Printer Friendly

Earnings differentials between natives and immigrants with college degree.

I. Introduction

The new Immigration Act of 1990, to take effect October 1991, is a major change in the country's 25-year-old legal immigration policy. The new law is designed to increase immigration of highly skilled manpower. Under the occupational preferences and labor certification program of the current law, an annual maximum of 54,000 visas are allocated to individuals, including spouses and children, who qualify for immigration due to their skills. This program is the only part of the immigration policy which screens and admits individuals based on labor market considerations. The new law would essentially keep the same program but would increase such immigration to at least 140,000.(1) This change constitutes an increase in skills-based admission from less than four to about ten percent of the total annual immigration flow. Given the unique screening of the labor certification beneficiaries and their conceivable productivity difference with other immigrants, relative earnings of highly skilled immigrants is an important issue which has not received sufficient attention in the literature. The legal basis of the current immigration policy is the Immigration and Nationality Act of 1952. This act established a preference system which applied only to quota countries (97 percent of the quotas were allocated to European countries, and 3 percent to all the other countries in the Eastern Hemisphere). The 1965 Amendment (current law) kept the preference system but changed the quota policy to provide equal treatment to all countries in the Eastern Hemisphere. A similar preference system and policy was established for the Western Hemisphere in 1977.

It is well documented that the 1965 Amendment helped to foster immigration of "independent" non-family individuals with college education during the late 1960s and the early 1970s. As of 1980, among all U.S. residents 25 years old and over, 16.2 percent had at least 4 years of college education. The corresponding figures for all immigrants, pre-1970 immigrants, and 1970-1980 immigrants are 15.8, 13.2, and 22.2 respectively. These figures along with the upcoming changes in the law suggest that the immigration flow will play an increasing role in the provision of college educated manpower needs of the U.S. labor market. With the decline in the flow of native-born graduates into advanced degree programs in recent years, particularly in science and engineering, it appears that immigration will play an even more important role in the labor market of highly trained manpower in the future. There is general consensus among economists that immigration of high-skilled labor constitutes a gain for the U.S. However, the relative success of such immigrants within the U.S. labor market, the impact of such immigrants on wages and employment of their American counterparts, and the consequent policy implications have not received sufficient attention in the literature.

This paper investigates the relative (compared to American-born) wages of U.S. immigrants with at least one college degree. In particular this study addresses the following questions: (1) Is there a significant wage differential between college educated immigrants and their American-born counterparts? and (2) What are the possible sources of this differential? Many past empirical studies have used the human capital framework to analyze the relative earnings of either total U.S. immigrants or of a particular ethnic group, regardless of level of education. These studies have focussed on the supply side of the labor market and have found the existence of earnings differentials between natives and immigrants. In general, the literature indicates that due to the imperfect transferability of country-specific human capital to this country the productivity and earnings of newly arrived immigrants in the U.S. are relatively low but overtake those of U.S. natives within 10 to 15 years of residence in this country.(2)

There are reasons to believe that the findings from studies on immigrant workers in general, may not be applicable to the particular group of immigrants with at least a college degree. First, it is reasonable to assume that formal college training is less country-specific than other forms of human capital, such as on-the-job training and experience. Second, as noted in the literature, approximately one half of all professional immigrants come from the potential pool of foreign students or exchange visitors who adjust to immigrant status after some length of residency and formal training in the U.S. labor market. These immigrants are most likely to be fluent in English and have acquired relevant skills and knowledge. Given these points, then the initial native-immigrant productivity gap and, therefore, the subsequent earning differentials between natives and immigrants are expected to be less for college educated immigrants than it is for overall immigrants. In the absence of productivity differentials, however, earning differentials might still exist due to imperfect information and/or discrimination in the labor market. The latter is a demand side possibility which should be taken into account in the study of relative earnings and is considered in this paper.

II. Model Specification

Following the human capital model of earning determination, the logarithm of earnings for native-born and foreign-born workers is assumed to depend upon a group of individual characteristics that influences worker's productivity. These characteristics include, but are not limited to, years of schooling, years of potential labor market experience and its square, and marital status. Previous empirical research of immigrant earnings, however, has found differences in the valuation of labor market experience acquired prior to immigration and experience acquired in the U.S. labor market. Thus the foreign-born wage model is modified to include separate terms for pre-immigration and post-immigration experiences, as well as a term representing the interaction between the two experiences. One interpretation of the interaction term is that the effect of post-immigration experience on earning growth is not independent of pre-immigration experience.(3)

Many empirical studies of black-white and female-male wage differentials have found significant differences in the wage structure among regions and occupations. To the extent that regional and occupational distributions of immigrant workers in the U.S. differ from those of native workers, an accurate measure of sources of earning differentials requires inclusion of region and occupation variables in the earning models. Inclusion of occupation variables in an earning equation may be justified as proxies for some unobserved human capital which affect one's productivity. Some occupations may have a greater demand for these unobserved characteristics and attract individuals with a greater endowment of them. In addition, in the case of skilled immigrants, region and occupation matter because the law is designed to channel such immigrants into tight labor markets.

The traditional approach to investigate the sources of earning differentials is the well known decomposition methodology. According to this technique, upon the estimation of two separate reduced-form equations, the gross earning differential between the two groups involved is decomposed into endowment and residual effects. The former is attributed to skills differentials and the latter is attributed to differences in the regression coefficients and is often considered as an estimate of discrimination. This approach implicitly assumes that the estimated residual effect is deterministic. However, this estimate is based on behavioral models (reduced form equations) and is affected by sampling variability. As such, this effect is stochastic, therefore, requiring statistical inferences.

A modification to the above decomposition approach is provided by Jackson and Lindley [9]. The modified alternative, based on a pooled sample of the two groups involved, not only provides estimates of the endowment and residual effects but also allows testing of the statistical significance of the residual difference. In addition, the approach allows testing of whether a particular coefficient is identical across the two samples. In the following section separate earnings equations are estimated for natives and immigrants with at least one college degree. Also, using the pooled sample of the two groups, the modified decomposition procedure and the corresponding statistical testings are performed.

III. Econometric Results

A. Data

The research population is confined to the non-black, male component of the U.S. labor force in order to control for sex- related interruptions (and the measured experience) and possible sex and race-related earning discrimination. Individuals are required to have at least one college degree, to be of age 24-65 and to have participated in the civilian labor force in 1980, as well as to have worked some time during 1979 (the year in which earnings are to be measured). The self-employed and individuals with any disability that limits work, as well as individuals attending college in 1980, are excluded from this population in order to better represent the earnings determination described previously.

Both an immigrant and a native sample of this population are obtained from the five percent one-in-one-thousand Public Use Sample of the 1980 Census. The immigrant sample is confined to those who perceive themselves as not having difficulty speaking English. The natural logarithm of 1979 weekly earnings (wage and salary) is regressed against the set of independent variables discussed earlier.

The means of socioeconomic characteristics of the two samples are provided in Table 1. This table indicates differences in several variables between natives and immigrants. For example, the average weekly earnings of immigrants is 3.7 percent less than that of natives. Notice also that immigrants have about .6 years more education and about .8 years less experience than natives. Table 1 also reveals differences in regional distribution between natives and immigrants. Compared to natives, immigrants have a greater tendency to reside in SMSAs, in the Northeast and the West and less tendency to reside in the North Central and the South. The higher concentration of highly educated immigrants, relative to that of their native counterpart, in metropolitan areas is consistent with the locational behavior of the overall U.S. immigrant population. Immigrants, in general, have a higher geographical concentration than natives. Bartel [2] found that 75 percent of recent immigrants live in the 25 largest SMSAs, (compared to 50 percent of natives) and that for all ethnic groups the location of fellow countrymen (past immigrants) has a positive impact on the immigrants' locational choice in the U.S. This impact is somewhat of a smaller magnitude, but highly significant, for the more educated immigrants. In addition, locating closer to family and friends may reduce the locational concentration of the Americans.(4)

TABLE 1 Sample Means, College Educated Native and Immigrant Males
Variable Name Native Immigrants Log
1979 weekly earnings 6.029 5.993
Personal Characteristics
Education 9.106 9.663
U.S. experience 15.017 9.563
Pre-immigration experience N/A 4.636
Married 0.760 0.805
U.S. Residence
Northeast 0.240 0.298
North Central 0.242 0.184
South 0.294 0.210
West 0.224 0.308
Metropolitan Area 0.827 0.950
Managerial and 0.307 0.231
Engineers 0.089 0.167
Math & Computer Sciences 0.014 0.017
Natural Sciences 0.018 0.028
Health 0.034 0.076
Teachers, Postsecondary 0.030 0.043
Teachers, Others 0.081 0.029
Humanities 0.091 0.054
Social Sciences 0.012 0.012
Technical and Adm. Support 0.202 0.194
Service 0.035 0.038
Blue Collar 0.086 0.110
Number of Observations 5080 4879

Source: See the text for the source and a detailed description of the samples.

Furthermore, Table 1 shows that immigrants are more likely to be in engineering, health, and postsecondary education (academicians), and less likely to be in managerial, humanities, and K-12 teaching occupations. The occupational distribution differences reflect the impact of U.S. immigration policy. The 1965 Amendment facilitated the immigration of individuals with advanced degrees in fields where labor has been in short supply, such as health and engineering. The Amendment also included the conversion of nonimmigrant status to that of an immigrant by graduating foreign students and exchange visitors such as professors.

B. Empirical Estimates

Table 2 presents the estimated regression coefficients for native Americans and immigrants. The signs of the estimated coefficients are consistent with the findings of other human capital studies of both native and immigrant earnings determination. In this regard, 1979 male earnings are augmented by additional years of education beyond an undergraduate degree, by potential experience (albeit at a declining rate), and by marital status. In addition, earning differentials by occupation are observed. Native individuals with engineering (reference group) or health related jobs earn relatively more than individuals in other occupations (For example, 29 percent more than academicians). Among immigrants, those with health related jobs earn the most, followed by social scientists and engineers. Among plausible explanations for occupational differences in earnings for both natives and immigrants are theories of compensating wage differentials and the corresponding nonwage aspects of jobs, as well as the hedonic theory of earnings.(5) The regressions also show that there are just under 6 percent differences in earnings among U.S. regions. In addition, native individuals living in SMSAs earn up to 18 percent more than those living in nonmetropolitan areas. The same is not true for immigrants. The t values are provided to test whether there is a significant difference between natives and immigrants with regard to the coefficient of each variable involved.(6) In this respect, notice that the returns to post-graduate education are the same for natives and immigrants (the difference between the two groups is equal to -.008 but statistically insignificant). The difference in returns to U.S. labor market experience is also statistically insignificant. This finding is in contrast with other studies which analyzed earnings profiles of all immigrants and have concluded that earnings rise faster with age or post-immigration experience for foreign-born than for native-born [3, 5, 7]. These results suggest a lack of rapid increase in earnings among immigrants with college degree. However, as expected, pre-immigration work experience has a much smaller effect on earnings than does experience gained in the U.S. (almost one-half).

The last column of Table 2 also shows that significant earning differentials in favor of natives are generated by marital status. The 5.8 percent lower earnings of married immigrant men is consistent with the findings by Borjas and Bronars [4]. They argue and demonstrate that the "intensity" of self-selection as well as the U.S. immigration selection processes with regard to unobserved "quality" is stronger among single than married immigrants. By controlling for the gains to specialization due to marriage, they provide evidence of up to a 20 percent reduced earning differentials between married and single immigrant men.

The results also indicate that natives require a premium of about 18 percent for living in more expensive metropolitan areas. Such a premium is not observed for immigrants. A plausible explanation for this difference is that immigrants may be willing to set lower reservation wages and, thus, accept lower offers in order to locate close to fellow ethnics. In addition, most foreign-born individuals emigrate from the large metropolitan areas of their home country. As such, they are accustomed to the life-style of populated areas and perhaps willing to pay a premium, in the form of lower real earnings, in order to reside in similar locations in the U.S. Finally, earning differentials in favor of natives are generated by some occupations such as humanities and technical and administrative support. In contrast, such differentials in favor of immigrants are generated by occupations such as social sciences, health, and postsecondary teaching (academicians). The contrast is a result of distributional differences between the two groups with respect to the fields of specialty within each occupation category. For example, an examination of the health occupation category for the native sample, revealed that about 54 and 23 percent are physicians and pharmacists, respectively. The corresponding figures for the immigrant sample are 79 and 10.


The above discussion provides plausible explanations of the earning differentials between highly educated natives and immigrants. The explanations are based on the differential characteristics of the two groups of workers--a supply side consideration. Another possible contributing factor to the earning differentials, however, may be the demand side of the labor market. This possibility is also investigated by decomposing the 3.7 percent gross differential in favor of natives into the endowment and residual effects.(7) The results are reported in Table 2. The endowment effect indicates a gap of 3 percent in favor of immigrants. The residual effect indicates a statistically significant "unexplained" differential of about 6.8 percent against immigrants. The latter may be a result of imperfect information and/or discrimination in the labor market.(8)

IV. Summary

This analysis has examined the relative earnings of U.S. immigrants with at least one college degree. Given the labor certification policy of the U.S. immigration law and the conceivable relative productivity-related differences between college-educated immigrants and immigrants in general, relative earnings of these highly trained immigrants is an important issue, which has been precluded by past research.

Human capital earning models are estimated and the wage decomposition methodology is applied. Contrary to the previous findings of relative earnings of all immigrants, regression results indicate a lack of difference in returns to education and U.S. labor market experience between natives and immigrants. However, there are differential returns to marriage, metropolitan area location, and some occupations between the two groups of workers. In addition, the endowment effect is about 3 percent in favor of immigrants, perhaps an indication of the effectiveness of the manpower provision of U.S. immigration policy. Accounting for differences in human capital characteristics, the estimated "unexplained" earning differentials against immigrants is less than 7 percent.


1. Current law provides for an annual admission of 270,000 legal immigrants under the preference system, 80% of the available visas are allocated to family reunification categories and 20% to occupational categories including spouses and children of the beneficiaries. In addition, immediate relatives of U.S. citizens and refugees are admitted outside these numerical restrictions. As a result, in recent years U.S. has experienced an annual immigration of more than 600,000. Under the occupational preferences and labor certification program of the current law a foreign-born professional, scientist, artist with exceptional ability, and worker in an occupation for which labor is in short supply can apply for immigration and employment in the U.S. An immigrant status will be granted to such applicant if an equally qualified American is not available at the place to which the applicant intends to work, and if the wages and working conditions of similarly employed Americans are not adversely affected. In addition, employment of such immigrant cannot be based on the acceptance of lower wages. The new law establishes an annual level of 700,000 for 1992-1994, 675,000 thereafter, and requires essentially the same conditions as the current law for employment- based immigration.

2. The extent of the immigrants' earnings growth has been the subject of extensive research. For examples of this work, see Borjas [3], and Chiswick [5]. An excellent survey of this literature is provided by Greenwood and McDowell [8].

3. The assimilation model by Chiswick [5] includes the total years of potential labor market experience (EXP) and the number of years in the U.S. (YSM). The latter variable is a proxy for post- immigration experience and accounts for the differential rates of return to pre- and-post immigration experiences. He argues that replacement of post-immigration experience by YSM does not substantially alter the interpretation of the results [p. 904]. Fujii and Mak [6], however, demonstrate that this model (inclusion of EXP and YSM) "is both analytically flawed and inconsistent with empirical evidence" [p. 1145]. They advance and empirically test a model which includes both pre-and post-immigration experiences but not YSM. This latter model was also employed by Gabriel and Schmitz [7].

4. Traditionally, Europeans have favored locating in New York, Boston, and other Northeastern cities. While Asians have favored Los Angeles, New York, and San Francisco. The fact that a large portion of college educated immigrants (according to the 1980 Census about three-fourths of post-1969) are from Europe and Asia explains their concentration in the Northeast and the West.

5. For a detailed discussion of theories of compensating wage differentials and the related empirical literature, see C. McConnell and Brue [10].

6. These t values are obtained by estimating one single equation from the pooled native-immigrant sample. In addition to the original variables, this equation includes an immigrant dummy variable as well as an immigrant dummy-interaction term for each of the independent variables. This model is equivalent to the estimation of the two separate equations reported in Table 2, except that the coefficients of dummy and dummy interaction terms are the differences in the native-immigrant estimates for the intercept and for each of the explanatory variables.

7. The endowment effect is the summation of differences in the mean of explanatory variables weighted by the regression coefficients of the American equation. The foreign experience terms are weighted by the coefficients of the immigrant equation and are included in the endowment effect. The residual effect is the difference between the gross differential and the endowment effect.

8. Adopting Kenneth J. Arrow's [1] notion of discrimination, lower earnings for reasons other than productivity differentials--the 6.8 percent "unexplained" difference may be attributed to labor market discrimination.


Arrow, K. J., "The Theory of Discrimination," in Orley Ashenfelter and Albert Rees, eds., Discrimination in Labor Markets, Princeton, NJ: Princeton University Press, 1973, pp. 3-33.

Bartel, A. P., "Where Do the New U.S. Immigrants Live?" Journal of Labor Economics, 7 (October, 1989), pp. 371-391.

Borjas, G. J. "Assimiliation, Changes in Cohert Quality, and the Earnings of Immigrants," Journal of Labor Economics, 3 (October, 1985), pp. 463-489.

Borjas, G. J. and S. G. Bronars, "Immigration and the Family," Journal of Labor Economics, 9 (April, 1991), pp. 123-149.

Chiswick, B. R., "The Effect of Americanization on the Earnings of Foreign-born Men," Journal of Political Economy, 86 (October, 1978), pp. 897-921.

Fujii, E., and J. Mak, "On the Specification of the Income Equation for Immigrants," Southern Economic Journal, 49, (April,1983), pp. 1141-1146.

Gabriel, P. E. and S. Schmitz, "The Relative Earnings of Native and Immigrant Males in the United States," Quarterly Review of Economics and Business, 27 (Autumn 1987), pp. 91-101.

Greenwood, M. J. and J. M. McDowell, "The Factor Market Consequences of U.S. Immigration," Journal of Economic Literature, 24 (December, 1986), pp. 1738-1772.

Jackson, J. D., and J. T. Lindley, "Measuring the Extent of Wage Discrimination: A Statistical Test and a Caveat," Applied Economics, 21 (October, 1989), pp. 515-540.

McConnell, C. R. and S. L. Brue, Contemporary Labor Economics, New York: McGraw-Hill Inc., 1992.
COPYRIGHT 1993 Omicron Delta Epsilon
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Daneshvary, N.
Publication:American Economist
Date:Sep 22, 1993
Previous Article:Determinants of income velocity in the United Kingdom: multivariate Granger causality.
Next Article:Replacing the autarky factor prices with the corresponding post-trade prices: an extension of Deardorff's factor-proportions theory of trade.

Related Articles
Shortage of college graduates? More evidence is needed.
Immigration and ethnicity.
Immigration and wage changes of high school dropouts.
Earnings of college graduates: women compared with men.
Immigrant earnings performance in Canadian cities: 1981 through 2001.
The economics of the melting pot: how have the 25 million immigrants arriving in America since 1980 affected the employment equation?

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters