The idea of America as a melting pot is predicated on this country’s unique ability to absorb peoples from all regions of the world and turn them into prosperous Americans. In a large measure, the success of the melting pot effect is determined by the relative wealth and upward mobility of the immigrant. It is well understood there is a strong link between earnings and income but does that link also apply to immigrants? This analysis looks at the returns to education based upon the immigrant’s country of origin and the year of immigration. The intention of this analysis is the attempt to ascertain whether a wage premium or deficiency exists for educated immigrants as a whole as well as across specific regions of the globe.
The effects of the returns to educations have been studied extensively across a wide cross section of categories and subjects including, career paths, age groups, immigrant status, country of origin and geographic locations to name but a few. Several studies focus exclusively on geographic regions. Two such studies, recently reviewed, focused on the geographic region of the United Kingdom. One study, by Kevin Denny & Vincent O’Sullivan in the journal Applied Economic Letters, looked at whether returns to education can compensate for low ability while the other, by Mary Silles in the Journal of Applied Economics, looked more generally at the returns to education for the UK. Both studies utilized time series data from a specific geographic region and regression analysis while attempting to provide evidence in support of their argument. Both articles showed that there were positive returns to education across all segments of UK society, however The Silles article found a slight decline for women and younger workers have come to experience unequal returns to education. The Denny & O’Sullivan study finds that education can indeed be a substitute for ability but only for those with low ability; therefore the returns to education are high for those with low ability and low for those with high ability. Both articles utilize data from a specific geographic region, the UK, both articles find that the returns to education are clearly positive.
Looking at the U.S. there have been numerous studies that address a uniquely American issue; immigration. As a nation of immigrants, U.S. policy makers and economists have found it to be of critical importance to understand the root causes of wage disparity between 1st, 2nd and 3rd generation immigrants and the native born population. Two such studies recently reviewed focused on educational obtainment as a root cause of wage disparity. The first study by the RAND Corporation, titled Immigration in a Changing Economy focused on the issue of educational obtainment as a critical factor in the success of immigrants. Chapter 4 of the article deals with this issue specifically and is titled The Success of Immigrants Increasingly Depends on Their Education. Using detailed California immigration data this chapter first looks to compares educational obtainment level across different immigrant segments broken out by country of origin. The study looks at the earnings of each immigrant segment while comparing both earnings and educational obtainment to the native born population. The study finds that educational obtainment levels do rise substantially from 1st generation immigrants to 2nd generation reaching near parity with the 3rd generation. The study also finds that the wage disparity between 1st generation immigrants and the native born population narrows further if the immigrant increases their human capital, particularly through the acquisition of English language training. Even here however a large difference is evident, immigrants with a higher level of education are shown to pick up English quicker and thus realize earnings gains faster. The study finds substantial wage disparities exist not only between natives and immigrants but also between immigrants of different national origin. The study is able to show that this wage disparity between immigrant groups can be attributed to different levels of education between immigrant groups. This is important for two reasons; first those immigrants with higher levels of education become fluent in English faster, thus realizing wage gains quicker, and second, the gains compound overtime ensuring that immigrants with a higher level of education continue to outpace those without a higher level of education throughout their lives. The study also deals with the 2nd and 3rd generation immigrants. The study finds that the 2nd and 3rd generation immigrants from parents with higher levels of education achieve parity in wage and human capital levels faster. The example cited shows a persistent reduction in both educational obtainment and earnings potential for Hispanics of Mexican origin, which have the lowest average rates of educational obtainment among all immigrants groups, continues through to the 3rd generation.
The second article by the Dallas Federal Reserve Bank, titled, Immigrant Assimilation: Is the U.S. Still a Melting Pot, deals with educational obtainments role in an immigrant successful assimilation into U.S. society. The study considers economic achievement to be successful assimilation, and like the RAND Corporation study, looks at the wage disparity that exists between immigrants and the native born population. The study uses data from the U.S. Census Bureau, Current Population Survey from March of 2002 on immigrant education levels and income. The study finds that a high percentage of immigrants come to the U.S. with low levels of educational obtainment. Immigrants of Mexican origin come with particularly low levels of educational obtainment. The study finds that roughly 32% of all foreign born immigrants have less than a high school diploma with another 33% having just a high school diploma. This general lack of educational obtainment coupled with structural changes in the U.S. labor market, such as a move up the technology and product value chain, has lead to a considerable wage gap between new immigrants with low levels of educational obtainment and the native born population. In a reverse of the overall trend, a higher percentage of immigrants have advanced degrees than the native born population, in these cases there does not appear to be a wage disparity. The study shows that immigrants who have been in the U.S. between 0-5 years earn close to 40% of what the native born population earns. This rises steadily the longer the immigrant resides in the U.S. eventually reaching 70% of native born population income by the 15th year.
The intention of this analysis is to look at the effect of education on the income of immigrants based upon country of origin and year of immigration. The purpose of this analysis is to ascertain how educated immigrants have done financially with respect to the native born population with similar levels of educational attainment. This analysis will look to find whether there is a wage premium or deficiency for educated immigrants versus the native population. This regression analysis will only be looking individuals with an educational attainment above the high school level. This regression specifically excluded any immigrant not classified to have had at least some college education. The principle variables relevant to this regression analysis are country of origin, year of immigration and education and how they compare to the native born population. As this regression is only concerned with educational attainment levels above high school, the inclusion of the country of origin and year of immigration variables holding education constant will allow the regression model to identify clearly the effects of country of origin and year of immigration on income.
Structural changes experienced in the labor market over the last several decades have lead to a global competition for highly skilled labor. The transformation of many of the world’s advanced economies from a primarily industrial economic structure to a service based economy has driven increased demand for human capital. The transformation of the advanced economies of the world has coincided with a technological revolution which has dramatically increased global commerce and mobility. The confluence of these trends has increased the demand for human capital which has in turn created a wage premium for educated labor. It is this author’s hypothesis that these long term trends coupled with a shortage of talented individuals has created a wage premium for educated immigrants with respect to a similarly educated native born population. This hypothesis is rooted in the assumption that in a globally competitive labor market where talented and educated individuals are recruited globally, educated immigrants will realize a wage premium over the native born population as the benefit of immigration. Holding education constant across the whole sample with log natural of wage & salary (INCWAGE) as my dependent variable and year of immigration and country of origin as my two principle independent variables, this analysis looks to show that there is statistical evidence of a wage premium for immigrants with respect to the native born population. The method devised to showcase the transformation in the global economy, which is this author’s hypothesis as to why a wage premium has developed for educated immigrants with respect to the similarly educated native born population, is the year of immigration binary variable, which provides the author with the ability to test whether the wage premium has always existed or if it is a more recent outgrowth of the transformation of the world’s advanced economies.
This regression analysis utilized continuous variables, age (AGE), wage & salary variable (INCWAGE), birth place of origin (BPL). This regression also utilized binary variables, region of the U.S. (REGION), educational attainment levels (EDUC99), sex (SEX), year of immigration (YRIMMIG), and marital status (MARST). The dependent variable is wage & salary (INCWAGE), the wage & salary variable was computed utilizing the log of INCWAGE. The independent variables were, age, region of the U.S., sex, educational attainment levels, race, marital status, country of origin and year of immigration. To achieve greater clarity of results, additional binary variables were created grouping year of immigration into decades. These binary variables include year of immigration pre 1950, and every decade from 1950 on until 1990 -1999 with 2000 – 2008 as my reference variable. In addition to grouping year of immigration by decade the country of origin is grouped by region of the world with the U.S. as the reference variable. This allows the regression to obtain an aggregate view of the educational and income trends of immigrants from a broad cross section of the globe.
The data used to this author’s hypothesis was compiled by U.S. Census’ Current Population Survey as collected and integrated by the University of Minnesota’s IPUMS-CPS project. The data includes all variables listed in the above paragraph. Income (INCWAGE) is measured as the median household income, where as educational attainment levels (EDUC99) are measured as percentage of the population with either some college, a bachelor’s degree or a post grad degree. Utilizing OLS regression techniques with log natural of (INCWAGE) as my dependent and year of immigration (YRIMMIG), country of origin (BPL), educational attainment levels (EDUC99) as my independent variables of interest allows this author to test, at the micro level, this paper’s hypothesis. Utilizing the IPUMS-CPS/U.S. Census’ Current Population Survey data I have provided a table below with descriptive statistics. The table below displays the national average for continuous variables, age, salary (INCWAGE) not in log form, weeks worked and hours worked, for my restricted model which includes only educational levels of some collage and above. Additionally, the descriptive statistics show, in percentages, the population’s educational obtainment levels, region of the U.S., race/ethnicity, as well as the global regional origin and decade immigrated.
Table 1: Descriptive Statistics for the Sample Size (N) of 59,501
The above table shows that the median household income for the sample is $50,505 with average hours worked per week of 39.91 and a work year comprised of 47.74 weeks. Nearly 50% of the sample fall into the some collage category, with 33% holding a BA and the remainder with post grad degree. The sample is made up of 51.5% female and 62.3% are married. The sample is spread evenly across the U.S. and the ethnic make up mirrors the U.S. population as a whole. When looking at the regional origin (BPL) and the decade immigrated a clear pattern emerges. Immigration across all categories, barring Western Europe, shows clear growth, with Latin American and Asian immigration showing very strong growth through the 70s and 80s with a decline in the 90s. Educated immigration from Asia in the 1980s was highest with almost a full 1% of all immigrants arriving holding an educational attainment level of some college or better. Surprisingly immigrants from the Subcontinent region were as a percentage of immigrants with some college or better quite low, which clearly disproves the common myth of the super educated Indian immigrant. Immigrants from the South America region came with the lowest levels of educational obtainment across all categories in this study.
The purpose of this analysis was to ascertain whether there exists a wage premium for educated immigrants with respect to the native born population. In other words is there an immigration premium for immigrants with high human capital. The 3 regression models below provide this author with the ability to test the effect of the inclusion of additional binary variables. The results of my regression analysis can be found in the table below titled OLS Multiple Regression Analysis of Earnings as Dependent Variable.
Table 2: Empirical Results for the Test of the Immigrant Wage Premium Theory
Dependent Variable: Log Natural of Median Household Income
Model 1 shows the effect on LN (INCWAGE) of the binary variables age, weeks worked, hours worked, racial ethnicities, gender and region of the U.S. in which the individual lives without screening for educational obtainment. Model 1 shows that the LN INCWAGE is 6.332. The data shows that there is a strongly negative relationship on income for all ethnicities other than Caucasian and Asian. The data in model 1 also shows a strongly negative relationship between income and gender when the gender is a female. Regionally, model 1 shows that there is a slight income premium for those that reside in the Northeast with a deficit for those residing in the Midwest and South. Model 2 adds educational obtainment to the regression. The addition of educational obtainment to the regression model sees the value of the constant fall slightly from Model 1. Overall the inclusion of the educational obtainment in Model 2 sees all values from Model 1 decline with the exception of the effects of being an of ethnic origin. The inclusion of the educational obtainment variable decreases the negative effect on income of being of an ethnic origin for all cases with the exception of Asians. In Model 1 being of Asian ethnicity had a positive effect on income whereas in Model 2, with the inclusion of educational obtainment variables, being of Asian ethnicity has a negative effect on income. Model 2 sees a strongly positive effect on income of having either a four year degree or post grad degree.
Model 3 includes all of the same variables as Model 2 with the addition of the global regional origin and decade of immigration variables. In Model 3 the value of the constant reaches its maximum for all 3 models. This clearly shows that the inclusion of the two new variables has had a positive effect on income. In model 3 the age variable is at its minimum as is weeks worked, while hours worked remains unchanged from model 2. The inclusion of the immigration variables continues the decline in the negative effect of ethnic origins on income seen in model 2. The exception being a slight rise in the negative effects of being of a Latino ethnic origin on income from the model 2 lows.
The negative effects on income of being female remain almost constant across all 3 models. The inclusion of the immigration variables sees a slight weakening of the positive effects of having either a four year degree or a post grad degree on income. Model 3 also includes the marriage status binary variable; however this variable is irrelevant to this papers analysis as such will not be explored further.
Looking at the included immigration variables we see that for the Latin American region there is no relevant data until the 1950s. Educated immigrants from Latin American who immigrated in the 1950s show a strongly negative effect of -2.14 on incomes. This is the minimum reach for educated immigrants from Latin America and the minimum across the whole regression analysis. From years 1960 through 1989 immigrants from Latin America see a slight education premium over the native born population; however this premium turns into slight a deficit in the 90s. Immigrants from the Caribbean region again have no relevant data on immigrant’s pre 1950s; however we see a wage premium for educated immigrants in the 1950s of 0.221. This premium turns to a deficit of negative 0.213 in the 1960s. Overall a wage premium is seen for educated immigrants from the Caribbean region. Like the previous two regions, the South American region lacks relevant data on the pre50s educated immigrants. From 1950s onward educated immigrants from the South American region show a general wage deficit with respect to the native born population. In fact, educated immigrants from the South American region show a wage deficit with respect to the native born population every decade, with the exception of the 1970s, where there was a slight wage premium over the native born population of 0.127. Now looking at Western Europe, again there is no relevant data for the pre 1950s time period. The picture that emerges from the regression analysis of educated immigration from Western Europe is a tale of two different periods. Educated immigrants saw a wage premium over the native born population up until the 1970s at which point the wage premium turned into a deficit. In the decade of 1950 Western European immigrants had a wage premium of 0.372. In the decade of 1960 Western European immigrants had a wage premium of 0.029. By the decade of 1970 the wage premium had turned into a deficit of -0,082. This trend continued through the decade of 1980 where Western European immigrants had a wage deficit of -0.446. There is no relevant data for the decade of the 1990s. Eastern Europe is a different story, unlike Western Europe which had a premium which turned into a deficit, Eastern Europe seesawed. In the decades prior to 1950 educated Eastern European immigrants had a wage premium over the native born population of 1.559, which is the largest of the whole regression analysis. The 1950s saw a dramatic turn around with the wage premium turning into a deficit of -0.375. The 1960s saw a premium of 0.14 while the 1970s brought deficits again, this time of -0.685. The 1980s were a better time for educated Eastern European immigrants with a wage premium of 0.081, however by the 1990s this premium had again turned into a deficit of -0.315. Educated Asian immigration is much like Western European immigration in that it starts out with a wage premium for several decades which then turns into a deficit. In the case of educated immigrants from Asia they had a wage premium of 0.183 in the period prior to the 1950s which continued into the 1950s growing to 0.506. The decade of the 1960s saw a continuation of the earlier trend with wage premium of 0.081. The same was true for the decade of the 1970s where the wage premium was 0.0055. The 4 decade continuation of the wage premium for educated immigrants of Asian origin is the longest streak in the regression analysis. By the 1980s the premium had turned into a deficit of -0.005 and continued to fall reaching -0.016 in the 1990s. The final global region of the world studies by this regression analysis was the Subcontinent region. The subcontinent region lacked sufficient relevant data to provide information on any potential wage premium/deficit that may exist prior to the 1950s. The 1950s saw educated immigrants from the Subcontinent region realize a wage deficit with respect to the native born population of -0.37. The deficit narrowed in the 1960s to -0.001 and turned positive in the 1970s where educated immigrants from the Subcontinent held a wage premium of 0.172. By the 1980 the premium had turned into a deficit of -0.279. There is no data on the 1990s for immigrants from the Subcontinent region. Model 3 has an adjusted R-square of 0.589. The R-Squared statistic shows us how much, as a percentage, of the increase in LN INCWAGE can be explained by the variables included in model 3. In this case you can see that slightly over 58% of the increase in LN INCWAGE can be explained by the variables included in model 3. The remaining 42% would be explained by all other unobserved factors such as ability, competition for employees and favorable employer tax policies.
Overall the data does not provide a clear global picture of if a wage premium exists for educated immigrants. It is clear there does exist a wage premium at certain periods for immigrants from certain regions; however this does not provide clear, consistent, statistically relevant data which points to a global educated immigrant wage premium. Thus the data does not support nor disprove this author’s hypothesis of an educated immigrate wage premium. The data does clearly disprove the author’s hypothesis that the structural changes in the global economy of the last 50 years have lead to the development of a wage premium for educated immigrants. It is clear from the inconsistency of the regressions finding that wage premiums have existed since the 1950s as have wage deficits.
The original question asked by this report, is there a wage premium for educated immigrants over the native born population has been shown to be inconclusive, however that does not change the importance of the question. Understanding how well an immigrant assimilates economically is of critical importance to the future of the U.S. Economic achievement is a critical step in living the American dream. Immigration policies which foster and support rapid integration into the American narrative must consider how well an immigrant will perform economically in this society. As we have seen from the 4 studies reviewed at the opening pages of this report, education matters, not only to lifetime earnings potentials but also to economic assimilation and success in this country. This regression has shown that there is no certain path to prosperity however starting out with an education can go a long way towards ensuring immigrants succeed in the United States, thus ensuring a continuation of the uniquely American melting pot phenomenon.
Works Cited
Denny, Kevin , and Vincent O'Sullivan. "Can education compensate for low ability> Evidence from British data. ." Applied Economic Letters 14 (2007): 675-660.
"IPUMS CPS." Census Micro Data from the Current Population Survey. http://cps.ipums.org/cps/index.shtml (accessed April 22, 2009).
McCarthy, Kevin, and Georges Vernez. "The Success of Immigrants Increasingly Depends on Their Education." Immigration in a Changing Economy 1 (1998): 13-21.
Orrenius, Pia . "Immigrant Assimilation: Is the U.S. Still a Melting Pot?." Federal Reserve Bank of Dallas 3 (2004): 1-6.
Silles, Mary. "The Returns To Education For The United Kingdom." Journal of Applied Economics 1 (2007): 391-413.