The number of Africans admitted as refugees increased in the mid-1990s and again in the mid-2000s. The numbers fluctuate across years, but as of this writing, about one-quarter of all African-origin legal permanent residents entered as refugees. The vast majority come from a small number of countries. One-third of refugees over the last three decades were born in Sudan/Somalia, one-quarter were born in Ethiopia/Eritrea, and most of the rest were born in Liberia and East Africa. Refugees are not typically drawn from among the most skilled in their home countries, and they often arrive with limited employment prospects.

These figures highlight two important facts. First, the growth in African-origin migration to the United States is driven by migrants who hail from a vastly more diverse array of countries than was the case 50 years ago. Second, substantial temporal and spatial heterogeneity exists in the types of visas under which the migrants have entered the United States, which has implications for their human capital, selectivity relative to those left behind, and resources available to them in the United States. These facts, combined with immense differences in human capital and financial resources between and within African countries, likely affect the patterns of labor market outcomes of African-born migrants living in the United States.

Background

Several theoretical perspectives offer insights into the factors that may influence migrants’ labor market outcomes. Human capital theory emphasizes the roles of education, training, and experience (Card 1999; Hout 2012; Mincer 1974), all of which are likely to be relevant for African migrants to the United States. Variation in the levels of schooling among African migrants is enormous. Moreover, the transferability of schooling to the U.S. labor market likely varies substantially as well, partly as a function of differences in country of birth but perhaps also differences in race, ethnicity, and native language. Human capital theory suggests that migrants whose schooling is less transferable and thus not rewarded by employers are more likely to work in the self-employed sector, at least when they first arrive. These migrants might switch to the market sector after obtaining work experience or education in the United States.

Age at migration and recency of the move are also likely to be associated with variation in labor market outcomes. Individuals who moved to the United States as school-age children will have completed at least some school in the United States and so will not suffer the credential penalty experienced by migrants who completed school in their origin country. More generally, their early life experiences differ substantially from those of migrants arriving at older ages. Experiences with schooling and early exposure to the U.S. social, political, and cultural environment may all be related to labor market outcomes (Thomas 2009, 2011b, 2012). On the other hand, the decision to move was typically made by these migrants’ parents. Depending on the degree of intergenerational transmission of skills and attributes, the children of migrants are probably less highly selected on characteristics associated with labor market success in the United States. For example, migrants are likely to be positively selected on entrepreneurship, but whether these skills are transmitted across generations of migrants has not been established (Dunn and Holtz-Eakin 2000; Fairlie 2013.)

African-origin migrants who moved to the United States many years ago are likely to be an especially select group. The costs of migration were relatively high (Jasso 2011), and networks of co-nationals were more sparse. Those still in the United States have stayed for the long haul and are likely to be relatively more successful than recent arrivals. Moreover, their U.S. experience may have payoffs for navigating demands of the U.S. labor market. More recent migrants have had less time to assimilate but may benefit from established social networks of co-nationals for job referrals and integrating into the U.S. environment more generally. Recent migrants are also increasingly diverse and have come to the United States for a variety of reasons—some to join family members, others because of their skills or luck in the diversity visa lottery, and others as refugees.

Several studies have investigated wage differences among white and black migrants from Africa to the United States. Using the 1990 census, Dodoo and Takyi (2002) showed that hourly wages of 25- to 64-year-old African-origin white male migrants are 32 % higher than those of comparable black migrants. The difference fell to 19 % after they controlled for human capital and years in the United States. Differences by race remained significant with additional controls for the migrant’s country of birth and for the subset with U.S. degrees.

Similarly, Kollehlon and Eule (2003), also using 1990 census data, documented significantly higher hourly earnings at ages 25–64 among African-born white men and women than among their black counterparts, with and without controls for educational attainment, English ability, years of work experience, marital status, year of immigration, region of residence, and children ever born (women only). The white wage premium was higher for men (34 %) than for women (13 %).² Kollehlon and Eule’s findings also pointed to country-of-birth heterogeneity in hourly wages among Africans living in the United States, especially among men.

Borch and Corra (2010) compared earnings of black and white men and women at ages 25–64 who were born in Africa and living in the United States in 1980, 1990, and 2000, based on U.S. census data. They reported a significant white wage premium that was larger for men than for women when adjusted for covariates that included human capital controls, year, whether the person migrated before age 16, region of residence, and marital status. For men, the black-white difference appeared to have grown over time.

Others have compared wages of foreign-born to U.S.-born blacks to gain insights into migrant adaptation and the role of racial discrimination (Borch and Corra 2010; Borjas 1999, Butcher 1994; Corra and Kimuna 2009; Dodoo 1997; Kollehlon and Eule 2003; Stewart and Dixon 2010). Wage differences for black male and female migrants from Africa relative to black migrants from the Caribbean and U.S.-born blacks have varied over time. For example, the pattern of lower wages of African-born male migrants relative to U.S.-born blacks and other black foreign-born males observed in the 1980 census data (Butcher 1994) were absent in analyses of the 1990 U.S. Census (Dodoo 1997; Kollehlon and Eule 2003; Kposowa 2002). Corra and Kimuna (2009), who studied hourly earnings of black female migrants residing in the United States in 1990 and 2000, found that the wages of female migrants from Africa deteriorated over time relative to wages of U.S.-born black women. These results are likely related to the changing composition of the African-born black population in the United States.

This research makes three contributions to the literature. First, we investigate variation in wages, distinguishing country of birth, race/ethnicity, gender, and timing of arrival in the United States, and interpret the results in the context of observed and unobserved selection of migrants. Second, because returns to schooling in the U.S. labor market vary by country of origin and race/ethnicity, we examine the role of self-employment as an alternative to wage income. Earnings from self-employment are likely to be less contaminated by the roles of signaling and credentialism relative to earnings in the market wage sector. Third, we use more recent data than prior studies. By drawing on annual waves of the ACS from 2000–2011, we more fully capture the tremendous increase in diversity among migrants from Africa since the late 1990s while also providing evidence on labor market outcomes of migrants over the longer-term.

Data and Methods

Twelve waves (2000–2011) of the public-use microdata samples from the ACS, collected and disseminated by the U.S. Census Bureau (2011) yield a sample of 38,546 male and 33,291 female African-born migrants to the United States. Pooling multiple waves provided a sufficiently large sample to examine migrants separately by country of birth and key human capital characteristics.³ The analytical sample includes noninstitutionalized males and females aged 25–64 for whom birthplace is recorded as a country in Africa. We include those in school, many of whom were also working; excluding individuals who were working while in school does not affect our conclusions. Among males, 91 % (34,904) had worked and had positive earnings in the year prior to the survey, as had 75 % of females (24,965).

Dependent Variables

Our main dependent variable in the regression analyses is the log of real hourly earnings, which we calculate by dividing total earnings (the sum of wages, salaries, and nonfarm self-employment income) the year before the survey by the number of hours worked (the usual number of hours worked per week multiplied by the number of weeks worked in the previous year). All hourly earnings are converted to 2010 U.S. dollars using the chained All Items Consumer Price Index (CPI) for all urban areas in the United States. Among men, we also examine factors that predict whether the individual was self-employed (yes, no) as well as whether earnings from self-employment and from wages varied with migrant characteristics along the lines predicted by theory.

Explanatory Variables

To facilitate interpretation of our results, we examine individual countries of birth or combine them into groups based on geographic location, primary language, and history. These countries, which together account for 80 % of all African-born migrants to the United States are Algeria/Morocco, Cape Verde, East Africa (Kenya, Tanzania, Uganda, and other East African countries not separately identified), Egypt, Ethiopia/Eritrea, Ghana, Liberia, Nigeria, Senegal/Cameroon, South Africa/Zimbabwe, and Sudan/Somalia (which includes South Sudan). The remainder (20 %) includes migrants from countries in North, West, and Southern Africa not individually identified in the ACS.

We also control for self-identified race: black, white, Asian, and other. We code everyone who self-identified as black/African American or black/African American and another race as black; those who self-identified only as white as white; those who self-identified as Asian or Asian and a race other than black/African American as Asian; and those not classified as “other” race (Mutchler et al. 2007; Tucker et al. 2002). We experimented with alternative codings of multirace individuals and combining the residual other category with blacks or whites, but none of our substantive conclusions was affected. We also control for marital status (never married, currently married, separated/divorced/widowed); and, for women, whether the individual lived with children younger than 6 years.

Our human capital characteristics include educational attainment (less than college, some college, college degree, post-college degree); whether at least some education was obtained in the United States (no, yes), estimated from age at arrival in the United States and the number of years of school completed; and level of English ability (does not speak English or does not speak well, speaks well but not at home, speaks very well but not at home, speaks only English at home).

Other explanatory variables are U.S. citizenship (not a citizen, citizen by virtue of being born to American parents, naturalized), number of years in the United States based on the survey date and date of arrival (<6 years, 6–10 years, 11–15 years, 16–25 years, 25+ years), age at which the individual arrived in the United States (coded into five-year age groups starting at ages 0–4 and up to 50+), state of residence in the United States, and year of survey. We include state to control for variation in labor market conditions and wage rates. We would have liked to include visa type at entry among noncitizens, but this information is not available in the ACS.

Empirical Models

We begin by modeling real hourly earnings. We estimate a linear regression model of log of hourly earnings, stratified by gender. In Model 1, we control for country of birth, race, age, years in the United States, age at arrival, and survey year. In Model 2, we add a set of controls related to human capital: level of education; self-reported English ability; marital status; citizenship; and state of residence in the United States at the time of the survey. Because the distribution of migrants by country of birth varied considerably across self-identified race categories, we stratify the analysis by race (black and white/Asian). In addition, because the selection mechanisms for migration as a child versus an adult are likely to differ, we stratify by whether the individual arrived in the United States before age 18..

To investigate self-employment rather than wage work as a strategy for capturing returns to migration, we identify characteristics that predict participation in the self-employed sector relative to the market wage sector. We also examine hourly earnings among self-employed individuals in comparison with wage earners. These analyses focus on men only because few women are self-employed. Again, the key variable of interest is country of birth, and we control for the same factors described earlier. We model self-employment (yes, no) with a linear probability model and hourly earnings using ordinary least squares (OLS).

In all models, standard errors and test statistics are estimated taking into account clustering by country of birth and heteroskedascticity of arbitrary form. All estimates are weighted to account for sampling probabilities.

Results

Table 1 presents sample characteristics for males and females, both overall and then further stratified by race and age of arrival in the United States. Migrants from Nigeria, Egypt, and Ethiopia/Eritrea account for about 35 % of all male migrants from Africa. Among male migrants, 69 % self-identify as black; 25.5 %, as white; 3.4 %, as Asian; and 2.2 %, as other races. In the regression analyses, we will separately examine whites and Asians (as one group: white/Asian) and blacks in combination with other races (in a second group). For expositional ease, we will refer to the second group as “blacks.”⁴

Table 1.

Description of Africans in America: By gender, race, and age at arrival in the United States, 2000–2011

Gender	Males	Females	Males				Females
Race	All	All	Black/Other		White/Asian		Black/Other		White/Asian
Age at U.S. Arrival			≥18	<18	≥18	<18	≥18	<18	≥18	<18
Sample Size	38,546	33,291	23,229	2,505	9,881	2,931	19,963	2,662	7,865	2,801
A. Country of Birth
South Africa/Zimbabwe	6.7	8.1	1.5	2.4	19.8	16.7	2.3	2.6	24.1	19.7
Nigeria	14.5	13.2	20.0	18.8	0.8	2.9	18.5	13.6	0.9	2.4
East Africa	9.1	10.8	8.4	8.3	10.2	12.2	10.3	7.7	12.8	12.7
Ghana	8.3	7.6	11.6	10.1	0.3	0.8	10.6	9.5	0.3	0.8
Senegal/Cameroon	2.6	2.5	3.7	2.1	0.2	0.5	3.5	2.6	0.3	0.7
Liberia	3.7	4.6	4.8	6.2	0.4	1.3	6.0	8.1	0.1	1.2
Egypt	10.1	8.4	0.9	1.6	35.1	23.4	0.5	0.6	31.0	20.4
Ethiopia/Eritrea	10.5	12.0	14.2	13.4	0.9	4.2	16.0	16.0	1.4	2.8
Cape Verde	1.6	2.5	1.7	5.6	0.3	0.9	2.5	8.4	0.6	0.8
Algeria/Morocco	6.6	5.5	1.2	1.3	21.2	14.3	1.0	1.3	17.9	14.5
Sudan/Somalia	6.4	6.5	8.7	7.5	1.2	1.2	9.0	6.7	1.0	0.9
Other African countries	19.9	18.2	23.1	22.6	9.6	21.5	19.9	22.9	9.4	23.1
B. Race
% Black	69.0	69.4	97.2	94.3			97.1	92.1
% White	25.5	24.0			88.0	88.9			84.6	88.0
% Asian	3.4	4.1			12.0	11.1			15.4	12.0
% Other	2.2	2.5	2.8	5.7			2.9	7.9
C. Work and Earnings
% Work for income	90.8	74.7	90.6	88.8	91.7	91.9	76.6	83.6	64.0	77.9
% Self employed	12.1	5.8	10.1	8.4	18.1	15.9	4.6	4.1	8.4	11.4
Hourly earnings	26.02 [0.26]	22.09 [0.37]	22.13 [0.31]	23.66 [1.06]	34.44 [0.54]	37.48 [1.06]	19.69 [0.35]	23.53 [1.01]	26.63 [1.27]	30.06 [1.94]
ln(hourly earnings)	2.91 [0.01]	2.77 [0.01]	2.79 [0.01]	2.86 [0.02]	3.14 [0.01]	3.26 [0.02]	2.69 [0.01]	2.84 [0.02]	2.88 [0.02]	3.02 [0.03]
ln(hourly earnings) if self-employed	2.90	2.70	2.70	2.64	3.15	3.29	2.55	2.74	2.99	2.95
ln(hourly earnings) if wage sector	2.91	2.80	2.80	2.88	3.14	3.25	2.70	2.84	2.87	3.04
D. Human Capital
Education (years)	13.53 [0.03]	12.3 [0.04]	13.2 [0.04]	13.6 [0.11]	14.3 [0.06]	14.3 [0.09]	11.6 [0.05]	13.3 [0.11]	13.4 [0.07]	14.1 [0.12]
% Completed college	75.2	64.7	72.3	76.2	81.0	82.1	58.4	74.1	73.6	82.1
% Some education in U.S.	22.9	23.8	13.6	84.1	12.0	91.5	11.7	82.9	11.7	91.9
% Only English spoken at home	24.9	24.0	19.1	38.4	27.3	61.4	17.4	34.1	26.4	62.1
% Speak English very well/not at home	49.9	45.6	52.4	48.4	47.8	33.0	47.1	51.6	43.3	31.8
% Speak English well/not at home	19.3	18.9	21.5	11.2	19.5	4.1	21.7	10.2	19.0	5.1
E. Years in U.S.	14.3	13.9	11.7	22.3	14.0	33.0	10.4	22.3	14.0	33.4
F. Citizenship
% U.S. citizen (American parents)	3.5	4.3	0.8	8.5	1.7	32.2	1.0	7.4	2.4	35.2
% U.S. citizen (naturalized)	43.3	42.2	39.2	51.6	49.3	54.7	36.1	58.4	49.8	53.8

Note: Standard errors are shown in brackets.

Source: 2000–2011 American Community Survey public-use microdata samples (sample sizes are not weighted; all estimates are weighted).

The differences in the country-of-birth distribution by race and age at arrival in the United States are substantial. For example, among males who arrived in the United States as adults (i.e., age 18 and older), close to one-half of black migrants were born in Nigeria (20 %), Ethiopia/Eritrea (14.2 %), or Ghana (11.6 %). In contrast, about three-quarters of white migrants who arrived as adults were born in South Africa/Zimbabwe (19.8 %), Egypt (35.1 %), or Algeria/Morocco (21.2 %). Male migrants who came as children account for a small fraction of all first-generation African-origin migrants: less than 10 % of blacks and less than 30 % of white/Asian male migrants. Most of them probably came with their parents. The distributions of black and white/Asian male migrants by country of origin for those who arrived as children are similar to the distributions for those who came as adults, although a higher fraction of those who arrived as children self-report as being an “other” race, suggesting that identification with race is more complex than for their parents.

The differences in the country of birth distributions by race point to potentially important differences between white/Asian and black males in the propensity to leave their countries for the United States. For example, the numbers of black versus white/Asian migrants to the United States from South Africa/Zimbabwe imply that only 1 in every 1,000 southern African blacks has moved to the United States, versus more than 50 of every 1,000 white/Asian southern Africans. Black movers from these countries are likely to be especially positively selected. As shown in the last four columns of Table 1, the distributions of the country of birth and age at migration among black and white/Asian females are broadly similar to those for male migrants.

Panel C displays summary statistics for labor market outcomes. The fraction with any earnings in the 12 months preceding the survey is reported in the first row of the panel, along with the fraction reporting self-employment income. Mean real hourly earnings (in 2010 dollars) and the mean of the logarithm of real hourly earnings are reported in the next rows.

White/Asian migrants from Africa earn significantly more than black migrants. For those arriving as adults, white/Asian males earn 55 % more than black males ($34 per hour and $22 per hour, respectively) and white/Asian females earn 35 % more than black females ($27 per hour vs. $20 per hour, respectively). Migrants who arrived as children have higher hourly earnings than those who arrived as adults, on average, and the racial gaps are slightly bigger for males and slightly smaller for females. Male migrants earn more than female migrants, although the gender gaps are much smaller for black migrants than for whites/Asians, especially among those who arrived as children.

These patterns may partly reflect differences in human capital, demographic characteristics, time spent in the United States, and assimilation, as well as price differences across states of residence. As seen in panels D–F in Table 1, educational attainment is higher for whites/Asians than for blacks. Whites/Asians are also more likely to speak only English at home (particularly if they arrived before age 18), to have lived longer in the United States, and to be U.S. citizens. Among those who arrived before age 18, about one-third of whites/Asians are the children of American parents, versus less than 10 % of blacks. In addition to these observed characteristics, unobserved characteristics likely to influence labor market outcomes, and they may vary by country of origin.

Hourly Earnings of African-Origin Male Migrants

Table 2 presents results from the linear regression models of the logarithm of real hourly earnings for African-born male migrants to the United States. For all males combined, we show results from the two empirical models described earlier, without human capital controls (Model 1) and with human capital controls (Model 2). All coefficients are multiplied by 100 and thus represent percentage differences in earnings relative to the excluded category. The omitted category for country of birth consists of African-born male migrants from countries in North, West, and Southern Africa that are not separately identified in the ACS.

Table 2.

Migrant characteristics that predict ln(hourly earnings) for males aged 25–64 at interview: Roles of country of birth, race, human capital, citizenship, and age at arrival in the United States, 2000–2011

Race:	All Males		Black/Other		White/Asian
Age at U.S. Arrival:			≥18	<18	≥18	<18
	(1)	(2)	(3)	(4)	(5)	(6)
A. (1) If Country of Birth Is:
South Africa/Zimbabwe	39.72** [3.20]	26.45** [3.41]	9.96** [3.48]	−7.75 [4.39]	37.93** [4.67]	3.65 [3.20]
Nigeria	22.21** [1.33]	4.26** [1.30]	6.79** [1.09]	2.70 [4.50]
East Africa	20.32** [2.48]	10.63** [1.60]	7.51* [2.76]	18.35 [11.59]	24.28** [5.09]	−11.26** [3.27]
Ghana	13.43** [1.17]	8.20** [0.71]	7.95** [0.90]	12.66* [5.30]
Senegal/Cameroon	13.68 [10.72]	1.24 [3.10]	2.02 [3.16]	−10.09 [6.45]
Liberia	4.39** [1.19]	0.50 [1.05]	0.18 [1.56]	1.60 [3.15]
Egypt	−0.11 [3.66]	−8.60* [3.40]	−1.46 [1.82]	1.53 [7.17]	−8.59 [4.47]	−2.62 [3.29]
Ethiopia/Eritrea	1.51 [1.20]	1.52 [1.29]	−0.29 [1.16]	5.95 [7.62]
Cape Verde	−8.44** [2.57]	8.91** [2.64]	9.36** [3.04]	5.32 [7.90]
Algeria/Morocco	−13.66** [3.57]	−11.85** [2.92]	−7.10 [3.90]	−22.45** [5.17]	−8.94* [3.94]	3.35 [2.26]
Sudan/Somalia	−7.70 [5.18]	−4.40 [3.31]	−6.61** [1.82]	−1.40 [8.31]
Other African countries	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
B. (1) If Race Is:
Asian	36.11** [4.48]	29.45** [4.17]			−2.56 [3.11]	5.99 [4.62]
White	31.54** [4.24]	24.84** [3.94]			Ref.	Ref.
Other	8.47 [5.21]	8.62* [4.11]	7.03 [5.26]	15.85** [5.12]
Black	Ref.	Ref.	Ref.	Ref.
C. Education (1) If:
Post-college		62.33** [2.19]	57.60** [2.53]	58.72** [5.01]	73.33** [6.35]	80.73** [5.71]
Completed college		30.37** [2.66]	27.04** [2.74]	28.20** [6.01]	37.38** [8.20]	53.60** [5.69]
Some college		9.11** [2.19]	9.29** [2.00]	2.53 [5.70]	10.36 [7.48]	16.84* [7.04]
Less than college		Ref.	Ref.	Ref.	Ref.	Ref.
Educated in U.S.		−1.42 [2.25]	−0.15 [2.09]	2.49 [4.98]	−4.83 [7.20]	−5.77 [8.31]
Not educated in U.S.		Ref.	Ref.	Ref.	Ref.	Ref.
D. (1) If Speaks English:
At home		19.38** [2.89]	16.12** [4.03]	48.39** [14.43]	16.49** [3.55]	29.63** [7.64]
Very well (but not at home)		13.90** [2.44]	12.03** [2.88]	43.54** [13.64]	13.65** [2.77]	19.44* [8.29]
Well (but not at home)		2.30 [1.92]	3.95 [2.11]	24.56 [14.87]	−7.50 [4.28]	14.75 [7.19]
Not well, not at home		Ref.	Ref.	Ref.	Ref.	Ref.
E. U.S. Citizen (1) If:
Citizen by birth		5.72 [3.28]	16.33** [4.28]	2.78 [7.08]	2.46 [9.22]	−2.38 [4.52]
Citizen by naturalization		6.68** [1.80]	7.15** [1.98]	5.84 [4.91]	2.82 [3.75]	9.56 [4.70]
Not a citizen		Ref.	Ref.	Ref.	Ref.	Ref.
F. (1) If years in U.S.:
>25 years	60.15** [4.40]	33.67** [5.81]	23.62** [4.50]	23.07* [9.43]	47.28** [4.69]	52.79** [14.72]
16–25 years	43.23** [2.89]	25.65** [3.53]	23.64** [4.16]	11.48 [7.41]	36.55** [5.02]	22.62 [13.85]
11–15 years	33.49** [1.84]	21.98** [2.06]	20.49** [2.49]	2.66 [7.41]	30.16** [4.83]	20.56 [16.24]
6–10 years	19.80** [2.07]	13.95** [2.33]	13.12** [2.90]		17.06** [3.87]
0–5 (or 0–10) years	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Sample Size	34,904	34,904	21,007	2,171	9,032	2,694
R2	.13	.23	.17	.19	.27	.25

Notes: Dependent variable is 100 × logarithm of annual earnings in 2010 dollars divided by annual hours of work. All models also include indicator variables for year of survey and age at arrival in the United States. Models (2)–(6) also include state of residence in United States at time of survey to adjust for price variation across United States. All variance-covariance matrices take into account clustering at the level of country of birth and are robust to heteroskedasticity of arbitrary form. Standard errors are shown in brackets below OLS coefficient estimates. All estimates are weighted by sampling weights.

Source: 2000–2011 ACS public-use microdata samples.

^*p < .05;

^**p < .01

There are large differences in men’s real hourly earnings by country of birth (column 1). Migrants born in South Africa/Zimbabwe earn the most: 40 % more than the omitted group of migrants. Men born in Algeria/Morocco, on the other hand, earn 14 % less, and those from Cape Verde earn 8 % less. Other migrants with relatively high earnings are Nigerians (22 %), East Africans (20 %), and Ghanaians (13 %). As shown in panel B of the table, white and Asian male migrants earn about 30 % more than blacks, and the differences between whites and Asians are not statistically significant. Male migrants of other races earn an 8 % premium over black male migrants.

Model 2 includes controls for education, English ability, citizenship, years in the United States, age at arrival, marital status, and U.S. state of residence. The country of birth differences diminish substantially. The most striking impact is for migrants from Cape Verde, whose educational attainment is relatively low compared with male migrants from the other African countries. Their relative earnings changed from a statistically significant –8.4 % deficit in Model 1 to a statistically significant 8.9 % advantage in Model 2 relative to the omitted category. In addition, the earnings advantages decline by between one-third and one-half for men born in South Africa/Zimbabwe, East Africa, and Ghana. Relative to the omitted category, the earnings advantages for men born in Nigeria and Senegal/Cameroon are nearly eliminated, the earnings disadvantage shrinks slightly for men born in Algeria/Morocco, and significant negative gaps open up for Egyptian males. The gradient in hourly earnings associated with educational attainment is steep, with some post-college education conferring an advantage of 62 % relative to less than a college education.

Real mean hourly earnings of white and Asian male migrants were significantly higher than those of black males in column 1. The gap shrinks with controls for human capital, but Asian men earn 29 % more, white men earn 25 % more, and other race men earn 9 % more than black men—and each of these gaps is statistically significant.

The results in columns 1 and 2 suggest that some roots of labor market success reflect social, political, economic, and cultural environments including experiences in early life. Given that these contexts vary as a function of race and age of arrival in the United States, the models are stratified by these characteristics. Results from Model 2 are presented in columns 3 and 4 for blacks and in columns 5 and 6 for whites/Asians. In the ACS, there are four country groups in which the fraction of migrants who are black is less than 90 %: the models separately identify black and white/Asian migrants for those countries. The composition of the excluded group in the regression for whites/Asians (whites/Asians from other African countries) is not the same as the composition of the excluded group as in the regressions for blacks, but the differences in earnings between the two excluded groups are small (and statistically insignificant), and none of our conclusions change if the models are estimated with a single excluded group of blacks and whites/Asians (results not shown).

Among black male migrants, there is more heterogeneity in earnings differences by country of birth relative to the omitted category among those who arrived in the United States before age 18 than among those who arrived at older ages. Among the former, migrants from Algeria/Morocco earn 22 % less, whereas those from East Africa earn 18 % more than the reference group. Differences among those who arrived as adults range from –7 % for males born in Algeria/Morocco to 10 % for men born in South Africa/Zimbabwe, relative to the omitted category. For those born in East Africa, Ghana, Egypt, and Ethiopia/Eritrea, the country-of-birth advantage is greater for those who arrived in the United States as children than for those arriving as adults. The reverse is true for migrants from South Africa/Zimbabwe, Senegal/Cameroon, and Algeria/Morocco. For those migrants, those who arrived as children earn substantially less than the reference group. Migrants who self-report as being an “other” race earn significantly more than those who self-identify as black, particularly among those who arrived as children, for whom the earnings advantage is 16 %.

There is also considerable heterogeneity in earnings by country of birth as well as variation in patterns by age at arrival among white/Asian men. Those from South Africa/Zimbabwe and East Africa who arrived as adults earn significantly more than the reference group: 38 % and 24 % more, respectively. In contrast, among those who arrived before age 18, earnings for South Africans/Zimbabweans are only 4 % higher, whereas earnings for East Africans are 11 % lower relative to the omitted category. White/Asian male migrants from Algeria/Morocco and from Egypt fare somewhat better (relative to their reference groups) if they came before age 18 than later. For white/Asian male migrants, we control for Asian race, and the racial gaps are small and not statistically significant.

We now turn to the results for human capital. Estimated returns to completing college or enrolling beyond college are high relative to having less than a college education. The educational gradient in wages is somewhat steeper for white/Asian males than for black males. Although being educated in the United States is not a statistically significant predictor of earnings, speaking English at home or speaking English well are rewarded in the labor market, regardless of race or age of arrival, although the effect is strongest for black males who arrived as children. English proficiency may be a particularly effective signal of successful assimilation for these migrants, and it may reflect differences in ability, linguistic skills, or other characteristics related to background and early life experiences. Migrants who have naturalized—another marker of assimilation—earn a 6 % to 7 % premium if they are black. Among whites/Asians, the premium to naturalization is 3 % for those who arrived as adults versus 10 % for those who arrived as children, although neither of these is statistically significant.

Longer duration of residence in the United States is associated with higher earnings among both black and white/Asian males, with greater benefits as duration rises for whites and Asians than for blacks. This likely reflects both the selectivity of longer-term migrants at time of arrival and that the probability of staying in the United States for the longer term is greater for those who have greater success in the labor market.

Self-employment and Earnings by Type of Employment Among Male African-Origin Migrants Individuals who leave their country of birth are likely to be selected on characteristics—some observed and some not—that are more highly valued in the destination than in the origin location. For international migrants to the United States, unobserved traits often hypothesized to be important in the theoretical literature include ambition and entrepreneurial ability (Evans and Jovanovic 1989). Neither is measured in the ACS, so we turn to labor market outcomes that are likely related to these traits and explore variation in the probability of self-employment and hourly earnings in self-employment. For some, self-employment may be an important route to success in the labor market, particularly if one is entrepreneurial or prone to risk-taking—skills that are potentially valuable but not highly valued (or readily observed) by employers in the United States. For others, self-employment may be the better option in a competitive labor market. Self-employment may also reflect a choice made in response to discrimination against workers with observable traits similar to those of the migrant. To provide evidence on these issues, we compare earnings from self-employment with earnings in the market wage sector. In the absence of plausible characteristics to adjust for selection into each sector, we interpret the estimates conditional on sectoral choice. Regression results predicting self-employment and earnings for men by self-employment status, adjusting for human capital and sociodemographic characteristics, are reported in Table 3.

Table 3.

Male migrants’ probability of being in the self-employment sector, and earnings in self-employment and wage sector: Roles of country of birth, race, human capital, citizenship, and age at arrival in the United States, 2000–2011

	Age at Arrival ≥18						Age at Arrival <18
	Black/Other			White/Asian			Black/Other			White/Asian
	Probability	ln(Hourly Earnings)		Probability	ln(Hourly Earnings)		Probability	ln(Hourly Earnings)		Probability	ln(Hourly Earnings)
	Self-Emp	Self-Emp.	Wage	Self-Emp.	Self-Emp.	Wage	Self-Emp.	Self-Emp.	Wage	Self-Emp.	Self-Emp.	Wage
	(1)	(2)	(3)	(1)	(2)	(3)	(1)	(2)	(3)	(1)	(2)	(3)
A. (1) If Country of Birth Is:
South Africa/Zimbabwe	3.5 [2.1]	−4.0 [19.1]	13.4** [2.5]	6.8** [1.6]	44.0** [7.9]	37.2** [4.5]	11.8 [5.9]	8.2 [18.6]	−10.2* [4.8]	3.1 [2.5]	−22.5 [19.9]	9.3** [3.0]
Nigeria	2.8** [0.8]	12.8** [2.5]	6.7** [1.0]				1.2 [0.9]	−7.1 [9.5]	−0.9 [4.4]
East Africa	−1.5 [0.9]	26.7** [4.4]	6.6* [2.8]	4.6 [3.2]	26.6** [9.4]	23.9** [4.4]	−3.6* [1.4]	−23.7 [27.8]	15.8 [12.6]	0.0 [2.1]	−33.0 [17.2]	−7.1 [4.1]
Ghana	−1.7* [0.8]	8.4* [3.4]	7.7** [0.8]				−2.7* [1.1]	−29.1 [18.3]	7.2 [4.9]
Senegal/Cameroon	4.1* [1.7]	−9.4* [4.4]	4.5* [1.9]				9.0 [7.6]	−51.3** [11.7]	−6.5 [4.6]
Liberia	−3.3** [0.8]	54.3** [4.3]	−3.5* [1.4]				−2.2 [1.2]	37.3* [15.6]	−6.4 [3.7]
Egypt	6.2** [1.1]	14.8* [5.7]	−3.7* [1.8]	0.2 [1.6]	−10.6 [8.5]	−7.3 [4.0]	12.1** [1.7]	−38.6 [32.1]	−2.0 [6.8]	2.6 [2.5]	2.9 [10.1]	−5.8 [3.4]
Ethiopia/Eritrea	3.8** [0.9]	−5.8 [3.5]	2.0 [1.1]				0.8 [1.2]	−19.2 [17.7]	5.5 [8.7]
Cape Verde	−5.2** [1.7]	31.3** [9.9]	7.5* [2.9]				−4.3 [2.4]	−42.7 [40.4]	5.5 [7.6]
Algeria/Morocco	6.5 [5.3]	−6.0 [5.5]	−5.9 [3.7]	0.8 [1.6]	−12.3 [8.7]	−7.6* [3.6]	2.1 [2.7]	−113.2** [26.8]	−11.9 [9.6]	−2.6 [2.2]	−3.6 [13.2]	2.7 [2.5]
Sudan/Somalia	1.2 [1.0]	−8.7** [2.7]	−5.7** [2.0]				−1.7 [2.3]	−40.9 [24.5]	−4.5 [6.6]
Other African countries	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
B. (1) If Race Is:
Other	−0.4 [1.5]	9.5 [17.7]	6.5 [4.2]				−3.5 [2.0]	99.9** [25.1]	12.3* [5.3]
Black	Ref.	Ref.	Ref.				Ref.	Ref.	Ref.
Asian				4.4* [1.7]	15.8 [15.4]	3.1 [3.1]				0.8 [3.1]	−1.1 [22.8]	13.4* [5.0]
White				Ref.	Ref.	Ref.				Ref.	Ref.	Ref.
C. Education (1) If:
Post-college	−2.9** [1.0]	50.5** [5.7]	57.3** [2.4]	−2.7 [1.8]	93.4** [12.7]	68.7** [5.7]	−4.1* [1.6]	152.4** [33.0]	54.0** [5.3]	−1.6 [2.6]	100.4** [19.8]	79.8** [6.7]
Completed college	−2.0* [0.9]	18.0** [4.2]	26.9** [3.1]	−2.5 [1.6]	33.0** [10.1]	37.6** [8.7]	0.0 [1.4]	61.0** [15.1]	28.5** [6.4]	−4.1* [1.9]	19.3 [17.9]	62.7** [5.5]
Some college	−0.3 [0.6]	5.0 [4.9]	9.5** [2.3]	3.7* [1.5]	15.9 [12.1]	9.9 [6.1]	−0.7 [1.7]	4.5 [18.4]	4.3 [5.5]	−3.5 [2.3]	13.2 [23.5]	20.8** [7.4]
Less than college	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Educated in U.S.	−0.7 [1.7]	−1.8 [8.1]	0.0 [2.3]	−3.3 [1.7]	−10.9 [10.1]	−4.5 [7.6]	0.7 [1.9]	17.0 [29.4]	0.2 [5.4]	−13.7** [4.5]	−22.2 [16.2]	−5.4 [5.4]
Not educated in U.S.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
D. (1) If Speaks English:
At home	−0.9 [1.3]	30.6** [8.5]	14.9** [4.2]	0.9 [1.9]	9.6 [18.3]	19.0** [5.1]	−6.8 [6.4]	84.2 [41.8]	40.5** [14.1]	14.1** [3.2]	111.8** [39.1]	23.2** [6.6]
Very well (but not at home)	−0.3 [1.5]	29.3** [5.7]	10.5** [2.9]	0.6 [1.6]	12.1 [19.2]	15.0** [3.3]	−7.9 [6.8]	125.3** [40.8]	29.6* [13.7]	15.1** [3.7]	102.0* [39.7]	12.3 [6.1]
Well (but not at home)	1.7 [1.0]	13.7 [8.8]	3.6 [2.2]	1.7 [1.9]	3.4 [21.2]	−8.7 [5.2]	−4.7 [6.9]	116.7* [41.9]	14.4 [14.0]	13.6 [7.5]	76.3 [46.3]	15.9 [11.6]
Not well, not at home	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
E. U.S. Citizen (1) If:
Citizen by birth	0.2 [2.3]	16.7 [24.4]	16.4** [5.2]	0.9 [4.1]	−13.4 [25.0]	5.8 [7.3]	1.5 [3.9]	−2.3 [32.9]	1.1 [5.4]	−6.0* [2.8]	3.0 [13.6]	−8.7 [4.2]
Citizen by naturalization	−0.1 [1.1]	−11.0* [4.6]	9.6** [1.9]	0.9 [1.4]	−2.9 [7.3]	3.4 [3.6]	1.8 [2.0]	−8.0 [17.3]	9.6 [5.2]	−0.2 [1.6]	22.9* [9.7]	8.3 [4.6]
Not a citizen	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
F. Years in U.S. (1) If:
>25 years	13.0** [0.9]	8.4 [15.3]	27.2** [4.4]	20.2** [2.4]	58.9** [12.1]	46.2** [5.0]	1.0 [4.5]	77.6* [34.2]	20.5* [8.8]	17.8** [3.3]	38.6 [76.5]	46.3** [14.9]
16–25 years	9.9** [0.8]	16.5 [16.0]	25.4** [3.5]	17.6** [1.4]	37.7** [11.5]	39.0** [4.6]	1.3 [2.9]	61.2 [38.7]	10.7 [7.8]	14.4** [3.2]	10.4 [83.5]	18.6 [14.0]
11–15 years	7.5** [1.2]	14.7 [12.0]	21.3** [2.2]	13.3** [1.5]	30.6** [9.7]	31.5** [4.6]	−1.4 [2.5]	47.4 [40.2]	−0.4 [8.0]	4.9 [4.5]	93.1 [84.3]	11.8 [14.6]
6–10 years	4.6** [0.7]	3.2 [13.8]	14.1** [2.8]	6.7** [1.4]	17.9* [8.4]	17.7** [3.5]
0–5 (or 0–10) years	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.

Notes: Linear probability model for probability of self-employment (coefficients × 100) in column 1; OLS for 100 × ln(hourly earnings) conditional on sector in columns 2 and 3. All models also include indicator variables for year of survey, years in United States, and age at arrival. All models also include state of residence in United States at time of survey. All variance-covariance matrices take into account clustering at the level of country of birth and are robust to heteroskedasticity of arbitrary form. Standard errors are shown in brackets below OLS coefficient estimates. All estimates are weighted by sampling weights.

Source: 2000–2011 ACS public-use microdata samples.

^*p < .05;

^**p < .01

Among migrants who arrived as adults, differences in the probability of being self-employed by country of birth are small in magnitude. Country-specific differences in the hourly earnings gaps between the self-employed and wage workers for whites/Asians are small, suggesting that the market wage sector does not value country-specific traits differently among these migrants. However, the patterns are quite different among blacks who arrived as adults, some of whom earn very high premiums in the self-employed sector over the wage sector. For example, relative to the reference group of migrants, self-employed Liberians earn nearly 60 % more than wage workers, Cape Verdeans and East Africans earn over 20 % more, and Egyptians earn about 17 % more in the self-employed sector. These earnings advantages for the self-employed relative to wage workers are consistent with black self-employed migrants having difficulty signaling their value to employers but reaping rewards when they strike out for themselves in the labor market.

Black migrants who arrived as children from Egypt and South Africa/Zimbabwe are more than 10 percentage points more likely to be self-employed than migrants born in other countries. South Africans/Zimbabweans who are self-employed earn almost 20 % more than wage workers, while Liberians earn 40 % more in the self-employed sector (paralleling the result for Liberians who arrived as adults). All other blacks who arrived as children earn less in the self-employed sector, with the gap being very large in some cases (more than 100 % among Algerians/Moroccans and more than 50 % among Senegalese/Cameroonians). Among whites/Asians who arrived as children, there are small differences in the probability of being self-employed, and wage gaps are small except for self-employed South Africans/Zimbabweans and East Africans, who earn 30 % to 40 % less than those in the wage sector.

We also examine how the probability of self-employment and sector-specific earnings vary by other markers of human capital. Sectoral choice is not highly related to education. The better-educated are less likely to be self-employed; these differences, however, are small and not always statistically significant, although whites/Asians who arrived as children and were educated in the United States are much less likely to be self-employed.

In contrast, returns to education vary substantially between self-employed and wage workers. Among blacks who migrated at older ages, the educational gradient in earnings is flatter for the self-employed than for wage workers; comparatively, for those who arrived before age 18, earnings of the self-employed rise far more dramatically with education than is the case for those earning a wage. This difference in the relationship between education and wages by age of arrival is apparent but more muted among whites. Among blacks and among whites who arrived as adults, speaking English well has a modest association with the probability of being self-employed, but whites who arrived before age 18 are much more likely to be self-employed if they speak English well. The rewards for speaking English well are large and significant, particularly for those who arrived before age 18 and work in the self-employed sector. Those who have been in the United States longer earn much more than those who arrived recently, which likely reflects a combination of changing selectivity of migrants over time and the effect of lower earning migrants returning to Africa. Citizenship also carries a substantial premium, particularly among blacks in the wage sector who arrived as adults. Over and above country of origin, there are large returns to different types of human capital, and the nature of these returns is different for those who arrived as adults relative to those who arrived as children as well as for blacks and whites/Asians. A fuller understanding of this heterogeneity is likely to be informative about the mechanisms underlying the success of African-origin migrants in the United States.

Hourly Earnings of African-Origin Female Migrants

We next discuss the results for real hourly earnings among women (Table 4). Because the models are estimated only for those who reported positive earnings, we stratify by race and age of arrival and estimate two models, the second of which introduces controls for human capital and all other explanatory variables.

Table 4.

Migrant characteristics that predict ln(hourly earnings) for females aged 25–64 at interview: Roles of country of birth, race, human capital, citizenship, and age at arrival in United States, 2000–2011

Race:	All Females		Black/Other		White/Asian
Age at U.S. Arrival:			≥18	<18	≥18	<18
	(1)	(2)	(3)	(4)	(5)	(6)
A. (1) If Country of Birth Is:
South Africa/Zimbabwe	26.36** [4.04]	14.72** [3.56]	3.76 [6.22]	−1.34 [16.47]	20.04** [4.26]	8.31 [9.05]
Nigeria	27.67** [2.60]	10.73** [2.03]	11.89** [1.98]	−4.44 [3.42]
East Africa	14.70** [4.08]	4.63 [3.30]	4.31 [2.91]	−3.10 [5.02]	1.91 [6.28]	9.96 [8.11]
Ghana	9.56** [2.53]	7.53** [1.96]	7.99** [1.78]	−8.08* [3.54]
Senegal/Cameroon	18.37** [6.50]	6.11 [3.74]	5.81 [4.70]	15.42* [5.82]
Liberia	1.64 [2.56]	0.81 [2.26]	−1.09 [2.08]	−2.02 [4.13]
Egypt	12.22* [5.00]	0.97 [4.04]	−1.48 [2.40]	0.11 [4.93]	2.13 [4.17]	10.74 [7.01]
Ethiopia/Eritrea	−1.92 [3.40]	0.52 [2.09]	−3.70 [1.92]	8.65 [6.87]
Cape Verde	−9.72* [3.92]	−0.71 [3.05]	−2.68 [2.51]	−5.51 [5.78]
Algeria/Morocco	4.22 [5.93]	2.19 [3.92]	0.15 [4.19]	0.89 [11.74]	0.43 [4.85]	13.38 [6.66]
Sudan/Somalia	−12.26** [4.38]	−4.21 [2.24]	−5.96** [1.98]	−7.95 [6.18]
Other African countries	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
B. (1) If Race Is:
Asian	9.45 [5.62]	3.33 [4.51]			3.16 [4.16]	14.45* [5.90]
White	7.06 [4.97]	0.43 [4.37]			Ref.	Ref.
Other			1.22 [0.2]	0.16 [0.0]
Black	Ref.	Ref.	Ref.	Ref.
C. Education (1) If:
Post-college		62.61** [2.88]	53.93** [4.04]	74.79** [6.43]	70.73** [3.54]	66.44** [5.46]
Completed college		36.82** [2.52]	35.40** [3.00]	47.87** [4.19]	32.15** [2.25]	42.78** [5.09]
Some college		15.23** [1.71]	15.65** [1.98]	12.53 [6.92]	12.41** [2.87]	18.21* [7.40]
Less than college		Ref	Ref	Ref	Ref	Ref
Educated in U.S.		1.91 [2.64]	3.47 [2.48]	1.52 [5.84]	−7.06 [7.15]	21.11* [10.13]
Not educated in U.S.		Ref.	Ref.	Ref.	Ref.	Ref.
D. (1) If Speaks English:
At home		16.38** [3.51]	10.36** [3.00]	43.00** [12.23]	29.72** [4.03]	−15.19 [15.14]
Very well (but not at home)		15.25** [3.53]	11.83** [3.84]	41.27** [12.15]	27.49** [5.16]	−23.59 [13.62]
Well (but not at home)		3.60 [2.62]	1.51 [2.80]	29.50* [12.53]	9.34 [5.02]	−31.84 [15.84]
Not well, not at home		Ref.	Ref.	Ref.	Ref.	Ref.
E. U.S. Citizen (1) If:
Citizen by birth		6.81 [4.15]	11.25** [3.95]	−3.80 [9.15]	2.21 [6.31]	8.64 [12.83]
Citizen by naturalization		7.94** [1.86]	7.33** [2.16]	6.18 [6.01]	9.26** [3.05]	6.73 [7.60]
Not a citizen		Ref.	Ref.	Ref.	Ref.	Ref.
F. (1) If years in U.S.:
>25 years	60.15** [4.40]	34.29** [3.07]	34.57** [3.38]	30.25** [10.00]	25.13** [8.15]	12.9 [17.99]
16–25 years	43.23** [2.89]	27.23** [2.88]	30.47** [2.40]	15.0 [9.42]	23.81** [7.11]	−2.0 [17.03]
11–15 years	33.49** [1.84]	21.42** [3.09]	23.17** [3.58]	9.7 [10.45]	18.80** [4.91]	−11.7 [13.00]
6–10 years	19.80** [2.07]	14.44** [2.92]	14.80** [3.17]		14.87** [4.58]
0–5 (or 0–10) years	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
Sample Size	24,965	24,965	15,498	2,232	5,059	2,176
R2	.10	.20	.19	.23	.20	.17

Notes: Dependent variable is 100 × logarithm of annual earnings/annual hours of work in 2010 dollars. All models also include indicator variables for year of survey, years in United States, and age at arrival. Models (2)–(6) also include state of residence in United States at time of survey to adjust for price variation across the United States. All variance-covariance matrices take into account clustering at the level of country of birth and are robust to heteroskedasticity of arbitrary form. Standard errors are shown in brackets below OLS coefficient estimates. All estimates are weighted by sampling weights. Source: 2000–2011 ACS public-use microdata samples.

^*p < .05;

^**p < .01

In comparison with males, heterogeneity in earnings across country of birth for females is smaller and the relative positions of countries vary. The sizes of the country-of-birth coefficients vary from –12.3 to 27.7 for women versus from –13.7 to 39.7 for men. With respect to particular countries, relative to the omitted group, the earnings of female migrants from Nigeria are the highest (28 %) and shrink to second place (behind South Africa/Zimbabwe) after controls for human capital are added. Racial differences are much smaller than for men: in fact, earnings for white women are no different from those for black women after controls for human capital are added. (Wages for Asian women remain slightly but not statistically significantly higher.)

Turning to the results that stratify by race and age of arrival, Nigerians are the highest earning workers among blacks who arrived as adults, with a 12 % advantage relative to the omitted group. The lowest-earning black females were born in Sudan/Somalia: they earn 6 % less than the omitted group. Among black women who migrated before age 18, women from Senegal/Cameroon earn 15 % more, whereas the earnings of women from Ghana and Sudan/Somalia are 8 % lower than those of the omitted group.

Differences by gender in the country-of-birth results for blacks suggest that the selection of black migrants from Africa who participate in the U.S. labor force differs in substantively important ways for males and females from the same country. For example, the earnings premium for Nigerians arriving as adults is 7 % for men and 12 % for women. Ghanaian men who migrated before age 18 earn 13 % more than those in the omitted group, but their earnings are 8 % lower among Ghanaian women.

Similar differences between males and females appear in the wage premiums of white/Asian male and female migrants from Africa. Although white/Asian females from South Africa/Zimbabwe have the highest earnings among those who migrated after age 18, that premium (20 %) is only one-half the size of the corresponding male premium (38 %). Furthermore, there are no significant differences in earnings by country of birth among white/Asian women who arrived before age 18, whereas East African men in this group earn 11 % less than the reference group. The evidence suggests complex and substantial heterogeneity in the selection of white/Asian males and females born in Africa, both into migration and then, for women, into the work force in the United States.

The regression models control human capital and sociodemographic characteristics, and the heterogeneity reflects factors that are unobserved in the models. We have pointed to the possible roles of educational quality, social and political context, and family background in driving cross-country differences in earnings. However, it is unlikely that the large differences in selection of males and females from the same country of origin are attributable to differences in education quality or family background.

Possibly at least part of the country-specific differences between male and female migrants in the labor force arise from different propensities to work. Among males aged 25 to 64, the proportion without earned income in the year before the interview is less than 10 % versus around 25 % for females. As a result, selection into the labor market is likely to account for more of the differences among female migrants than among their male counterparts. Labor force participation also varies among women by country of birth. Black female migrants from Arabic-speaking countries—Egypt, Algeria/Morocco, and Sudan/Somalia—are far less likely to be in the labor market than other females, particularly among those who came to the United States as adults. Black Algerian and Moroccan and Egyptian females who arrived as adults were 18 % to 20 % less likely to be working than the omitted group, but no such difference emerges for white/Asian female migrants from these countries. Age at arrival does not seem to be a large part of the story. Differences in labor force participation for females from the same country but who arrived at different ages are relatively small, and after we control for covariates, migrants from most African countries who arrived at younger ages are generally less likely to participate in the labor force.

With respect to differences in women’s earnings by educational attainment, returns to college and to education beyond college are high for all groups of females and are similar in magnitude to those for males. As is the case for males, speaking English at home is associated with higher earnings for all black females and for white and Asian females who arrived as adults; among all women, it is also associated with a substantially higher propensity to be working. In general, longer duration of U.S. residence is also associated with higher earnings among females; but unlike men, the benefits of longer duration of residence appears to be somewhat stronger for black women than for white/Asian women, regardless of age at migration.

Discussion and Conclusions

Over the past several decades, the number of African migrants to America has grown exponentially. Accompanying the growth in numbers, dramatic increases in the diversity of new entrants with respect to country of origin, background, and reason for entry have also occurred. Thomas (2011) analyzed these trends in detail, concluding that they reflect complex interactions among factors that include the evolution of U.S. immigration policies and structural features of sending countries.

The increasing diversity of African migrants raises the question of how labor market outcomes vary for individuals born on the same continent but hailing from disparate countries within it. We focus on these comparisons. By doing so, we contribute a complementary angle to the existing literature that draws out differences between migrants and native-born Americans of the same race or between migrants of the same race but from different regions. The patterns more fully unfold when we examine differences by country of birth in analyses that also stratify by gender, race, and age of arrival, which we establish are important.

Our first main result is simple: the results paint a complex and nuanced picture of labor market outcomes of African-born migrants. This reflects the vast differences across African countries in history and levels of development and the levels and quality of human capital embodied in the population of each country; the diversity of languages, cultural, and racial backgrounds represented by migrants from Africa; and opportunities at home and the option to leave, including country-specific differences in U.S. visa and immigration policies that have afforded very different opportunities for migration across countries and over time. Our understanding of international migration is substantially enriched by embracing this diversity rather than treating African-born migrants as a homogenous group.

Overall, we find substantial differences in labor market outcomes by country of birth. Because we control for race, these differences are more than a byproduct of correlations between race and country of birth. Moreover, they persist in models that are stratified by race. One potential explanation is that they arise as a function of country-specific differences in education and language ability. When we add controls for education, English skills, and citizenship, these differences are reduced but not eliminated.

The patterns of several countries stand out. Hourly earnings of migrants from South Africa/Zimbabwe, many of whom come on employment-related visas, are relatively high; whereas those for migrants from Sudan/Somalia—countries from which a high proportion of new entrants arrive as refugees—are relatively low. Males from Cape Verde, a country with a long history of sending migrants to America, earn considerably more than the reference group after controlling human capital. For each of these countries, significant nuances emerge in our additional analyses, which stratify by race and age of arrival and (for men) consider earnings from self-employment separately from those from the wage sector.

Consider immigrants from Southern Africa, for whom earlier work has documented higher earnings (Borch and Corra 2010; Kollehlon and Eule 2003). In our overall models (Model 2), the earnings premiums are 26.5 % for men and 14.7 % for women from South Africa/Zimbabwe. These figures mask the fact that the premiums are much higher for individuals who arrived after the age of 18 than for those who arrived before the age of 18. Among those arriving after age 18, earnings are a striking (and statistically significant) 38 % higher for white males, 20 % for white females, and 10 % for black males. Still, the earnings of South African/Zimbabwean males and females who arrived when they were under the age of 18 are not statistically different from the reference groups.

The experience of migrants from Sudan/Somalia stands in stark contrast. As shown in Fig. 4, the vast majority of Sudanese/Somalian migrants are refugees—a situation unlikely to equip them with the skills to compete effectively in the workplace. Indeed, in the overall models, earnings of both males and females are at or near the bottom with respect to country-specific differences. Among males, stratifying by age of arrival reveals that the earnings penalty is more negative for men who arrived after 18 than for those who arrived at earlier ages, suggesting that country of birth does not necessarily portend lifelong disadvantage. Moreover, among those born in Ethiopia/Eritrea (other countries from which many refugees arrived in the 1980s and 1990s), those still in the United States experience no wage penalty irrespective of gender or whether they arrived as adults.

Overall, men born in Liberia earn no more than the reference migrants, but this masks the fact that those who are self-employed earn far more than those in the wage sector, and this gap is large irrespective of whether the migrant arrived as an adult or a child. This is important because a high proportion of recent immigrants from Liberia have come as refugees, again suggesting that refugee status may not confer permanent disadvantage in the United States.

Earnings of men from Cape Verde are relatively high after human capital is controlled for and are not substantially different for those who arrived as adults and those who arrived as children. However, among those who arrived as adults, there is a very large premium to working in the self-employed relative to the wage sector, but the reverse is true among those who arrived before age 18. Apparently, self-employment has provided a means for Cape Verdeans who arrived as adults to capitalize on attributes that are not rewarded in the wage sector.

Age at arrival plays an important role in understanding earnings differences among immigrants. Several characteristics exert opposing influences on labor market outcomes by age at arrival. As examples, we note the reversal in the signs of the South Africa/Zimbabwe coefficients on hourly earnings for black males who arrived after versus before age 18. Among the whites/Asians, similar sign reversals occur for the coefficients on being from Algeria/Morocco. With respect to human capital controls, English-speaking abilities appear to offer greater rewards in the labor market for those who arrived before age 18, particularly among blacks.

On average, immigrants who arrived in the United States before age 18 are better educated and report better English skills than those who arrived after age 18. These differences are in the direction one might expect if arriving at earlier ages provides educational opportunities and facilitates assimilation. Migrants who arrived before age 18 also earn more on average. However, among blacks, the overall variation in labor market outcomes across countries of birth is as great for those who arrived before age 18 as for those who arrived after age 18. This is not true for whites, for whom country of birth differences in earnings are smaller among those who arrived before age 18. This racial difference in patterns by age of arrival is consistent with the idea that processes of assimilation work differently for blacks and whites/Asians.

Distinguishing earnings among the self-employed from earnings for those who work for wages also sheds light on labor market outcomes for migrants. As discuss earlier, self-employment potentially plays myriad roles, ranging from a refuge of last resort for those unable to find work in the wage sector to an avenue of opportunity for those with attributes that are not as highly rewarded in the wage sector. Country-of-birth differences in earnings tend to be more dramatic among the self-employed than among those in the wage sector. Earlier, we noted the large premiums that self-employed black males from Liberia, Cape Verde, and East Africa enjoy. Interestingly, the benefits of speaking English well tend to be greater for those who are self-employed than for those in the wage sector (apart from whites/Asians arriving after age 18). The patterns of differences in labor market outcomes between self-employed and wage sector workers are also quite different by race, which is consistent with the idea that processes of signaling and discrimination work differently for black and white/Asian migrants from Africa.

Our discussion to this point has focused on men. To what extent are patterns for women similar? Overall, differences in earnings by country of birth and by race are much smaller for women than for men. With respect to country of birth, smaller differences for women than for men are also noted in work using earlier data (Borch and Corra 2010; Kollehlon and Eule 2003). With respect to race, the advantage of being white relative to black is eliminated in the model that controls for human capital—a dramatic difference relative to the results for men, where the gap remained 24 %. In this respect, our results differ from Borch and Corra (2010), who documented high earnings for white women based on earlier waves of census data.

The patterns for specific countries differ for men and women as well. The disadvantages of being from Egypt or from Algeria/Morocco observed for men are not observed for women. It is important to note that the selection processes almost certainly operate differently for men and women—not only the processes through which migrants come to the United States but also the processes that determine whether they participate in the labor market, something that is far closer to universal for men than for women.

In conclusion, we demonstrated considerable heterogeneity in labor market outcomes by country of birth for migrants from Africa to the United States—heterogeneity that can only be partially explained by observed human capital and demographic characteristics of migrants. The mosaic we described is likely to reflect different selection mechanisms by country of birth and gender that contribute to the variation in the success of African-born migrants to integrate and assimilate in the U.S. labor market. Lack of information on reasons for migration, including type of entry visa and strength of local migrant networks in the United States, is an important limiting factor. A novel research design will be necessary to advance understanding of factors that underlie the decision to migrate and factors that contribute to the success of migrants in the United States. In the absence of randomly assigning people to move (through random assignment of visas, for example) and a longitudinal follow-up of movers and nonmovers, at a minimum studies are needed that capture experiences in the country of origin and provide evidence on the life of the migrant before and after moving.