The Earnings Disadvantage of 21st Century Immigrants in the United States

Abstract

This article provides a comprehensive analysis of the earnings disadvantage of 21st century immigrants in the United States. The study is the first to decompose the earnings disadvantage faced by recent immigrants to present the channels through which immigrants lag behind their native counterparts. The decomposition of the earnings disadvantage reveals that the time spent in the United States is the key determinant of the earnings disadvantage. Other important sources of the earnings disadvantage of immigrants are the levels of English-language proficiency and educational attainment. The decomposition analysis also suggests that low levels of human capital cause an even larger disadvantage for immigrants in the years following the 2008-2009 recession as compared with the corresponding relative returns of the prerecession period. The decomposition analysis and trends in returns to human capital variables highlight the merits of a selective immigration system that favors young, English-speaking, and highly educated individuals.

JEL Classifications: J1, J3, J6

Keywords

relative earnings the Great Recession earnings decomposition American Community Survey 21st century immigrants

Introduction

The percentage of immigrants in the U.S. population declined from 12% in 1920 to 5% in 1970. Since then, there has been a substantial increase in the percentage of foreign-born individuals residing in the United States (see Figure 1). In 2015, 13.5% of the U.S. population, or 1 in 7, was foreign-born (Zong & Batalova, 2017). With such a substantial growth, immigration has attracted a great deal of attention in recent decades. The recent surge in illegal immigration has further fueled the debate over the implications of immigration on society and the economy. From an economic standpoint, the discussion on immigration has largely focused on the effects of immigrants on the labor market outcomes, such as employment and wages, of natives rather than on understanding immigrants’ labor market outcomes.

Figure 1.

Population of immigrants in the United States.

This article analyzes the wage disadvantage experienced by 21st century immigrants relative to natives. In particular, the article presents a decomposition of immigrants’ earnings differential with natives and the trends in the key predictors of immigrants’ labor market success. With a large number of foreign-born workers joining the U.S. labor force, immigrants will have a profound effect on the future of the social security system, the government’s fiscal policy, and the competitiveness of the U.S. economy. Lee and Miller (2000) suggest that selective immigration may help support the social security system and may also relieve fiscal pressures in the short and in the long run. Storesletten (2000) concludes that selective immigration may avert a possible fiscal crisis resulting from retiring baby boomers. Chellaraj, Maskus, and Mattoo (2008) show that an increase in skilled immigration is associated with higher rates of innovation in the United States. However, the formulation of a selective immigration policy, in an effort to maximize the benefits of immigration to the U.S. economy, requires a thorough understanding of the characteristics of recent immigrants that are causing them to lag behind their native peers in the labor market. This study presents such an analysis.

This study extends the existing literature on immigrants’ labor market outcomes in two ways. First, it decomposes the earnings disadvantage of immigrants to ascertain the relative contribution of various characteristics. We find that once the variables of age, state of residence, and survey year are controlled for, earnings of 21st century immigrant men are 25% lower than the earnings of their native counterparts. The decomposition results indicate that time spent in the United States explains 19.6 percentage points of the earnings disadvantage of recent immigrants. We also find that English-language proficiency and educational attainment explain 3 and 1.5 percentage points of the earnings disadvantage, respectively.

Second, the study provides an analysis of the changes in the returns to English-language proficiency and educational attainment over three different periods of the 21st century. These periods are (a) the years before the Great Recession, (b) the years of the recession, and (c) the postrecession years. The results show that low levels of English-language proficiency and educational attainment generate lower earnings for immigrants in comparison to natives in the postrecession period as compared with the prerecession period.

The results presented in this article can inform economic aspects of the immigration policy. For example, the findings that highly educated and English-proficient immigrant men are subject to a much lower earnings disadvantage, and that this phenomenon is even more pronounced in the postrecession labor market, can be used to formulate a selective immigration policy that favors highly educated and English-proficient individuals.

Literature Review and Motivation

Most studies pertaining to the labor market outcomes of immigrants in the United States calculate the earnings disadvantage of immigrants with minimal controls and then focus on issues such as the rate of economic assimilation of immigrants and changes in cohort quality over time.¹ To the best of our knowledge, there are no studies that decompose the earnings disadvantage of recent immigrants. However, many studies have explored the trends in the earnings disadvantage of immigrants. Two prominent studies that present the trends in the relative earnings of 21st century immigrants are mentioned below.

Borjas and Friedberg (2009) try to establish the reason behind the increase in the relative earnings of immigrants in the 1990s and to see if this increase has continued for 21st century immigrants. Using Current Population Survey (CPS) data for 1994 to 2009, they conclude that the increase in relative earnings of immigrants has not continued for 21st century immigrants. Borjas and Friedberg (2009) do not detail the earnings structure of 21st century immigrants in their analysis, as their focus is on explaining the increase in relative earnings of immigrants in the 1990s.

Borjas (2015) uses 1970, 1980, 1990, and 2000 Census data along with the pooled 2009-2011 American Community Survey (ACS) data to find the reasons behind the slowdown in the rate of economic assimilation of recent immigrants. By developing a human capital accumulation model and using English-language proficiency as a proxy for human capital, he concludes that some of the slowdown in the rate of growth of earnings is attributable to a decline in the rate at which immigrants learn English. The focus of the study by Borjas (2015) is to find the reasons behind the slowdown of growth in earnings of recent cohorts compared with earlier cohorts of immigrants and not to disentangle the sources of the wage disadvantage.

Hence, the current literature on the relative earnings of recent immigrants lacks a study that decomposes the earnings disadvantage faced by 21st century immigrant men. This article presents such a decomposition analysis of the earnings disadvantage of immigrants in terms of immigrants’ characteristics. It also examines the changes in the returns over time to English-language proficiency and educational attainment. The latter analysis highlights the large earnings disadvantage of low-skilled immigrants in the postrecession period.

Data

The empirical work uses ACS data from the Integrated Public Use Microdata Series (IPUMS), which is compiled by Ruggles et al. (2015), for the years 2001-2013. ACS is administered by the U.S. Census Bureau and provides a large cross-sectional dataset, including information on economic, demographic, and educational characteristics of individuals and households.

Before reviewing the descriptive statistics, it is useful to mention some details of the variables and the sample construction. First, the study concerns itself only with men between 18 and 64 years of age. It covers all 50 states and the District of Columbia. All respondents who have either not worked at all in the past year or who are attending school or college are dropped. Only individuals who work for a wage or salary are included in the sample.² The education variable is a categorical variable with eight categories: primary or less than primary education, less than high school, high school graduate, some college education, associate’s degree, bachelor’s degree, master’s degree, and advanced degree (PhD or a professional qualification beyond bachelor’s degree). IPUMS occupation codes and classifications are used to construct 22 occupation dummies representing different industries and job categories.

Finally, an immigrant is defined as an individual who was born in a foreign country and now resides in the United States. This definition includes both people who entered the United States legally and those who entered illegally.³ It also includes refugees, asylees, and nonimmigrant visa holders, such as individuals who hold temporary work visas. As this study concerns itself with the immigrants of the 21st century, all immigrants who came to the United States before 2000 are dropped from the sample. Table 1 presents summary statistics.

Table 1.

Earnings and Education of Natives and Immigrants.

	Natives	Immigrants
Age	42.17	34.21
Age	(12.13)	(10.26)
Income
Wage Earnings	49,329	34,531
Wage Earnings	(50,661)	(45,263)
Hourly Wage	24.67	18.57
Hourly Wage	(69.79)	(40.01)
Education
Primary or Less	0.01%	6.23%
<High School^a	7.90%	26.45%
High School	31.58%	23.18%
Some College	22.19%	8.80%
Associate’s Degree	7.91%	3.24%
Bachelor’s Degree	19.96%	16.43%
Master’s Degree	7.19%	9.79%
Advanced Degree	3.23%	5.92%
Observations	5,199,240	192,441

Note. All wage figures are converted into 2005 dollars. Standard deviations are in parentheses.

The less than high school category includes respondents who have education levels between primary school and high school. The category of “Advanced Degree” includes doctoral and professional degrees.

The sample includes 5.2 million natives and 192,411 immigrants. The average age of an immigrant in the sample is 34 years, whereas the average age of a native in the sample is 42 years. It can be argued that the difference in average ages of immigrants and natives may limit the meaningfulness of the earnings-differential analysis between these two groups. But as detailed in the next section, this study employs decomposition techniques that account for differences in the characteristics between the two groups.

Table 1 shows that natives have a marked earnings advantage over immigrants. The annual wage earnings of natives are on average US$14,798 more than those of immigrants. Immigrants also earn about 25% (US$6.10) less in hourly wages than natives. Natives are also more educated than immigrants. The differences between the educational attainment of immigrants and natives are particularly large at lower levels of education. A total of 6.23% of the immigrants have not attended secondary school compared with only 0.01% of natives. However, the differences between immigrants and natives decrease for higher levels of education. In fact, a greater proportion of immigrants have a master’s or advanced degree.

Table 2 details English-language proficiency and citizenship status of immigrants. The table shows that the majority of immigrants do not have U.S. citizenship. With regard to English-language proficiency, 21,111 immigrants, or 11% of the immigrant sample, report to have the highest of the five levels of English proficiency. By contrast, 32,870 immigrants report to have no proficiency in English, and 49,355 immigrants report their English proficiency as “barely able to speak” or Poor. The large number of immigrants with little or no proficiency in English is not surprising given the fact that about 50% of immigrants in the sample are from Mexico, Central America, or South America, as shown in the appendix. According to Borjas (2015), recent Hispanic immigrants have a lower incentive to learn English because of a large Hispanic population in the United States.

Table 2.

Citizenship and English-Language Proficiency of Immigrants.

	English-language proficiency					Total
	None	Poor	Well	Very well	Native	Total
Citizenship Status
Naturalized Citizen	525	2,684	5,106	7,126	2,617	18,058
Not a Citizen	32,303	46,520	34,612	41,341	17,313	172,089
Missing Information	42	151	274	646	1,181	2,294
Total	32,870	49,355	39,992	49,113	21,111	192,441

Note. Respondents self-report one of the 5 language-proficiency levels, and, hence, this measure is likely to suffer from perception bias.

Empirical Approach

The empirical strategy is based on the following model:

\begin{array}{l} \ln (w a g e_{i t}) = α + β I_{i t} + + γ' D S t a t e_{i t} + δ A g e_{i t} \\ + ζ^{'} D R a c e_{i t} + θ^{'} D C o h o r t_{i t} + κ^{'} D E d u c_{i t} \\ + λ' D E n g_{i t} + μ' D O c c_{i t} + η_{t} + ε_{i t}, \end{array}

(1)

where subscript i refers to an individual and t to a survey year. In Equation 1, wage denotes real hourly wage income. The dummy variable I is 1 if an individual is an immigrant. DState denotes a vector of 51 indicator variables that specify the state in which individual i resides at time t.⁴ Vector age consists of two continuous variables: age and age squared. Drace is a vector of six race indicator variables. DCohort is a vector of 12 indicator variables, each representing a year of immigration. DEduc is a vector of eight binary variables that represents various levels of education. Vector DEng is a vector of five dummy variables, each representing a different level of English-language proficiency. DOcc is a vector of 22 0/1 indicator variables that account for various occupations. Survey-year fixed effects are identified by $η_{t}$ . In Equation 1, the β coefficient represents the wage (dis)advantage associated with being an immigrant.

The studies of Oaxaca (1973) and Blinder (1973) note that differences in the earnings of two groups can arise because of (a) differences in the endowments of certain characteristics, (b) differences in the returns to these characteristics, (c) interactions between differences in endowments and returns to endowments. Following Jann (2008), the threefold decomposition of the earnings disadvantage of immigrants can be written as,

D i s a d v a n t a g e = ({\bar{X}}^{N} - {\bar{X}}^{I}) {\bar{τ}}^{I} - {\bar{X}}^{I} ({\bar{τ}}^{N} - {\bar{τ}}^{I}) + ({\bar{X}}^{N} - {\bar{X}}^{I}) ({\bar{τ}}^{N} - {\bar{τ}}^{I})

(2)

where Disadvantage represents the average earnings disadvantage of immigrants; superscript N represents natives, and superscript I represents immigrants; $\bar{X}$ represents the mean characteristics, and $\bar{τ}$ is the mean return to the characteristics. The first term on the right-hand side represents the difference between the earnings of natives and immigrants that results from the gap in mean characteristics of the two populations. The second term represents the difference between the earnings of natives and immigrants because of the differences in the mean returns to the characteristics. The third term quantifies the contribution of interactions of the differences in endowments and the differences in returns.

Immigrants have different levels of endowments of various characteristics compared with natives. They are also highly likely to have different returns to various characteristics. Many immigrants received education in their country of origin, which is often of a lower quality than American education. Even if some of the education is gained in the United States, the certificate may not translate into higher productivity in the same way as it does for natives. This can be due to the lack of unobservable characteristics, such as job search skills or cultural awareness, in immigrants. In the same way, age may have different returns for natives and immigrants. To capture the potentially different effects of such variables for immigrants, we interact the immigrant dummy variable with the age, education, and race variables, and add these interactions as new variables to Equation 1,⁵

\begin{array}{l} \ln (w a g e_{i t}) = α + β I_{i t} + γ^{'} D S t a t e_{i t} + + δ A g e_{i t} + ζ^{'} D R a c e_{i t} + θ^{'} D C o h o r t_{i t} + \\ κ^{'} D E d u c_{i t} + λ' D E n g_{i t} + μ' D O c c_{i t} + Π + η_{t} + ε_{i t} \end{array}

(3)

where : Π = δ_{1} (I \times a g e_{i t}) + {ζ^{'}}_{1} (I \times D R a c e_{i t}) + {κ^{'}}_{1} (I \times D E d u c_{i t})

Equation 3 takes account of the differences in endowment levels of characteristics and includes interaction terms of some of the characteristics with the immigrant dummy variable. We estimate numerous variants of Equation 3 to decompose the earnings disadvantage of immigrants. As Equation 3 includes many interaction terms that include the immigrant dummy, β no longer represents the earnings disadvantage of an immigrant. To account for this, we partially differentiate Equation 3 with respect to I and use the resulting expression to estimate the overall earnings disadvantage.

Results and Discussion

Decomposition of Wage Disadvantage of Immigrants

As detailed in Equation 2, the total earnings disadvantage can be divided into three sources: (a) differences in endowments, (b) differences in returns to endowments, and (c) interactions of differences in endowments and returns. Using the control variables of Equation 1 and the Blinder-Oaxaca algorithm written by Jann (2008), we estimate the earnings disadvantage of immigrant men to be 36.2%.⁶ The results from the Blinder-Oaxaca decomposition suggest that exactly half of the 36.2% disadvantage is due to differences in endowments; only 2.7 of the 36.2% disadvantage is due to the differences in the returns (coefficients) to immigrants’ characteristics. The remaining 15.4 of the 36.2% wage disadvantage comes from the interaction of differences in return and endowment levels. Equation 3, which accounts for differences in both returns and endowments of immigrants and natives, is estimated to parse out the effects of various characteristics on the earnings differential between immigrant and native men.

Table 3 presents the estimation results for variants of Equation 3. Only the estimates for the earnings disadvantage of immigrants are shown. The specification in the first column of Table 3 controls only for the time fixed effects of the survey years and the state identifiers; each subsequent column adds additional controls until all the variables specified in Equation 3 are included. The idea is to note the change in the earnings disadvantage as additional controls are added. The change associated with each additional set of controls is interpreted as its contribution to the overall earnings disadvantage.

Table 3.

Estimation Results for Variants of Equation 3.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
Earnings Disadvantage	−0.400	−0.250	−0.238	−0.042	−0.027	−0.031	−0.008
Earnings Disadvantage	(0.002)	(0.002)	(0.002)	(0.015)	(0.010)	(0.009)	(0.009)
R ²	.038	.165	.182	.183	.306	.359	.360
Controls
Census Year	X	X	X	X	X	X	X
State	X	X	X	X	X	X	X
Age & Age × I		X	X	X	X	X	X
Race & Race × I			X	X	X	X	X
Year of Immigration				X	X	X	X
Education & Education × I					X	X	X
Occupation						X	X
English Proficiency							X

Note. Dependent Variable: Log of Hourly Wages. All regressions use 5,391,681 individuals. Standard errors are in parentheses.

All coefficients are significant at p < .01 level.

Column 1 of Table 3 identifies the wage disadvantage of immigrants to be about 40%. Column 2 adds the control variables of age, age squared, and their interactions terms with the immigrant dummy variable. The addition of these variables reduces the earnings disadvantage to 25%. This suggests that a large part of the earnings disadvantage is attributable to the lower average age of immigrants compared with natives. Older individuals tend to have more work experience when compared with the younger individuals. More work experience leads to greater accumulation of unobservable job-specific skills and soft-skill, such as communication, critical thinking, and team-working skills, which is reflected in productivity and earnings.

The third column adds race dummy variables and their interaction terms with the immigrant dummy variable. This absorbs 1.2 percentage points of the earnings disadvantage. The fourth column adds the cohort dummies to the specification, which controls, at least in theory, for differences in the quality of different cohorts of immigrants. However, if one can assume that the cohort quality does not vary substantially over the brief period of 13 years, the cohort dummies primarily control for the number of years an immigrant has been in the United States. After the cohort dummies are added to the model, the estimate of the earnings disadvantage reduces from 23.8% to 4.2%. This suggests that about 20 percentage points of the earnings disadvantage experienced by immigrants is associated with the number of years an immigrant has been in the United States.

The fifth column adds education level dummies and their interactions with the immigrant dummy variable. The addition of these variables reduces the earnings disadvantage to 2.7% or about US$1,300 a year for a full-time immigrant worker. This suggests that 1.5 percentage points of the earnings disadvantage of immigrants is associated with their educational attainment. Column 6 adds occupation dummies to the model. This changes the earnings disadvantage of immigrants very little and indicates that occupational choice is not a key driver of the earnings disadvantage of immigrants. The final column of the table adds dummy variables that represent various levels of English-language proficiency to the specification. Addition of these variables reduces the estimate of earnings disadvantage to less than 1% and makes it statistically insignificant. This suggests that somewhat less than 3 percentage points of the earnings disadvantage of immigrants is attributable to their lower level of English-language proficiency.⁷

The results in Table 3 show that the time spent in the United States, English-language proficiency, and the educational attainment are the main predictors of the earnings disadvantage of immigrant men. The contribution of the time spent in the United States toward the earnings disadvantage of 21st century immigrants is substantially larger than the contribution of English-language proficiency and educational attainment.

The magnitude of the effect of time spent in the United States, however, should not be surprising as time spent in the United States allows immigrants to learn U.S. work-culture and earn work experience in the United States that makes them more marketable in the U.S. job market. The immigrants of the 21st century also face a large earnings disadvantage as compared with their native peers because of their low levels of English-language proficiency. A large proportion of the 21st century immigrants is Hispanic. Like their 20th century counterparts, as documented by McManus, Gould, and Welch (1983), they experience a substantial earnings disadvantage because of their limited proficiency in English. The relatively smaller contribution of educational attainment to the earnings disadvantage is discussed in the next section, which examines the returns to different levels of education.

Breakdown of the Returns to Human Capital

In this section, we present the returns to different levels of two important human capital variables, namely, English-language proficiency and educational attainment. We first present this analysis using the full sample, as in the analysis so far, and then focus on the three subsamples of the full sample.

The 21st century started with the dotcom stock market crash, rising interest rates, and the 9/11 terror attack. The result was the recession of 2001. This was followed by strong economic growth until the financial crisis of 2007/2008. The 2008-2009 recession had far-reaching implications for the U.S. economy, especially for the construction and financial sectors of the economy. Since the end of the recession, the economy has experienced modest growth, and the labor market has been improving. Given the volatility in the U.S. economy and substantial changes in many sectors over the first 13 years of the 21st century, it appears useful to study the returns to English-language proficiency and educational attainment over different periods of the 21st century. To this end, we create three subsamples: a prerecession period from 2001 to 2007, the recession period from 2009 to 2010, and the postrecession period from 2012 to 2013.⁸

Respondents report their wage earnings from the 12 months preceding the interview. This means that a person who gets surveyed in January 2010 reports earnings of 2009, and a person who gets surveyed in December 2010 reports earnings that are mostly earned in 2010, while both individuals appear in the 2010 round of the Census data. Hence, I use the years 2009-2010 instead of the years 2008-2009 to ensure that the effect of the recession is accurately captured. The years 2008 and 2011 are dropped from this analysis, to avoid the noise generated by the survey methodology that makes them the transition years “into” and “out of” recession, respectively.

The first column of Table 4 presents the results for the full sample; the following columns cover the three subsamples. The first column of Table 4 shows that individuals who have less than a high school education earn 12.4% less than high school graduates, which is the omitted category in the specification. Interestingly, the wage disadvantage associated with this level of education is 7 percentage points less for immigrant men, as shown by the coefficient of the interaction term of the immigrant dummy variable with the dummy variable of “less than high school” level of education. Borjas and Friedberg (2009) attribute this anomaly to the fact that the recent low-skilled immigrants tend to work in well-paid, low-skilled occupations such as construction. The estimates for the full sample also show that individuals with a bachelor’s degree enjoy a substantial advantage over individuals with only high school qualification. For immigrants, the return to a bachelor’s degree is about 3.4 percentage points lower than that for natives. An immigrant holding a master’s degree earns 6.6% more than a native with a master’s degree. However, for advanced education levels, which include doctoral degrees, immigrants earn about 22 percentage points less than natives.

Table 4.

Pre- and Postrecession Returns to Education and Language Proficiency.

	All 13 years	Prerecession	Recession	Postrecession
Education
<HS	−0.124**	−0.122**	−0.130**	−0.126**
<HS	(0.001)	(0.002)	(0.003)	(0.003)
<HS × I	0.070**	0.080**	0.075**	0.065**
<HS × I	(0.004)	(0.007)	(0.009)	(0.008)
Bachelor	0.366**	0.369**	0.367**	0.367**
Bachelor	(0.001)	(0.001)	(0.002)	(0.002)
Bachelor × I	−0.034**	−0.039**	−0.025*	−0.032**
Bachelor × I	(0.005)	(0.009)	(0.010)	(0.009)
Master’s	0.515**	0.506**	0.526**	0.528**
Master’s	(0.001)	(0.002)	(0.003)	(0.003)
Master’s × I	0.066**	0.039**	0.061**	0.086**
Master’s × I	(0.006)	(0.011)	(0.012)	(0.010)
Advanced	0.692**	0.675**	0.708**	0.719**
Advanced	(0.002)	(0.002)	(0.004)	(0.004)
Advanced × I	−0.220**	−0.286**	−0.225**	−0.168**
Advanced × I	(0.007)	(0.013)	(0.014)	(0.013)
English Proficiency
None	−0.259**	−0.239**	−0.251**	−0.271**
None	(0.004)	(0.007)	(0.010)	(0.009)
Poor	−0.205**	−0.175**	−0.213**	−0.221**
Poor	(0.003)	(0.005)	(0.007)	(0.007)
Well	−0.127**	−0.104**	−0.137**	−0.144**
Well	(0.002)	(0.004)	(0.005)	(0.005)
Very Well	−0.052**	−0.050**	−0.052**	−0.051**
Very Well	(0.001)	(0.002)	(0.003)	(0.003)
R ²	.360	.336	.372	.383
Observations (000s)	5,392	2,339	1,015	1,021

Note. Standard errors are in parentheses. HS = high school.

p < .01. **p < .001. Dependent Variable: Log of Hourly Wage.

The higher returns to a master’s degree for immigrants are consistent with the fact that a substantially greater proportion of immigrants holds a master’s degree in science, technology, engineering, and mathematics (STEM) fields as compared with natives (National Science Foundation, 2014). The individuals that hold degrees in STEM fields earn substantially more than the non-STEM degree holders (Langdon, McKittrick, Beede, Kahn, & Doms, 2012). Furthermore, the proportion of foreign students in graduate STEM programs in the United States is also significantly higher than the proportion of native students (Anderson, 2013). Hence, it is likely that immigrants with a master’s in STEM fields are being compared with natives with master’s in non-STEM fields, which results in immigrants with master’s degrees to exhibit a large earnings advantage over natives.

The results for the full sample also show that advanced degrees earn a significantly lower return for immigrants than for natives. This may be surprising given that National Science Foundation (2014) also reports that a higher proportion of immigrants hold advanced degrees in STEM fields. One possible explanation is that a majority of immigrants who hold advanced degrees may have earned these degrees outside the United States, which are often valued less than American degrees (Chiswick & Miller, 2009).

As noted in the previous section, the contribution of the educational attainment to the earnings disadvantage of 21st century immigrants is relatively small as compared with the contribution of years in the United States and English-language proficiency. The interactions of the education-level dummy variables with the immigrant dummy variables help in explaining this point. The results show that the returns to less than a high school level of education are higher for immigrants than for natives. Given that a large proportion of immigrants is educated below the high school level, the advantageous returns for this subset of the immigrant sample cause the total effect of education on the earnings disadvantage to be limited.

However, it should be noted that this anomaly may not necessarily be desirable. If the goal is to have a well-educated and highly skilled workforce, a greater proportion of immigrants with higher levels of education and with an earnings disadvantage in comparison to natives of comparable education may be more desirable than a large, low-skilled population of immigrants who derive higher returns from low levels of education. The results for English-language proficiency show that individuals with no proficiency earn about 26% less than individuals with native-level proficiency, the omitted category. This disadvantage reduces to 5% for immigrants who report the ability to speak English “very well.”

The prerecession, recession, and postrecession analysis also shows some interesting trends. The estimates suggest that higher levels of education generate higher returns in the postrecession period as compared with the prerecession period, whereas less-educated individuals face a greater degree of disadvantage in the labor market in the postrecession period as compared with the prerecession period. For example, in the postrecession period, a master’s degree generates a premium of 52.8% over the returns on a high school diploma.⁹ This premium is about 2.2 percentage points higher than the corresponding premium for the prerecession period. A more interesting finding is that the changes in returns are amplified for immigrants. For natives, a less than high school education earns a 0.4 percentage points lower return in the postrecession period compared with the prerecession period. For immigrants, the corresponding change is 1.9 percentage points.¹⁰ By contrast, an advanced degree generates a 4.40 percentage points higher return in the postrecession period for natives, but a 16.20 percentage points higher return for immigrants, both compared with the corresponding returns in the prerecession period.

Conclusion

This article makes two key contributions to the existing literature. First, it estimates and provides a decomposition of the earnings disadvantage of men who have immigrated to the United States since the turn of the century. Second, it highlights the trends in returns to two key human capital characteristics, English-language proficiency and education, over different time periods of the 21st century.

The empirical analysis uses a comprehensive set of characteristics to decompose the earnings disadvantage. Following the Blinder-Oaxaca framework of decomposing an earnings disadvantage, it also includes interaction terms of the immigrant dummy variable with key individual characteristics. The results show that controls for age and years in the United States explain large portions of the earnings disadvantage of immigrant men. The article finds English-language proficiency to be responsible for about 3 percentage points of the earnings disadvantage, while educational attainment is found to be keeping immigrants behind their native counterparts by about 1.5 percentage points. The relatively smaller contribution of education to the earnings differential of immigrants with comparable natives is at odds with the widely held belief that education is the main driver of the earnings disadvantage of immigrants.

The breakdown of the returns to education shows that immigrants with lower levels of education are at a wage advantage over comparable natives. This advantage disappears for higher levels of education, except for immigrant men with a master’s degree. The analysis of changes in returns to English-language proficiency and educational attainment, over different periods of the 21st century, shows that individuals with low levels of these human capital attributes face a greater earnings disadvantage in the postrecession period. The results also show that, for immigrants, lower levels of these human capital attributes are penalized to a greater extent, as compared with natives, in the postrecession period. By contrast, the positive changes in the returns to higher levels of education are larger for immigrant men than for native men, from the prerecession period to the postrecession period.

The results presented in this article can guide the policy discussion on selective immigration. The high returns to the interrelated variables of age and years in the United States suggest that employers value traits such as U.S.-specific work experience and cultural awareness. This suggests that an immigration policy that attracts younger people might help close the earnings gap between immigrants and natives. With regard to English-language proficiency, the United States could follow the example of some other developed countries that have added a reasonable level of English-language proficiency to the list of eligibility requirements for various types of visas. The contribution of educational attainment to the earnings disadvantage of immigrants is relatively small, but its importance for broader economic goals is not. Highly educated immigrants may earn less than highly educated natives, but they contribute much more than low-skilled immigrants to the fiscal strength of the U.S. government, the social security system, and in driving innovation in the United States. Given the importance of an educated work force, an immigration policy that favors highly educated individuals, especially those who are educated in the United States, may increase the benefits from immigration.

Finally, it needs to be noted that “earnings of immigrants” is only one of the many aspects of the debate on immigration policy. The findings of this article highlight the merits of selective immigration but do not endorse a policy solely based on economic considerations. An appreciation of the reasons behind people’s decision to migrate can help in formulation of a comprehensive immigration policy. In recent years, millions of people have been displaced because of war, violence, persecution, and natural disasters (United Nations High Commissioner for Refugees [UNHCR], 2016). Some of these people have applied to enter the United States as refugees, asylees, or have crossed the border illegally (Center for American Progress Immigration Team and Michael D. Nicholson, 2017; UNHCR, 2016). The United States has ratified United Nations 1967 Protocol relating to the status of refugees, which, among other rights, gives refugees the right of not being sent back into harm’s way under most circumstances (Fitzpatrick, 1997). Comprehensive immigration reform will not only be informed by the economics of immigration, but also by the U.S. values of life, liberty, pursuit of happiness and justice, along with its commitments to the international community.

Footnotes

Appendix

Country/Region of Origin.

Country/region of origin	Observations	Percentage
Mexico	65,734	34.16
India	19,311	10.03
Central America	18,924	9.83
South America	13,223	6.87
Africa	9,369	4.87
West Europe	8,162	4.24
China	7,779	4.04
Philippines	7,650	3.98
West Indies	7,171	3.73
West Asia (mainly Arab)	5,488	2.85
Cuba	4,171	2.17
United Kingdom	4,121	2.14
East Europe	3,986	2.07
Canada	3,446	1.79
Russia	3,392	1.76
Korea	2,760	1.43
Vietnam	2,547	1.32
Japan	2,476	1.29
East Asia	1,640	0.85
Australia and New Zealand	1,091	0.57
Total Immigrants	192,441	100.00

Note. Source countries that sent relatively small number of immigrants have been lumped into regions.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Fahad Gill is an assistant professor of economics at West Virginia University (Tech.). He is an applied microeconomist with research interests in the areas of labor economics, development economics, and financial economics. He is particularly interested in labor market outcomes of minorities in the United States, and rural population in South Asia.

Waseem Ahmad is an assistant professor of business management sciences, at the University of Agriculture, in Faisalabad, Pakistan. He is an applied economist with research focus on agribusiness and marketing. He is also interested in issues of immigration and emigration.

References

Anderson

(2013, July). The importance of international students to America (National Foundation for American Policy, Policy Brief). Retrieved from http://www.nfap.com/pdf/

Blinder

A. S.

(1973). Wage discrimination: Reduced form and structural estimates. Journal of Human Resources, 8, 436-455.

Borjas

G. J.

(1985). Chicago assimilation, changes in cohort quality, and the earnings of immigrants. Journal of Labor Economics, 3, 463-489.

Borjas

G. J.

(1995). Assimilation and changes in cohort quality revisited: What happened to immigrant earnings in the 1980s? Journal of Labor Economics, 13, 201-245.

Borjas

G. J.

(2015). The slowdown in the economic assimilation of immigrants: Aging and cohort effects revisited again. Journal of Human Capital, 9, 483-517.

Borjas

G. J.

Friedberg

R. M.

(2009). Recent trends in the earnings of new immigrants to the United States. Cambridge, MA: National Bureau of Economic Research, Working Paper No. 15406. Retrieved from: http://www.nber.org/papers/w15406.

Center for American Progress Immigration Team and Michael D. Nicholson. (2017). The facts on immigration today: 2017 edition. Retrieved from https://www.americanprogress.org/issues/immigration/reports/2017/04/20/430736/facts-immigration-today-2017-edition/

Chellaraj

Maskus

K. E.

Mattoo

(2008). The contribution of international graduate students to U.S. innovation. Review of International Economics, 16, 444-462.

Chiswick

B. R.

(1978). The effect of Americanization on the earnings of foreign-born men. Journal of Political Economy, 86, 897-921.

10.

Chiswick

B. R.

Miller

P. W.

(2009). The international transferability of immigrants’ human capital. Economics of Education Review, 28, 162-169.

11.

Duleep

H. O.

Regets

M. C.

(2002). The elusive concept of immigrant quality: Evidence from 1970-1990. Bonn, Germany: IZA Institute of Labor Economics Discussion Paper No. 631. Retrieved from: ftp://repec.iza.org/RePEc/Discussionpaper/dp631.pdf.

12.

Fitzpatrick

(1997). The international dimension of US refugee law. Berkeley Journal of International Law, 15, 1-26.

13.

Funkhouser

Trejo

S. J.

(1995). The labor market skills of recent male immigrants: Evidence from the current population survey. Industrial & Labor Relations Review, 48, 792-811.

14.

Jann

(2008). The Blinder-Oaxaca decomposition for linear regression models. The Stata Journal, 8, 453-479.

15.

Krogstad

J. M.

Passel

J. S.

Cohn

(2017). 5 facts about illegal immigration in the U.S. Retrieved from http://www.pewresearch.org/fact-tank/2017/04/27/5-facts-about-illegal-immigration-in-the-u-s/

16.

Langdon

McKittrick

Beede

Kahn

Doms

(2012). STEM: Good jobs now and for the future. Retrieved from http://www.esa.doc.gov/reports/stem-good-jobs-now-and-future

17.

Lee

Miller

(2000). Immigration, social security and broader fiscal impact. AEA Papers and Proceedings, 90, 350-354.

18.

McManus

Gould

Welch

(1983). Earnings of Hispanic men: The role of English language proficiency. Journal of Labor Economics, 1, 101-130.

19.

Migration Policy Institute. (2017). U.S. immigration trends. Retrieved from http://www.migrationpolicy.org/programs/data-hub/us-immigration-trends

20.

National Science Foundation. (2014). Science and engineering indicators. Retrieved from https://www.nsf.gov/statistics/seind14/index.cfm/digest.

21.

Oaxaca

(1973). Male-female wage differentials in urban labor markets. International Economic Review, 14, 693-709.

22.

Ruggles

S. J.

Alexander

Genadek

Goeken

Schroeder

M. B.

Sobek

(2015). Integrated Public Use Microdata Series (Version 5.0) [Machine-readable database]. Retrieved from https://usa.ipums.org/usa/

23.

Storesletten

(2000). Sustaining fiscal policy through immigration. Journal of Political Economy, 108, 300-323.

24.

United Nations High Commissioner for Refugees. (2016). Global trends: Forced displacements in 2015 (Rep.). Geneva, Switzerland: UNHCR, The UN Refugee Agency.

25.

U.S. Bureau of Labor Statistics. (2014). Local area unemployment statistics. Retrieved from http://www.bls.gov/lau/

26.

Zong

Batalova

(2017). Frequently requested statistics on immigrants and immigration in the United States. Retrieved from http://www.migrationpolicy.org/article/frequently-requested-statistics-immigrants-and-immigration-united-states