Education: Productivity Booster or Merely a Signal? Examining role of education in the US labor market using the immigrant population

Transcription

1 NORTHWESTERN UNIVERSITY MATHEMATICAL METHODS IN THE SOCIAL SCIENCES PROGRAM Education: Productivity Booster or Merely a Signal? Examining role of education in the US labor market using the immigrant population Weijia (Cassie) Zhang 5/9/2014 Advisor: Professor Richard Walker Instructor: Professor Joseph P. Ferrie

2 Abstract It is generally established that education is positively correlated with income in the US labor market. This paper examines the role of education in the labor market by attempting to distinguish between the human capital and signaling theories. If the signaling theory is truly at play, then a worker s level of education relative to his cohort should be significant in estimating his / her income. US immigrants are the ideal population to study in that they offer a great variety of education levels and qualities. I modify Mincer s model and create a measure of relative education to test the significance of the human capital and signaling theories. The results found that both absolute and relative measures of education were extremely statistically significant. This evidence supports the conclusion that both theories are in effect in the US labor market. 1

3 ACKNOWLEDGMENTS I would like to take this opportunity to thank the numerous individual that have made this thesis possible. Professor Richard Walker, my thesis advisor, has provided me invaluable insights and encouragement throughout the development of this thesis. This thesis would not be in existence without him. I owe a great deal to Professor Joseph P. Ferrie, Professor William Rogerson and Sarah M. Ferrer for not only running the thesis program, but also the MMSS program, from which I gained the analytical tools and mind to complete this thesis. I would also like to thank Mr. John J. Hernandez n the University Library for helping me to find data for this study. Lastly, I would like to thank my parents, Lin He and Yuning Zhang, for their continual guidance and support over the last 22 years of my life and for always providing me with the best education opportunity possible. 2

4 EDUCATION AND TALENT IN US IMMIGRATION Weijia (Cassie) Zhang Northwestern University May 2014 Introduction Undisputedly, education is a key aspect of a country s developmental progress and significantly contributes to the its human capital growth. In fact, expanding and improving education has always been a key tenet of development strategy because it is believed to provide many positive externalities on a country s economic and social well-being. It has always been assumed that education adds skills and knowledge to the labor force. The fact that the wealth of a nation increases with improvements in education follows the understanding that people with more education tend to receive higher wages. Hence, if a country has more educated individuals, the average and total income of the country should increase as well. Despite the fact that education should increase workers earnings, the pursuit of education is costly at an individual level, both monetarily and time-wise. The high opportunity cost of education is often the largest deterrent for a worker, especially those in developing countries, where the education systems are often extremely inadequate and inefficient. So, while education increases earnings in the long-run, it involves a costly short-term trade-off between time and money while the workers are in school. Thus, each worker must make a complex decision to determine his / her optimal level of education. In this paper, I build upon the existing literature regarding the study of the role of education in the labor market. I explore the role of education by testing whether workers and firms view education as a method to acquire more skills and knowledge, thus increasing worker 3

5 productivity and long-run income, or as a signal of the worker s innate talent. My study is unique in two ways. First of all, my study uses the latest US census data from the 2010 Census, which provides an updated view of the market. Second of all, unlike previous studies, my study examines the differences in effect between US natives and immigrants. Immigrants are the ideal population to examine in this research for a couple of reasons: (1) they provide a varied sample of talent distributions from different countries (2) they offer us insight into the different rate of returns of education across different levels of qualities of education and (3) they most likely derive a relatively high utility from career opportunities. Thus, this paper rests nicely at the intersection of immigration and education as a signal of talent. Using a modified version of Mincer (1974) model of education and schooling, I explore the explanatory power of a worker s amount of education (in terms of years of school) and his / her relative position in his / her talent distribution using the data from the 2010 US Census. The results demonstrate that in the US labor market, education both adds human capital and serve as a signal of worker quality. Section I: Background and Review of Education Education and Productivity It is generally accepted that workers education and earnings are strongly correlated at a microeconomic level. However, there is little consensus regarding the connection between education and productivity. Three schools of thoughts have emerged regarding this relationship. All three schools use the assumption that the labor demand and supply determine, in equilibrium, a schedule of wages for all workers participating in the labor market, which creates an earning function for the market. Also, all three schools assume that every worker decides the level of 4

6 education that he / she will obtain. The optimal level is when the marginal cost of education is equal to the marginal benefit given by the earning function. Education Builds Human Capital The first school of thought argues that education builds the human capital of the labor force by teaching workers skills and making them more productive. Becker s 1962 study on human capital asserted that when workers determine their optimal level of education, they compare the present value of lifetime earnings associated with different levels of schooling. They remain in school as long as the marginal benefits of schooling outweigh the marginal costs (Jones). Becker s model makes two key assumptions: (1) a worker s wage is a perfect reflection of his / her marginal productivity and (2) all variables correlated with education s effect on wages are included in the regressions. Thus, education may (1) widen a worker s access to different sources of information and (2) increase a worker s ability to learn from past experience (Jones) Through education, a highly educated worker develops an comparative advantage over the less educated in certain activities, such as technological adoption, since education enhances a worker s ability to learn and synthesize information. Education as a Signal of Innate Talent The second school of thought is derived from the studies of Spence in 1973 and It argues that education serves purely as a signal of innate ability. The model makes several assumptions, the first of which is that asymmetric information occurs in the labor market in that only the employees know their position along the country s talent distribution. Firms / employers only know the talent distribution but not an individual s position along the distribution. The second assumption for this model is that education is costly. Workers must devote a significant amount of time to attain a degree, as well as financial resources. Even if there are no 5

7 tuition associated with the education (e.g. countries that provide free public education), there is a high opportunity cost to education, especially in countries where low-skilled labor is readily available and relatively well compensated. Then, by being a student, the worker is also losing out on potential earnings. The model s third assumption is that the cost of education decreases as we move further to the right along the distribution. In other words, it is more difficult for a less talented worker to attain the same level of education as a more talented worker and the marginal difference in difficulty increases over education levels. Hence, each worker s choice of education attainment is specific to their position along the talent distribution. Since the cost of getting an education is greater for workers with lower ability, then only workers with high ability can and will obtain high levels of education. So, through education, workers can send the firm / employers signals to reveal their relative position along the talent distribution. In this model, the signaling equilibrium, equilibrium in which the sender (workers) signals and the receiver (firms / employers) trusts that signal and the information it implies, is that a worker s level of education acts as a signal to the firm / employers the worker s marginal productivity, and hence the appropriate level of compensation that specific worker should receive. Hence, under the signaling theory, workers receive higher wages, not necessarily because education has increased their marginal productivity, but because firms view education as an indicative signal of the worker s quality and talent. In other words, education may add little or no value to a worker; workers would only pursue education to demonstrate their innate ability. Jones explains that in a signaling world education still reflects productivity, even if it does not cause all of it. Thus, firms can use level of education as an indicator of worker quality. 6

8 Unfortunately, it has been difficult to determine the extent to which education is a signal of talent and ability, as opposed to enhancing worker productivity, since all both explanations imply that the correlation between wages and education is positive. Education in Dynamic Environments The third school of thought on the relationship between education and productivity is that differences in environment yields different levels of productivity for workers with the same level of education. For example, returns to education are probably higher in dynamic environments since education improves workers access to key skills and knowledge. However, due to the limitation in data availability, this paper will only be able to test the human capital and signal schools of thoughts. Return to Education Traditionally, research examining return to education has typically used years or degree attained as the primary measurements of productivity for each potential immigrant. However, this approach makes several unrealistic assumptions. For example, by using years of education without any sort of quality equalizer, the traditional approach assumes that the quality of education is identical in every country. It is easy to argue that this conjecture is inaccurate; the quality of a year of education should be higher in developed countries, such as Denmark, than those in a developing country, such as Laos. By this logic, years of education should be discounted by the quality of education in the subject s native country. The traditional approach also assumes that education builds human capital and ignores the role of education as a signal of talent. Immigration is a good natural experiment to examine the role of education as a signal of innate talent by examining this relationship with various different education systems and innate talent distributions. 7

9 Section 2: Overview of US Immigration In order to fully use the immigration data, it is important to understand the US immigration landscape. In this section, I provide a brief overview of the history, change and current state of US immigration. In addition, I compare and contrast US family-based immigration policy with that of the point-based skills system that countries like Canada and Australia are currently employing. I include this last analysis because if education is truly a signal of talent for employers, then the point-based skills system should be an excellent method to identify the most productive workers in the immigration candidate pool. Since the US has faced a chronic shortage of highly skilled workers, especially in the STEM fields (Science, Technology, Engineering and Mathematics), this analysis could provide further evidence in support for the implementation of the point-based skills system in the US. History of US Immigration Immigration in the US can be divided into two distinct time periods: Pre-1965 and Post In the Pre-1965 time period, the vast majority of immigrants were from Europe, with the exception of a large but brief influx of Chinese immigrants during the Gold Rush in the mid- 1800s. During this period, migrants were viewed unfavorably by the Anglo-Saxon Protestant natives. Racial, anti-catholic and anti-immigration sentiments often ran high. The Naturalization Act of 1790 explicitly established the racial criteria that limited citizenship to Whites only. In 1882, President Chester A. Arthur signed the Chinese Exclusion Act, which blocked the migration of all Chinese workers into the US. These policies demonstrate a clear bias against certain migrant groups, namely, non-europeans. In the 1920s, Congress began to establish a national quota system. The Immigration Act of 1924 created a quota system that restricted entry 8

10 to 2 percent of the total number of people of each nationality in America as of the 1890 national census a system that favored immigrants from Western Europe and prohibited immigrants from Asia. ("U.S. Immigration Before 1965"). Thus, the pre-1965 immigration period can be seen as race-based and culturally biased in favor of Anglo-Saxon. The Hart-Cellar Immigration Act of 1965 can be considered as the single most important document in the history US immigration. It abolished the origin-based quota system that was established in the 1920s. The system was replaced by a preference system based on the immigrant s skills and familial relationship with US citizens and/or residents. This replacement dramatically changed the demographic mix of the inflowing immigration stream. As Table 1 and 2 demonstrate below, we can see that Pre-1965 US immigrants were predominantly European. There were very few Asians and South American who were allowed to immigrate to the US. Post-1965 US immigration is characterized by a much more even distribution of immigrants across continents. In particular, Asia and South America has had the greatest amount of growth in the past few decades. The abolishment of origin-based quota system created a more attractive immigration environment for a varied immigrant group to enter the US. 9

11 Table 1 US Immigration by Region, REGION Number of Immigrants Percentage Total 10,330, % Africa 2, % Asia 120, % Europe 8,881, % North America 1,179, % Oceania 8, % South America 137, % Unknown 10, % Source: 1900 US Census Table 2 US Immigration by Region, 2012 Region and country of citizenship 2012 REGION Number of Immigrants Percentage Total 53,887, % Africa 573, % Asia 11,062, % Europe 15,690, % North America 19,996, % Oceania 1,618, % South America 4,651, % Unknown 294, % Source: 2012 Yearbook of Immigration Statistics Current US Immigration Policies and Comparison with that of the Skill- Based Points System The Immigration Act of 1965 has made it easier for immigrants from all over the world to immigrate to the US, as seen in Figure 1. As a result, the total number of US immigrants has increased significantly over the years. 10

12 Number of Immigrants Figure 1 Legal Permanent Resident and H1B Visas, Fiscal Years 1980 to 2012, United States 2,000,000 1,800,000 1,600,000 1,400,000 1,200,000 1,000, , ,000 Legal Permanent Resident Total Refugee Arrival Total H1B Total 400, ,000 0 Year Note: Data for H1B visas is unavailable , and The H1B visa is used to employ foreign workers in occupations that require highly specialized knowledge in fields such as science, engineering, and computer programming. Source: 2007 Yearbook of Immigration Statistics Since 1965, family reunification has been the focus of US immigration policies. The quota for close family members and spouses of US citizens and Lawful Permanent Residents is unlimited, whereas those for employee-sponsored is extremely limited and coveted every year. The general categories of US immigration visas are documented in Table 3 below: 11

13 Table 3 US Immigration Types and Explanations Visa Category Description Visa Types Immediate Relatives of US Citizens Immediate family of US Citizens of Lawful Permanent Residents Spouses of US Citizens of Lawful Permanent Residents (IR-1) Unmarried Children of US Citizens of Lawful Permanent Residents (IR-2) Orphans adopted abroad by US Citizens of Lawful Permanent Residents (IR-3) Quota Type Unlimited Orphans adopted in the US by US Citizens of Lawful Permanent Residents (IR-4) Parents of US Citizens of Lawful Permanent Residents (IR-5) Family First Preference (F1): Unmarried sons and daughters of US Citizens and their minor children, if applicable 23,400 Family sponsored Preferences Visas for more distant family members of US Citizens of Lawful Permanent Residents Family Second Preference (F2): Spouses, minor children, and unmarried sons and daughters (age 21 and over) of Lawful Permanent Residents. At least 77% of all visas available for this category will go to the spouses and children; the remainder is allocated to unmarried sons and daughters 114,200 Family Third Preference (F3): Married sons and daughters of US Citizens and their spouses and minor children 23,400 Family Fourth Preference (F4): Brothers and sisters of U.S. citizens, and their spouses and minor children, provided the U.S. citizens are at least 21 years of age 65,000 12

14 Visa Category Description Visa Types Quota Type Employment Based Diversity Refugees, Asylees, Parolees Employment based immigrant visas are divided into five preference categories. Certain spouses and children may accompany or follow-to-join employment-based immigrants. To be considered for an immigrant visa under some of the employmentbased categories below, the applicant's prospective employer or agent must first obtain a labor certification approval from the Department of Labor. Once received (if required), the employer then files an Immigrant Petition for Alien Worker, Form I-140, with the U.S. Citizenship and Immigration Services (USCIS) for the appropriate employmentbased preference category. Drawn from random selection among all entries to persons who meet strict eligibility requirements from countries with low rates of immigration to the United States Refugee status or asylum may be granted to people who have been persecuted or fear they will be persecuted on account of race, religion, nationality, and/or membership in a particular social group or political opinion. Employment First Preference (E1): Priority Workers 1. Persons with Extraordinary Ability in the arts, sciences, education, business or athletics 2. Outstanding Professor or Researchers with at least three years of experience in teaching or research, who are recognized internationally 3. Multinational managers or executives who have been employed for at least one of the three preceding years by an oversea affiliate, parent, subsidiary, or branch of the US employer 40,040 Employment Second Preference (E2): 1. Professionals holding advanced degrees: must hold at least a baccalaureate degree and at least five years progressive experience in the profession 2. Persons of exceptional ability: n the sciences, arts, or business. Exceptional ability means having a degree of expertise significantly above that ordinarily encountered in the sciences, arts, or business 40,040 Employment Third Preference (E3): 1. Skilled workers: persons whose jobs require a minimum of 2 years training or work experience that are not temporary or seasonal 2. Professionals: members of the professions whose jobs require at least a baccalaureate degree from a U.S. university or college or its foreign equivalent degree 3. Unskilled workers: persons capable of filling positions that require less than two years training or experience that are not temporary or seasonal 40,040 Employment Fourth Preference (E4): Certain Special Immigrants (includes certain religious workers, Afghan and Iraqi interpreters, former employees of US government abroad, etc) 9,940 Employment Fifth Preference (E5): Immigrant Investors N/A The law and regulations require that every DV entrant must have at least: 1. A high school education or its equivalent; or 2. Two years of work experience within the past five years in an occupation requiring at least two years' training or experience. 55,000 To apply for Asylum, file a Form I-589, Application for Asylum and for Withholding of Removal, within one year of arrival to the United States N/A 13

15 Note: A lawful permanent resident (commonly known as a Green Cardholder) is a foreign national who has been granted the privilege of permanently living and working in the United States Source: US Department of State - Bureau of Consular Affairs US s heavy focus on family reunification as the central tenet of its immigration policy is a controversial choice. Family-based visas account for roughly two-thirds of all permanent visas allotted in an average year. Just 14 percent of all permanent visas in 2010 were employmentbased (The Hamilton Project). While the current policies embrace America s family-based values, many groups are disadvantaged by the system. For example, a group of valuable and significant immigrant group is slighted from the current US immigration system is the highlyskilled immigrants. It is extremely difficult and costly for an immigrant to immigrate to the US without any family relations in the US. While family members of US citizens and permanent residents are reunited in the US, highly-skilled immigrants with little to no family connection in the US often struggle to enter and stay. Employment-based immigration is a drawn-out and expensive process for both the immigrant and their employers. Permanent Resident status often takes more than 5 years to approve. The demand for the highly skilled work visa (H1B) is also clearly much larger than that of the supply. In fact, the cap for 2015 H1B visas were filled in merely 6 days in 2014 ("USCIS Reaches FY 2015 H-1B Cap"). Especially in a period like the present, there is a severe shortage of skilled workers in the US, the US unnecessarily complicated immigration system discourages companies to hire skilled workers from other countries, as the legal service fees for hiring such a worker often proves to be astronomical. To solve the chronic shortage of skilled workers, especially in the STEM fields, the US should consider migrating to the points-based skills system, like Canada and Australia have done. 14

16 Points-Based Skills System On the diametric end of immigration systems, countries, such as Australia and Canada, have long instituted a simpler skills-based selection system that awards points to immigrants who demonstrate English fluency and possess skills for needed jobs. For example, Canada s system is designed to select the immigrants who are most likely to integrate into the Canadian workforce based upon their background ("Points-Based Immigration Systems: Canada"). Six dimensions of an immigrant s background are measured by the points system: 1. Education (up to 25 points): individuals holding a Master s Degree or a Ph.D. and at least seventeen years of full-time or full-time equivalent study are awarded the full 25 points. Persons with only a high school degree are only awarded 5 points, which is the minimum. 2. Languages (up to 24 points): An applicant can be awarded up to twenty-four points for basic, moderate, or high proficiency in English and French. Language proficiency is determined through written and oral language exams. 3. Experience (up to 21 points): The maximum number of points can be earned by applicants in certain occupations. However, certain professions are restricted to prevent labor surpluses and competition. The full amount of 21 points can be earned by applicants with at least 4 years of experience in the approved occupation. Two points are deducted for each year less than 4 and one year of qualified experience will earn the applicant 15 points. 4. Age (up to 10 points): The maximum is awarded to applicants between 21 and 49 years old. Outside of the range, every year costs 2 points deduction. 15

17 5. Arranged Employment (up to 10 points): Awarded if applicant has a permanent and confirmed job offer. 6. Adaptability (up to 10 points): determined by spouse education, previous work experience in Canada, as well as family relations in Canada. Applicants must possess at least 67 points out of the full 100 to qualify as a skilled worker to immigrate to Canada. Source: Library of Congress As seen in the description of the Canadian points system, the only advantage that family relationships offers an immigrant in their application to Canada is in the adaptability criteria, which only counts for 10% of the overall selection process. Clearly, Canada has chosen to prioritize skilled immigration above family reunification, unlike the US. The advantages of such a points-based system are numerous and significant. For one, it is clearly designed to attract immigrants who are skilled and who will complement the Canadian labor market. These immigrants will also be able to integrate into and contribute to their future communities in Canada. Another key advantage of the system is its transparency. Applicants have a clear understanding of the selection process and criteria, which will allow them to selfselect more efficiently. The system also gives skilled immigrants more freedom from their Canadian employers since the validity of their visa is not permanently tied to their employer, which allows them to move easily in the Canadian labor market. The skills-based system is not without flaw, however. For one, applicants have complained about some of the criteria, such as adaptability, as unfair since they are very subjective. Also, this system has actually increased the processing time and backlog times for Canadian immigration services. 16

18 Section 3: Analysis and Results Data The analysis presented in this paper is drawn from the 2010 US Census One Percent Public Use Microdata Sample. It captures over 500 occupations that are separately identified. Since this study examines the dynamics of current labor market, non-current participants of the labor market were excluded from the sample. This study restricts pool of subjects to year old males who have been employed at some point in the last 5 years. Females were purposefully chosen to be excluded from the study to avoid effect of gender discrimination in the labor market. I was able to exclude the unemployed from the sample, as they provide no significant insights into the current labor market. The data contains a total of 762,800 US natives and 118, 381 male US immigrants. It includes important information about the subjects such as highest level of education attained, wage, birthplace, race / ethnicity and military experience. For the sake of this study, I define an immigrant to be any foreign-born individual contained in the dataset. A full demographic breakdown of the dataset can be found in the appendix. The sample contains 6 variables for each subject: birthplace, occupation (coded according to ACS 2010 code), annual income, marital status, level of education and race. I was able to create a dummy variable for military experience using the ACS 2010 occupation code. From the birthplace variable, I was able to able to create a dummy variable for immigrants, with the variable equaling 1 if the subject was born abroad and equaling 0 if the subject was born in a state in the US. Since the survey coded certain degrees (i.e. high school graduate, college graduate, professional degree beyond college) as the highest level of education attained rather the absolute 17

19 number of years of education received, I had to convert each degree to number of years. I make the assumption that the number of years of education required to attain each degree is the same across different countries and that it follows the structure of the US education system. In the US, 12 years is equivalent to a secondary school degree (or high school), 16 years to a college degree, 18 years for Masters and around 20 years for any degree beyond a Masters (e.g. PhD). Table 4 Degree and Years of Education Conversion Degree Attained Years of Education Completed High School 12 College 16 Masters 18 Professional Degree beyond Masters 20 Figure 2 Years of Education for US Natives and Immigrants Percentage Years of Education Natives Immigrants Note: Immigrant is a dummy variable equaling 1 if the subject was born abroad and 0 if the subject was born in the US Figure 2 shows the distribution of education of the US natives and the immigrant population. The mean of the distribution for US natives is around and the variance around The mean of the distribution for the US immigrants is around and the variance 18

20 around Two important observations can be drawn based on the information presented above about this dataset. First, the distribution of years of education for US immigrants is much more bipolar than that of the US natives. This observation can be evidenced by US immigrant s large variance (almost double that of US natives). If we inspect both distributions closely, we can also see that the percentage of US immigrants at lower education levels (0-5 years) is at least 6 times higher than that of the US native. At the same time, the percentage of US immigrants at higher education levels (16+ years) is also significantly higher than that of the US natives. This observation tells us that the US is attracting the two extreme ends of the education spectrum: the extremely uneducated and the extremely well-educated. The second observation is especially interesting. At first, it seems that, on average, US natives are slightly more educated than US immigrants based on the mean of absolute number of years of education. However, I doubted that we should measure skill of immigrants in terms of absolute years of education. In order to create an accurate and comparable measure of education levels, I converted each immigrant s years of education in their home country to US years of schooling using the following transformation: Normalized Years i = Absolute Years ij Mean Years j + US Mean In other words, I took worker i s number of absolute years of education and subtracted his / her home country s mean years of education to first obtain the number of education that worker i has over / under his / her home country cohort. Adding US mean of 13.3 years would tell us worker i s education level in the US talent distribution. Figure 4 below show the results. After normalization, we can see that, in terms of education relative to home-country mean, US immigrants are actually more educated than US natives. This finding is evident especially when we see that US immigrants range of years of education is wider and skewed to 19

21 0 Density.5 1 the right than that of US natives. In other words, the difference between Figure 3 and 4 that indicate that relative education level differs than absolute education level. Figure 3 Absolute Years of Education Years of Education US Natives US Immigrants 20

22 0 Density.5 1 Figure 4 Years of Education Normalized to US Mean Years of Education US Native US Immigrant Turning to the pattern of earnings, it is also important to note that the mean annual income of an immigrant is lower than that of the US natives by almost $2,000, as seen in Table. There are various explanations that can reconcile this difference. First, as noted above, immigrants have a lower mean for years of education than that of natives. So, if education and wages are positively correlated, as shown in many previous studies, then it makes sense that a group with a lower mean of absolute years of education, the immigrants, would also receive lower wages than the natives. This finding, as mentioned above, does not help us distinguish the role of education as a signal from building human capital, as both roles implies that wage and education levels be positively correlated with each other. Second, as Chiswick and Miller notes in their study of education mismatch, there are immigrant-specific factors that may contribute to a greater mismatch of education and 21

23 occupation among the foreign born in the labor market. These factors include limited international transferability of skills and selectivity in migration. Limited international transferability of skills refers to the fact that the skills and education attained in their home country may not be recognized to the same degree in the destination country. Examples include occupation-specific certifications (e.g. medical licenses, BAR, CPA, etc), language barriers and government regulations. In many occupations, such as healthcare, immigrants often complain that there is little recognition for their pre-immigration education, skills and experience in their destination country. This obstacle may prevent immigrant to be compensated at the same level as a native, especially for highly educated immigrants. Another explanation is that foreign schooling is, on an aggregate level, less productivityenhancing than that of the US. In other words, differences in quality of education could play a role into describing differences in skill and compensation match in the US labor marketplace. If firms / employers accept that foreign schooling is of lower quality than that of the US, then it would make sense that they would compensate immigrants at a lower level than natives. Discrimination against immigrants is also an obvious explanation for this income discrepancy. Unfortunately, race and ethnicity are obvious examples of such discrimination, as well as xenophobia. Highly skilled immigrants, especially those who are able to stay in the US using the H1B visa, are especially vulnerable to their employers. Since US Citizenship and Immigration Services requires that all H1B visa holders be employed and employer-sponsored at all times, employers wield a tremendous amount of power over their immigrant employees. For this reason, highly skilled immigrants often are willing to accept a lower pay, with respect to that of US natives, in order to maintain their employer s sponsorship. 22

24 Table 5 Income of US Natives and Immigrants. tabstat inctot, by (immigrant) Summary for variables: inctot by categories of: immigrant (Dummy =1 if subject is born abroad) immigrant mean Total Methodology Mincer (1974) established that the relationship between a worker s education levels and his / her wages is log-linear. The earnings function can thus be written as, (1) ln w i = α + βschyr i + ɛ i The ln w i is the logarithm of a worker i s wage, Schyr the years of the worker i s education and ɛ i represents other factors that affect worker i s wages, such as race / ethnicity and marital status. Mincer s model makes several assumptions. (1) The only cost of an additional year of schooling is foregone earnings and (2) the marginal increase in earnings due to the additional year of schooling is constant during the worker s lifetime (Jones). Thus, β can be understood as the rate of return of each additional year of schooling. This rate is reflected in the proportional change in the worker s income for each additional year of schooling. Thus, β represents the role education plays in adding productivity to a worker, as reflected in his / her wage levels. To measure to role of education as a signal, I modify the variables in Mincer s model. Kroch and Sjoblom (1994) and Johnes (1998) argued that employers use education to measure a worker s ability relative to the cohort. Therefore, if education is truly a signal of ability, then employers should infer a worker s ability from his / her education levels relative to that of his / her cohort in their home country. I define relative education as follows: 23

25 = Talent_dist ij Education i Mean Education j where i represents an individual worker and j represents i s home country. Therefore, Talent_dist i is the measure of an individual s talent relative to his / her home country cohort. The modified Mincer model is thus as follows (2) ln w i = γ + βschyr i + ϰtalent_dist ij + θschyr i *Immigrant i + ɛ i The ln w i is the logarithm of a worker i s wage, S i the number of years of formal education that the worker i received, RE ij the measure of a worker i s education against that of his / her cohort in his / her home country and ɛ i represents other worker-specific factors that affect worker i s wages, such as race / ethnicity and marital status. Schyr i *Immigrant i measures effect of being an immigrant has on worker i s marginal return to education. The returns for an additional year of education are thus: MRE i = β+ϰ If Immigrant i = 0 = β+ϰ+θ If Immigrant i = 1 Where MRE i is worker i s marginal returns to education and Immigrant i is a dummy variable that equals 0 if worker i is born in the US and 1 if the worker i is born abroad. We run the regression over the entire dataset. The human capital theory is confirmed if β is significant. The signaling theory is confirmed if ϰ and θ are both significant. If all three coefficients are significant, then the result tells us that both the human capital and signaling theories are in practice in the US labor market. We expect β and ϰ to be positive if they are significant and θ to be negative if found to be statistically significant. Results The results reveal several key findings. First of all, the fact that β, ϰ, θ are all extremely statistically significant, as evidenced by the t-static, demonstrates that both the human capital 24

26 theory and the signaling theory are valid in the US labor market. β of demonstrates that, for natives, each additional year of education below the mean level of education in the US (13.3 years) yields a 10.16% increase in annual wage. In other words, employers believe that education increases the marginal productivity of a worker. Thus, the facts that β is significant and positive establish the validity of the human capital theory in the US labor market. ϰ of also informs that relative educations matters; for natives, each additional year of education above the mean level of education in the US (13.3 years) induces a 15.58% increase in annual wage. This additional increase of 5.42% in annual wage shows that employers care about worker s education level relative to the mean of the home country and that they are willing to compensate those with an education level above the mean at a much higher rate than those below the mean level. Thus, the significance and positive sign of ϰ demonstrates that the signaling theory is also in play in the US labor market. It is also important to note for each year of education above the mean of the home country, the effect of relative education is accounts for around one-third of the total change in annual wage, which demonstrates that the signaling is an important mechanism at play in the US labor market. The negative signage of θ validates Chiswick and Miller s thoughts on skills mismatch between the immigration population and the US labor market. θ of shows us that, ceteris paribus, immigrants are compensated at a lower level that US natives. Specifically, immigrants are compensated 1.54% lower than US natives for each year of additional education. Again, this mismatch could be caused by language barriers, licensing obstacles, discrimination and employer s discount of previous experience from the home country of the immigrant. 25

27 Table 6 Regression Result. reg lnincome schyr talent_dist schyr_immigrant Black_Negro American_Indian Chinese_Japanese Other_Asian Multi_Racial > militaryservice married Source SS df MS Number of obs = F( 10,839491) = Model Prob > F = Residual R-squared = Adj R-squared = Total Root MSE = lnincome Coef. Std. Err. t P> t [95% Conf. Interval] schyr talent_dist schyr_immi~t Black_Negro American_I~n Chinese_Ja~e Other_Asian Multi_Racial militaryse~e married _cons Section V: Closing Remarks Considerations This study faces several limitations. First of all, due to limited data availability, I was not able to include experience into the model, which undoubtedly introduced omitted variable bias into the results. When making a decision regarding hiring and compensation, employers place the same, if not more, emphasis on experience as on education. Therefore, by omitting experience from the model, the results could be distorted. It is possible to create a proxy of a subject s previous work experience through using his / her age. We can make the assumption that all subjects included in the dataset have constantly worked after they completed their education. WE can then produce a synthetic measure of experience by subtracting the years of education that an individual obtained from his / her age. However, I have decided not to do so as previous studies, such as that of Kroch and Sjoblom (1994), have demonstrated that this approximation is misleading as it would tend to inflate the coefficient on years of schooling and favor the human capital hypothesis (Kroch and Sjoblom). Another challenge that this study faces is in the categorization of immigrant. In this study, for sake of simplicity, I categorized anyone who was born abroad as an US immigrant and 26

28 anyone who was born inside the US as a native. However, the more accurate categorization would have been according to where the subjects attend school, since there are immigrants, such as myself, who was born abroad but immigrated to the US at an early age and was completely educated in the US. Under the study s categorization, this group would have been compared against the mean level of education in their home country, when they should have been compared to the mean level of education in the US. It is unclear what direction this inaccuracy affects the model, but future researchers should take care to collect data regarding the subject s age upon education and the location of the schooling. Conclusion In this study, I have proposed a new way of testing the human capital and signaling theories of the value of education. If education acts as a signal in the US labor market, then a worker s level of education relative to his / her cohort should be significant predictor of his / her income. This finding would give evidence of the signaling theory. We constructed the measure of relative education by subtracting the mean number of years of education in a worker s home country from his / her highest level of educational attainment. This variable would thus tell us if there is a difference the marginal return of education for workers above and below the mean. If such a difference exists, then we can conclude that firms / employers view education as a signal of innate talent, giving further proof to the validity of the signaling theory. The regression results demonstrate that both absolute and relative measures of education are extremely statistically significant predictors of a worker s income. The fact that the relative measure accounts for around one-third of the change in worker s wage change further demonstrates the validity of the signaling theory. In addition, the negative signage of the interaction term between absolute measure of education and the immigration tells us that 27

29 immigrants experience a higher degree of skills mismatch than US natives, as predicted by Chiswick and Miller. I will now turn to implications of this study s finding. First of all, the validity of the human capital theory tells us that education does provide significant value to workers, as well as the market. Thus, this finding provides evidence for increasing public funding and investment into education systems. Secondly, the fact that signaling theory is shown to be valid in this study demonstrate that firms / employers care about the relative level of a worker s education because they believe that education is an accurate measure of innate ability. Hence, US should seriously consider migrating to a points-based skill system to admit more highly educated and skilled workers. The points-based skills system would provide a fairly accurate measure of a worker s level of productivity through examining potential immigrants level of education and work experience. 28

30 References Becker, Gary S. Human Capital: A Theoretical and Empirical Analysis with Special Reference to Education. 2nd ed. Chicago: U of Chicago, Print. Chiswick, Barry R., and Paul W. Miller. "Educational Mismatch: Are High-skilled Immigrants Really Working at High-skilled Jobs and the Price They Pay If They Aren t?" IZA Discussion Paper Series (2009): n. pag. Print. Greenstone, Michael, Adam Looney, and Harrison Marks. The Hamilton Project. The Hamilton Project. N.p., May Web. 17 Nov Johnes, Geraint. "Human Capital versus Sorting: New Data and a New Test." Applied Economic Letters 5.2 (1998): n. pag. Taylor and Francis. 2 Nov Web. 31 Jan Jones, Patricia. "Are Educated Workers Really More Productive?" Journal of Development Economics 62 (2001): Jstor. Web. 15 Jan Kroch, Eugene A., and Kriss Sjoblom. "Schooling as Human Capital or Signal: Some Evidence." The Journal of Human Resources 29.1 (1994): Jstor. University of Wisconsin Press. Web. 03 Apr Mincer, Jacob. Schooling, Experience, and Earnings. New York: National Bureau of Economic Research; Distributed by Columbia UP, Print. "Points-Based Immigration Systems: Canada." Library of Congress Home. The Law Library of Congress, Mar Web. 09 May Spence, Michael. "Competition in Salaries, Credentials, and Signaling Prerequisites for Jobs." The Quarterly Journal of Economics 90.1 (1976): Jstor. Web. 28 Jan Spence, Michael. "Job Market Signaling." The Quarterly Journal of Economics 87.3 (1973): Yale University. Yale University. Web. 3 Feb

31 "U.S. Immigration Before 1965." History.com. A&E Television Networks, n.d. Web. 08 May "USCIS Reaches FY 2015 H-1B Cap." U.S. Citizenship and Immigration Services. U.S. Citizenship and Immigration Services, 10 Apr Web. 09 May

32 Appendix Table 1 Immigrant-Specific Talent Regression. reg lnincome talent_dist Black_Negro American_Indian Chinese_Japanese Other_Asian Multi_Racial militaryservice mar > ried if immigrant==1 Source SS df MS Number of obs = F( 8,107126) = Model Prob > F = Residual R-squared = Adj R-squared = Total Root MSE = lnincome Coef. Std. Err. t P> t [95% Conf. Interval] talent_dist Black_Negro American_I~n Chinese_Ja~e Other_Asian Multi_Racial militaryse~e married _cons

33 Table 2 Immigrant-Specific Education Level Regression. reg lnincome schyr Black_Negro American_Indian Chinese_Japanese Other_Asian Multi_Racial militaryservice married if > immigrant==1 Source SS df MS Number of obs = F( 8,114145) = Model Prob > F = Residual R-squared = Adj R-squared = Total Root MSE = lnincome Coef. Std. Err. t P> t [95% Conf. Interval] schyr Black_Negro American_I~n Chinese_Ja~e Other_Asian Multi_Racial militaryse~e married _cons

34 Table 3 Native-Specific Education Level Regression. reg lnincome schyr Black_Negro American_Indian Chinese_Japanese Other_Asian Multi_Racial militaryservice married if > immigrant==0 Source SS df MS Number of obs = F( 8,732358) = Model Prob > F = Residual R-squared = Adj R-squared = Total Root MSE = lnincome Coef. Std. Err. t P> t [95% Conf. Interval] schyr Black_Negro American_I~n Chinese_Ja~e Other_Asian Multi_Racial militaryse~e married _cons