Within Income and Education Inequality Characteristics of Immigrants in the United States: Social Mobility and Adaptation

Transcription

1 Within Income and Education Inequality Characteristics of Immigrants in the United States: Social Mobility and Adaptation Mehmet E. Yaya August, 2012 ABTRACT Until recently, adaptation and assimilation of immigrants are viewed as an individualistic process rather than community-based in the literature. However, Hatton and Leigh (2011) have showed that immigrants integrate to the host country as communities not as individuals; hence group characteristics play as much of a role as individual characteristics on the level of adaptation. In this paper, I introduce another dimension to the host of immigration adaptation proxies that reflects a specific community-based characteristic: income and education inequality. I assert that inequality reflects not only social mobility of immigrant groups, but it can also be used as a proxy of adaptation. I argue immigrant groups that exhibit income and education distributions similar to the natives can be regarded as well adapted groups while those who exhibit vastly different distribution should be considered as groups with limited adaptation process. Using the American Community Survey (ACS) 2010 cohort, I present the initial findings on the within-group inequality for immigrants who came from different regions of the world and countries. Moreover, I employed a cross-sectional regression to analyze the determinants of income and education inequality across immigrant groups. The results suggests that immigrant community-based characteristics such as income, education, and number of years that immigrants spent in US can indeed explain most of the variation in withingroup inequality. Keywords: Inequality, Immigration, Social Mobility, Adaptation JEL classifications: D31, D63, F22 Assistant Professor of Economics, 703M Pray Harrold Hall, Department of Economics, Eastern Michigan University, Ypsilanti, MI 48197, myaya@emich.edu.

2 I. INTRODUCTION Inequality can be defined as the economic attainment differences between individuals in a population. It is an Anglo-Saxon problem that has important social mobility implications (Champernowne and Cowell 1998). Historically, inequality evolves differently across the countries. For instance, income inequality is fairly flat in Germany but it has been increasing in the last 30 years in the United States. Although it is well documented in the literature, there is ample disagreement about the causes of such increase in income inequality in US. For instance, Borjas (2010) attributed the increase in inequality to supply shifts such as the effect of immigration, international trade, skilled-based technological change, and institutional changes in the U.S. labor market. Furthermore, McCall (2000) suggest that within-group income inequality, inequality between racial and gender groups, also contributed to the rise in inequality in United States. Education inequality, on the other hand, is closely related to social mobility and researchers paid close attention to it. For instance, Blanden and Machin (2004) have analyzed the trends in education inequality in UK and found higher education expansion has not been equally distributed across people from richer and poorer backgrounds during the years of 1970s to 1990s. Inequality is an important economic issue due to its implications on the social mobility of people, especially at the bottom of the distribution. Equality of opportunity is an important concern in the societies since the variation in social mobility is aligned with the variation in inequality where low intergenerational mobility is observed with high inequality. Individuals in societies with high inequality find it much harder to change their position in the social hierarchy (Andrews and Leigh 2009). In these societies, the economic cycle breaks down when employees can't afford to buy all the products they actively participate in the production process. Therefore, 1

3 inequality leads to polarization in societies in which members of the society tend to engage in group-specific rent seeking (Siebert 1998). 1 Consequently, these social problems provide a rationale for policymakers closely following the evolution of inequality (Nolan 2009). Immigrants approximately 36 million, approximately to 12% of the U.S. population makes the U.S. the largest single destination country in the world. 2 They are important source of population growth; the high skilled ones add to the overall productivity, and their incomes contribute to private sector consumption and to public sector tax payments. These high skilled immigrants are also less likely to be unemployed (Greenwood and McDowell 2011). Literature on immigrants is vast, from their labor market impact to their adaptation process, i.e. assimilation. In this string of literature, a considerable amount of attention has been given to the immigrant assimilation in terms of economic, social, and civic adaptation. 3 Labor market outcomes such as employment, wages, and income are generally accepted as the economic assimilation proxies while language use, martial status, and family structure are the proxies for social assimilation. Finally, naturalization rates and voting behavior of immigrant groups are regarded as the civic adaptation proxies. Interestingly for the most part, adaptation and assimilation are viewed as individualistic rather than community-based in the literature. These studies used personal income, language ability, education, investment and health insurance preferences as a proxy of such adaptation process to the host country. However, Hatton and Leigh (2011) have recently showed that immigrants integrate to the host country as communities not as individuals; hence group 1 Winkelmann and Winkelmann (2010) have examined the impact of inequality on middle class and found that there indeed exist an inverse relationship between inequality and satisfaction as well Yearbook of Immigrant Statistics and Department of Homeland Security estimates show an additional 10 million unauthorized immigrants. 3 Vidgor (2010) presented a general overview of the types of assimilation that immigrants are experiencing in this recent article. 2

4 characteristics rather than individual characteristics play an important role on the level of adaptation. In this paper, following Hatton and Leigh s assertion of community-based adaptation, I propose to add yet another dimension to the host of immigration adaptation proxies that reflects a specific community-based characteristic: income and education inequality. I assert that within-group inequality reflects not only social mobility of immigrant groups, but it can be used as a proxy of how much an immigrant group progresses on the adaptation process. Hence, this paper argues immigrant groups that exhibit similar income and education distributions compared to the natives in the host country can be regarded as groups that are well adapted while those exhibit vastly different distributions should be considered as groups with limited adaptation process. The paper presents evidence on three distinct areas that are closely related to inequality, social mobility, and adaptation. First, it provides a brief comparison of the income and education inequality characteristics of immigrants and natives in the United States. The findings can be used to compare and contrast the social mobility of immigrants and natives. Next, it illustrates descriptive inequality statistics within immigrant subgroups based on the world region and the country of origin. It is fairly well documented that immigrants are not a homogenous group. The paper presents further evidence that the immigrants from different regions are heterogeneous, especially in terms of inequality, which can be used as a proxy of assimilation. Finally, the paper examines the determinants of within-group inequality of immigrant subgroups based on the country of origin. The descriptive statistics suggests that there are significant differences in within-group inequalities across these immigrant subgroups. The fundamental claim is that these immigrant subgroups live in the same host country, so the community-based 3

5 demographic factors should contribute to the variation in within-group inequalities; hence determines the social mobility of immigrants in the United States. The rest of the paper is organized as follows: section two reviews the literature on immigration and inequality. Section three presents the data, and section four presents inequality characteristics of immigrants. Section five contains the model specification; section six presents the empirical study on immigrant subgroups and section seven concludes. II. LITERATURE REVIEW United States hosts the largest number of immigrants in the world so it is not surprising that there are many studies published in the literature that have examined the socio-economic conditions of immigrants. Among those, some have analyzed the assimilation path of immigrants. These studies examined income, language ability, education, investment and health insurance preferences as a proxy of adaptation process to the host country. For instance, Borjas (1985) have used wages as a proxy of economics adaptation to the host country. Amuedo- Dorantes and de la Rica (2007) considered immigrants labor market outcomes such as employment likelihood, occupational attainment and occupational upward mobility as a proxy of adaptation. Meng and Gregory (2005) have shown that intermarried immigrants earn significantly higher incomes than endogamously married immigrants. The common feature of these papers is the fact that immigrant adaptation, i.e. assimilation, is explained with observed characteristics of individuals. In this paper, following Hatton and Leigh (2011), assimilation is assumed to progress at the community level. Hence, I used immigrant groups based on world region and country of birth as the underlying community definition and examined a certain community characteristic (income inequality) rather than individual characteristics to gauge the 4

6 adaptation level of these groups to the host country. I assert that the degree of adaptation by each of these immigrant groups is approximated with the level of within-group inequality. 4 Most of the earlier studies in the literature that has examined income inequality differences among immigrants and natives are descriptive in nature and did not extend to the issues of social mobility and adaptation. For instance, Moore and Pacey (2003) and Basavarajappa (2000) used Canadian immigration data for the previous two decades and identified significant differences between the characteristics of immigrants and natives Canada. Moore and Pacey (2003) found that immigrants education level in Canada was much higher than that of the native-born population. On the other hand, Basavarajappa (2000) examined the older immigrants in Canada and found that immigrants as a group have roughly the same income distribution as their Canadian-born counterparts; however income distribution within the immigrant subgroups differs considerably. Gray, Mills, and Zandvakili (2003) have also studied income inequality of immigrants in Canada. The latter two of these studies found that significant inequality differences between natives and immigrants. Their findings suggested that most of the observed inequality was due to income dispersion within each group in Canada. However, immigration studies with the exception of Basavarajappa have not examined the within-group income inequality of immigrants; hence they did not provide substantial discussion on the adaptation and social mobility of immigrants. Furthermore, these two studies did not extend to specific immigrant groups from different regions or countries of birth; hence they did not address heterogeneity of immigrants. Income and education inequality, i.e. social mobility that are used as adaptation proxies in this study, depend several different factors. Literature has documented a significant 4 Hoover and Yaya s (2010a) study is one of the few that examined the gender differences in within-group income inequality of immigrants in the United States. 5

7 relationship between economic development and inequality. For instance, in his seminal work, Kuznets (1955) proposed an inverted U-shaped relationship between income inequality and economic development. He argued that income inequality increases up to the threshold level of economic development. Only after this threshold is passed does income inequality start to decrease. Nielsen and Alderson (1997), however, suggested a great U-turn in the relationship between economic development and inequality. The authors argued that, due to changes in economic dynamics such as sectoral, gender-based, and racial dualism, the relationship between development and inequality in the United States has reversed since Aigner and Heins (1967) introduced median school years completed, unemployment rate, median population age, and mean family income as determinants of inequality. They found that the median age of the population negatively affected income inequality because of the older population s high degree of skill rigidity. The authors argued that these older people enforced and proliferated equality barriers. Aigner and Heins also showed that family income had a positive effect on income inequality, but unemployment had mixed effects. Furthermore, Odedokun and Round (2004), who studied factors affecting income inequality in Sub-Saharan African countries, used economic development, regional factors, government spending, inflation, unemployment, corruption, openness to trade, and education to explain income inequality in their cross-sectional study. III. DATA For the empirical study, I have used the American Community Survey (ACS), 2010 cohort. The ACS data set includes detailed social, demographic, and economic information about approximately three million households. Data is collected in all 3,141 US counties and American Indian, Alaska Native, and Native Hawaiian areas where population is greater than 6

8 55,000. Following Aly and Ragan (2010), immigrants are defined as the individuals who were born to non-u.s. parents in a country that is other than United States. Based in this classification, respondents in the survey fall into two categories: the first category is composed of natives, who were either born in one of the 50 states or US territories or born to US parents. The natives account for 85.6% of the sample. The second category is comprised of immigrants. 5 These immigrants are either citizens of other countries (almost 55% of immigrants surveyed) or naturalized U.S. citizens (the remaining 45%). In ACS survey, the total number of useable observations was 1,910,353 where the number of immigrants was 275,388. The ACS is an annual survey that provides researchers with a large set of information about both natives and immigrants: income, education, English language ability, naturalization status (of immigrants), employment status, year that the immigrant entered the United States, country of birth, gender, etc. Income is calculated as the sum of eight different sources of income: wages or salaries; net self-employment; interest, dividends, net rental, or royalties; estates and trusts, social security, or railroad retirement; Supplemental Security Income (SSI); public assistance or welfare payments; retirement, survivor, or disability pensions; and all other income. The annual ACS survey provides information about the educational attainment of respondents as well. Educational attainment is recoded as ranked order where zero value corresponds to less than high school; one corresponds to less than college; two corresponds to some college; three corresponds to a college degree; and four corresponds to an advanced degree. Number of years spent in US and Language use are the other immigrant-specific variables. The former can be calculated by the difference between the year the immigrant 5 In this classification, the study considers non-immigrant visa holders as immigrants because the survey does not allow differentiating between types of visas individuals hold. 7

9 entered the United States for the first time and the survey year. The latter measures if a language other than English is being frequently spoken in the household. As mentioned above, the immigrants in the survey are divided into two groups. Immigrants who are citizens of the United States through naturalization constitute the first group. These people have all the residency and employment rights of ordinary U.S. citizens and do not have any restrictions on their labor market participation and earning income. Immigrants who are citizens of other countries constitute the second group. Some of these people have limited employment rights (F-1, H-1B, E-1, E-2, and E-3), and some do not have a legal right to work during their stay in the United States (F-2, H-5, and B-1). Amuedo-Dorantes and Mundra (2007) suggested that the migrants legal status affect their earnings independently of their personal and human capital characteristics. Therefore, I have constructed a variable that shows the percentage of naturalized within the immigrant sub-group to separate these two types of immigrants (noncitizens and naturalized citizens). Furthermore, I have utilized the immigrants employment status which gives employment information during the last three months of their stay in the United States. IV. CHARACTERISTICS OF IMMIGRANTS Standard descriptive statistics of inequality such as relative mean deviation, coefficient of variance and standard deviation of logs offer limited information about the shape of the distribution. Moreover, most of them fail to capture the higher moments of income and education distribution sufficiently. Although, these simple statistics have been used as an inequality metric in the literature, they have been heavily criticized for their inability to capture the pattern of distribution among the extremes (Temkin, 1993). Rather than exploring how to choose an appropriate inequality measure, this paper derives results and their policy implications 8

10 from the following inequality measures. I used the following inequality metrics and respective calculation methods: Gini coefficient (GINI), GI NI 1 2n 2 n n i 1 j 1 y i y j Theil Index (THEIL), T H E I ll n y () l n g ( ) m l n y g m Atkinson s inequality measures (ATK) with inequality aversion levels, ε = / 1 ) ( 1 1 y At ki nseon 1 i n i 1 n where μ is mean income or education, n is the number of individuals in the group, y i is the income or education of the i th individual in the group, y j is income or education of the j th individual in the group, GM is the geometric mean. As stated in Campano and Salvatore (2006), the Theil addresses a very important socioeconomic question that is based on human characteristics. For large populations, almost all human characteristics are normally distributed and hence symmetric distributions. The Theil Index measures the movement from the norm or symmetry, which is a critical characteristic of distributions. Atkinson Index, on the other hand, incorporates the inequality aversion, measured with a parameter called epsilon. The larger the parameter, the more inequality averse the Atkinson Index becomes. Larger values of parameter make the index more sensitive to the transfers at the bottom of the distribution. Since, income transfer is beyond the scope of this paper, the parameter ε is selected consistent with what the Census Bureau reports. Finally, the Gini Coefficient is bounded with zero and one and the larger the coefficient, the greater the 9

11 inequality is. The coefficient is commonly used in the literature and it has wide appeal since it is more easily understood and calculated (Hoover and Yaya, 2011). A. Income Inequality of Characteristics of Natives and Immigrants Table 1 shows the descriptive statistics on income, education, and inequality of the respondents in ACS Natives have higher median income than their immigrant counterparts. The income difference between these groups is approximately 27.5%. Moreover, natives have higher educational attainment than immigrants as shown in Table 1.a. 6 Table 1.b presents the percent distribution of natives and immigrants. As it can clearly be seen from the table, immigrants have much higher percentage with less than high school degrees while they have slightly higher percentage of advanced degree holders. This suggests a higher education inequality among immigrants. Table 2.a-b shows the distributional characteristics of income and education among the natives and immigrants. 7 The top quadrant of the table suggests that income is more unequally distributed among immigrants than natives. Figure 1 also confirms that income distributions of both natives and immigrants were skewed to the right, as it is suggested by a lower median income compared to mean income but more so for immigrants. These differences are statistically different from each other based on t-test with unequal sample size and variance. The lower quadrant table also shows the inequality in education among natives and immigrants. Immigrants have higher inequality in education which coincides with higher income inequality for immigrants compared to natives. [Table 1 and 2] 6 Sana (2010) suggests that the scientists and engineers ratio between foreign-born and natives doubled from 0.16 in 1994 to 0.32 in These statistics are very comparable with the Census Bureau reports related to Gini, Theil, and Atkinson with inequality aversion level of

12 [Figure 1] Immigrants are then split into six different subgroups based on the regions in the world: Latin America, Asia, Europe, Africa, North America, and Oceania. The Latin American region includes all immigrants from South and Central America, including Mexico and the Caribbean island countries. Canada and Bermuda are the only two countries in the North American region. Australia, New Zealand, Fiji, Micronesia, Tonga, and Samoa are included in the Oceania region. Table 2 demonstrates the sample statistics of immigrants based on world region and presents a great deal of evidence on the heterogeneity of immigrants. The table suggests that 51% of immigrants in the United States are from Latin America. Latin American immigrants have, on average, the lowest education and income compared to other regions. Latin American immigrants also pull the overall average of income and education of the immigrants down. North American immigrants all from Canada and Bermuda, on the other hand, have higher levels of education and income than any other immigrant groups. Immigrants from Africa have the lowest median income after Latin Americans. The lower part of table shows a clearer picture of educational attainment among immigrants. Latin Americans have the highest percentage of less than high school while Asian, European, and North American immigrants have the highest percentage of advanced degree holders. [Table 3] Table 4 shows the income inequality metrics for these six regions listed above. Although the immigrants from the Latin American region have the lowest income and education level, their income is the most equally distributed one. This could be explained by a large cluster of Latin American immigrants earning low incomes. The results are consistent across the three inequality metrics. On the other hand, education inequality is significantly higher for Latinos 11

13 compared to other groups supporting the heterogeneous characteristics of immigrants especially visible in income and educational attainments of immigrants from Latin America. [Table 4] Extending these findings to the adaptation process of each immigrant group from different regions of the world suggest some interesting findings. For instance, compared to natives, the average income and education is closest among the immigrants from North America (mainly Canada) and Oceania (mainly Australia and New Zeeland). Not surprisingly, the income inequality metrics of these immigrant groups resemble very much to those natives as well. In contrast, immigrants from Latin America and Africa exhibit substantially different income and education patterns which is reflected on a significant difference on income and educational inequality; hence the level of adaptation. B. Income Inequality of Immigrants Based On Country of Origin Immigrants, then, are divided into subgroups based on their country of origin. 8 These countries are those recognized by ACS and each group contains a sample size of fifty or more immigrants. 9 There are also few immigrant groups that are from smaller countries that are not recognized by ACS but lumped in groups based on world regions. The largest immigrant group was from Mexico, which constituted almost 29% of all immigrants in the United States. Mexican immigrants have the lowest education levels of those are the least-educated households in from the Latin American region. This finding is comparable to Census averages, which Card s (2009) paper recently presented. Immigrants from the Philippines, India, and China were the other large immigrant groups. Australian, South African and British immigrants 8 Sample statistics of immigrants income and education from different countries are available from the author upon request. 9 This excluded Guinea, Iceland, and the West Indies. 12

14 had the highest mean and median income, while immigrants from Somalia and Yemen had the lowest. Then, the income inequality metrics that are introduced above were calculated for these immigrant subgroups. As it is mentioned above, there are significant differences in within income inequality across these immigrant subgroups. Figure 2 shows these subgroups in descending order. As the figure suggests, the Gini coefficient for personal income has a minimum of 0.47 (Senegal) and maximum of 0.70 (Saudi Arabia) across these subgroups. [Figure 2] Close examination of these ranked inequalities across the subgroups also suggests an interesting story about the social mobility and the degree of adaptation among these immigrant groups. For instance, immigrant groups from Barbados, Bermuda, El Salvador have significantly more equal distribution than natives while immigrant groups from Saudi Arabia, Kuwait, and Yemen has very unequal distribution compared to natives. I assert that the immigrants from these countries have a long way to adapt to the US. In comparison, immigrants from Australia, England, Canada has very similar income inequalities compared to natives and therefore they can be considered as well adapted immigrant groups in the United States. Furthermore, assume that there are two immigrant groups residing in the United States with two distinct inequality levels. These two subgroups have the same set of opportunities available, i.e. social mobility, to every other immigrant subgroup. Unless there is some sort of a nationwide discrimination against one of these subgroups, general level of macroeconomic factors such as unemployment rate, per capita income, or business cycles in the host country should not explain the difference in their within-group inequalities. Therefore, the possible explanatory variables should be specific to the particular immigrant subgroup as suggested by Hatton and Leigh (2011). This observation leads to an important question that this paper endeavors to answer: If immigrant subgroups live in the 13

15 same host country and have similar mean and median incomes and if common factors in the country of destination should not explain the variation in within income inequalities, what are the significant subgroup specific variables that explain the within-group variation? This question leads to an important path to understand the immigrant groups (Latino or non-latinos) and degree of their adaptation to the United States. Policymakers can better evaluate the underlying factors forcing such disparity in adaptation process and craft policies targeting these populations. V. MODEL SPECIFICATION The previous section showed that income inequality among immigrant subgroups varies significantly due to the heterogeneity of immigrants in US. This section employs a crosssectional regression analysis to determine the factors that cause inequality differences among immigrant subgroups. Following an augmented model specification by Hoover and Yaya (2010b), I used the regression model below: Inequality i Income Education i i Language Unemployment % Naturalized i 7 i % Females # of Years 3 i i i 4 i (1) where y i is one of the income or education inequality metrics introduced in the previous section. As mentioned in the previous section, each inequality metric depicts a different characteristic of the distribution of the underlying subgroup. Therefore, as a robustness check for the explanatory variables, three inequality measures (Gini, Theil, and Atkinson) are used separately as dependent variables in the successive regressions. One of the challenges of examining the determinants of inequality is exposing the relationship between economic development and inequality. If all immigrant subgroups reside in 14

16 the same host country, one cannot use host country-specific income to approximate the economic development level of those immigrants. Therefore, this study employs median income as an explanatory variable that is assumed to be the proxy of development of each immigrant subgroup as in Aigner and Heins (1967). The descriptive statistics of dependent variables are given in Table 5. The top part of the table shows the income based inequality metrics while the bottom part shows the education inequality metrics. The number of observations is 152 and these within inequality metrics are calculated for the ACS recognized subgroups. Income inequality metrics are naturally higher since the underlying factor (income) a continuous variable while education is rank ordered with values between zero and four. On the other hand, the standard deviation of education inequality is higher suggesting more disperse distribution of education within immigrant subgroups based on country of origin. [Table 5] As far as the explanatory factors of inequality, income related variables include the income and the squared income term to capture a possible nonlinear relationship between income and inequality. Earlier studies in income inequality used either mean or median income as an explanatory variable, and there is no consensus on which statistic to select. However, due to its skewed distribution, over the last 40 years median income has become widely accepted in the literature. 10 Following earlier research, median personal income is selected as an explanatory variable over mean income. Nielsen and Alderson s (1997) work, the nonlinear relationship between income and inequality is examined for sign and significance. They suggest a U-shape relation between these variables. Figure 3 depicts the scatter diagram of these immigrant subgroups with respect to their income and inequality. Each observation is demonstrated based on the size of the group. A larger filled circle represents a larger immigrant subgroup. There is, 10 See Gardner (1969), Betz (1974), Danziger (1976), Davies and Heather (1998), Lee and O Leary (2008). 15

17 indeed, a visible U-shaped link between income and inequality as suggested by Nielsen and Alderson (1997). [Figure 3] Immigrant group characteristics include mean values of the following variables for the subgroups: education, number of years spent in US, English use at home, percentage of females, unemployment rate, and percentage of naturalized immigrants. Some of these variables have been extensively used in the literature, but English language use, year of entry, and nativity are important immigrant group-specific factors that have not been explored as a determinant of inequality in the past and have the potential to capture the variation in within-group income inequality. There is also extensive research studying the types of workers who are more likely to migrate. Borjas (1987) suggested that if a strong correlation existed between immigrants expected income in the country of origin and in the country of destination, and if the income in the country of destination was more unequally distributed than in the country of origin, then a positive selection would be observed. Borjas (1987) and Chiswick (1978) both argued that immigrants more motivated than the citizens of a country origin and are a self-selected group. According to the authors, immigrants income surpasses that of native populations because immigrants are a select not random group of source country and have stronger investment incentives than native workers. 11 Cohen and Haberfeld (2007) examined the immigrants from former Soviet Union in the United States and found that these immigrants are positively selected. This finding is consistent with Borjas and Chiswick s assertions since income inequality in United States is one of the highest among the developed countries and the county became a 11 Hatton (2004) empirically tested the positive selection hypothesis with UK migration data. Hatton found an insignificant effect of income inequality, measured by the Gini coefficient, on immigration, suggesting that the UK income inequality level did not affect the decisions of immigrants migrating to the United Kingdom. 16

18 magnet for positively selected immigrants from countries that have more equally distributed income inequalities. This is especially true for immigrants coming from developed countries who are not only positively selected but obtain skills that are more likely to be transferred to the United States. For this reason, a dummy variable for countries that are members of the Organization of Economic Development and Cooperation is also considered. The regional analysis of immigrants from these countries in the previous section shows that immigrations from these regions have more education and income in the country of destination. Furthermore, it is possible that immigrants from politically and economically suppressed countries are more willing to migrate from their country of origin to another country. 12 In order to account for such differences across countries, I used Freedom House, Inc. (2005) political freedom index which takes values between one and seven, with one corresponding to complete freedom and seven to no freedom. The remaining unobserved time invariant geographical differences across these immigrant subgroups that are not captured by the explanatory variables are controlled with dummies for each continent. These variables are South America, Asia, Europe, Africa, North America, and Oceania. 13 [Table 6] Table 6 shows descriptive statistics of the independent variables that are used to explain the variation in inequalities among immigrant subgroups. Average of the median income across the subgroups is $24,312. Immigrant subgroups, on average, have attained associates degrees and/or attended some college. They have also spent an average of years in the United States. The percentage of females is 51.67% and percentage of naturalized immigrants across 12 Sylwester (2004) used political stability to explain the cross-country variation of income inequality measured by the Gini coefficient. 13 Bermuda and Canada are the only two countries in the North American group. 17

19 the subgroups is approximately 51.91%. Moreover, 9.55% of them are unemployed and 73% of those groups are using a language other than English at home. VI. RESULTS Table 7 shows the pairwise correlations along with statistical significances for the explanatory variables. Close examination of the table suggest that there should be no concern for multicolliniarity issue and correlation coefficients take the expected signs for most of the cases. For instance, income is positively correlated with education and negatively with unemployment rates within the immigrant subgroups. Furthermore, the longer the immigrants stay in US, the higher income they earn. Immigrants groups from OECD countries or politically stable countries earn more and have lower unemployment rates, consistent with the positive selection theory of Borjas. [Table 7] Table 8 shows the regression results for the empirical study of immigrants in the United States. The regressions use the OLS method with clustering around the size of the immigrant subgroups since the size varies significantly across the immigrant subgroups. Although the average group size is around 2,000 individuals; the standard deviation of the size is hovering around 7,000 immigrants. The income inequality is the dependent variable in 6 consecutive regressions in the table. The dependent variable in the first two regressions is Gini Coefficient, while the next two is Theil Index, and last two is Atkinson Index. The first, third, and fifth regressions have no source country controls such as political stability or OECD member dummy while second, fourth, and sixth include these variables. The number of observations is 152 for regression without source country controls and it is 127 with these controls. R 2 ranges from 58% to 70%. 18

20 The results suggest that income is an important and statistically significant factor determining the within-group inequality of immigrants. Moreover, the relationship between income and inequality is nonlinear and has the U-shape suggested by Nielsen and Alderson (1997). The coefficient for income is negative and income-squared is positive. This result confirms the U-shaped relation depicted in Figure 3. Moreover, gender, education and number of years spent in US are statistically significant. Immigrant subgroups that have higher levels of education or have spent more time in US have higher income inequality. 14 On the other hand, immigrant subgroups with higher percentage of females exhibit lower levels of inequality consistent with the gender inequality literature. This string of literature states that females work on smaller subset of industries and have more equally distributed income compared to males. 15 The political stability variable does not seem to explain the income inequality of across immigrant subgroups. 16 However, immigrants from OECD countries have higher income inequality compared to other immigrant subgroups. Finally, F-test shows that all the geographical control variables equal to zero could not be rejected. [Table 8] The regression estimates for the education inequality metrics are shown in Table 9. All of the models in this table use the OLS robust cluster estimator. The results show that education, language and unemployment are statistically significant for almost all of the education inequality metrics. Education has the negative and significant coefficient for all six consecutive regressions, suggesting that immigrant subgroups that have higher education have lower 14 There are also a sizeable number of immigrants who reside in the United States for higher education purposes. I have captured the status of these immigrants with a variable called percentage of students in the survey. However, the impact of the student status is insignificant. 15 For more information, check Hoover and Yaya (2010a). 16 For robustness check, I have also estimated the same model using the robust regression estimation. The results are almost identical to the OLS. The models have higher R-squared values compared to OLS estimates; however, overall findings of OLS do not change with robust regression estimation. These estimation results are available from the author upon request. 19

21 education inequality within each immigrant group. On the other hand, language variable has a strong and positive effect on education inequality. Immigrant groups that are more likely to talk a different language other than English at home generally have higher education inequality. Gender, i.e. percentage of females, has an impact towards more equally distributed education within immigrant groups. Finally, immigrant groups with higher percentage of naturalized citizens have more equal distribution of education. However, this effect subsides once the OECD dummy enters the regressions in the second, fourth, and sixth models. [Table 9] Consequently, the results suggest that the unique characteristics of immigrants require additional exploration for distinct variables related not only to the national characteristics of immigrants, but also to the relative performance of these subgroups in terms of adaptation to the host country. The results obtained from the regression analysis are as follows: First, median income, educational attainment, and gender are the significant factors determining income inequality of immigrant subgroups in the United States. Furthermore, the relationship between income and inequality is non-linear. On the other hand, the education inequality mainly depends on the average education, gender, and naturalization status. VII. CONCLUSION The number of immigrants has been steadily rising in the United States. Immigrants now account for more than 12% of the population, and they constitute a large portion of population growth. Consequently, immigrants are at the center of political disputes related to their demographic characteristics. On the other hand, for the last two decades in the United States, the causes and consequences of increasing inequality have still been controversial, and changes in immigration policies add more justification for further immigration and inequality-related 20

22 research. Furthermore, level of social mobility and adaptation by immigrants is also an important piece of information for policy makers. This paper examines the within income and education inequality of immigrants using information from approximately three hundred thousand immigrants surveyed by the ACS in 2010, and it provides extensive results for these broad issues related to immigrants and their inequality characteristics. Several findings are worth summarizing. First, this paper presents a snapshot of the current state of immigrant characteristics while examining the income and education inequality differences between U.S. natives and immigrants. There is strong evidence that income and education is more unequally distributed among immigrants than among natives. The most evident justification for this finding is the difference in characteristics of natives and immigrants. Specifically, U.S. natives have more education than immigrants, earn more income, and most importantly, do not experience market frictions such as language barriers, cultural adaptation periods, etc. This finding has important implication on the lack of social as well as provides a more complete view of degree of adaptation mobility among immigrants residing in US Second, the study examines the determinants of income inequality among immigrant subgroups based on world region and country of origin. Immigrants have unique characteristics compared to natives; hence, the analysis requires additional attention and further control variables to account for these diverse characteristics of immigrants. In addition to the standard explanatory variables such as income, gender, educational attainment, and unemployment rate, this study introduced immigrant group characteristics such as the number of years spent in the States, language use, and naturalization status to explain the variation in income and education inequality among immigrant subgroups. The results suggest that income inequality among immigrant subgroups can best be explained by median income, educational attainment, and 21

23 gender. Furthermore, there is a nonlinear relationship between income and inequality. Furthermore, education inequality depends on the average education level, gender, and naturalization status. The results provide valuable information to researchers and policymakers, since they illustrate the current state of inequality of each immigrant subgroup. Most of the existing studies examined the Hispanic population in the United States in isolation but this more comprehensive immigrant study both reaffirms the earlier findings related to Latin American immigrants and provides additional information on non-latino immigrants that is not available to the researchers. For example, compared to other immigrant subgroups, Mexican immigrants have lower income inequality; hence, convey lower risk of social mobility problems and polarization. However, immigrants from China exhibit much higher income inequality; thus, subject to long-term intergenerational mobility problems. Furthermore, the findings suggest some potential policy tools to the policymakers who regulate both the immigration and redistribution policies. With the additional information provided in this research, policymakers may evaluate the adaptation process of immigrant groups; they may adjust visa and naturalization approvals, some of which are subject to country quotas. One valuable extension of this work would be incorporating additional surveys provided by the ACS and producing a time dimension to the existing methodology. Using an appropriate panel data technique would not only introduce time dynamics, but also would allow for controlling the unobserved heterogeneity across the immigrant groups. However, one shortcoming of incorporating the earlier surveys is the fact that the ACS has only been published for a short period of time and does not track individuals across time. 17 Since income inequality 17 The full implementation of ACS has begun in 2005 with a sample of about three million addresses throughout the United States. 22

24 is persistent over time, a time dimension of a couple of years may have only a limited benefit. Nevertheless, the American Community Survey offers insightful, extensive, and valuable information about the immigrants on a yearly basis; this research will hopefully lead to significant future works on the inequality of immigrants. 23

25 REFERENCES Aigner, D. J., and A. J. Heins On the determinants of income equality. American Economic Review 57(1): Aly, A. E., and Ragan. J. F Arab immigrants in the United States: how and why do returns to education vary by country of origin. Journal of Population Economics 23(2): Amuedo-Dorantes, C. and Mundra, K Social Networks and Their Impact on the Employment and Earnings of Mexican Immigrants. Demography 44(4): Amuedo-Dorantes, C. and de la Rica, S Labour Market Assimilation of Recent Immigrants in Spain. British Journal of Industrial Relations 45(2): Andrews, D., and A. Leigh More inequality, less social mobility. Applied Economics Letters 16(15): Atkinson, A. B On the measurement of inequality. Journal of Economic Theory 2: Basavarajappa, K.G Distribution, Inequality and Concentration of Income among Older Immigrants in Canada. International Migration 38(1): Betz, M. B A comparative study of income inequality in cities. Pacific Sociological Review 17(4): Borjas, G. J Self-selection and the earnings of immigrants. American Economic Review 77(4): Borjas, George J Assimilation, changes in cohort quality, and the earnings of immigrants. Journal of Labor Economics 3(4): Borjas, G. J Labor Economics. New York: McGraw-Hill/Irwin. 24

26 Blanden, Jo, and Stephen Machin Educational Inequality and the Expansion of UK Higher Education. Scottish Journal of Political Economy 51(2): Campano, Fred, and Dominick Salvatore Income Distribution. New York, New York: Oxford University Press. Card, D Immigration and inequality. American Economic Review: Papers & Proceedings 99(2):1 21. Champernowne, D. G., and F. A. Cowell Economic inequality and income distribution. New York: Cambridge Univ Press. Chiswick, B. R The effect of Americanization on the earnings of foreign-born men. Journal of Political Economy 86(5): Cohen, Y., and Haberfeld, Y Self-Selection and Earning Assimilation: Immigrants from the Former Soviet Union in Israel and the US. Demography 44: Danziger, S Determinants of the level and distribution of family income in metropolitan areas. Land Economics 52(4): Davies, H., and J. Heather Gender and income inequality in the UK, : The feminization of earnings or of poverty? Journal of the Royal Statistical Society 161(1): Gardner, B. L Determinants of farm family income. American Journal of Applied Agricultural Economics 51(4): Gray, D., J. Mills, and S. Zandvakili Immigration, assimilation and inequality of income distribution in Canada. Univ of Cincinnati, Department of Economics, Economics Working Papers Series. 25

27 Greenwood, M. J., and McDowell, J. M USA immigration policy, source-country social programs, and the skill composition of legal USA immigration. Journal of Population Economics 24: Gujarati, D. N Basic Econometrics. 4th ed. New York: McGraw-Hill/Irwin. Hatton, Timothy J., and Andrew Leigh Immigrants assimilate as communities, not just as individuals. Journal of Population Economics 24 (2) (October): Hatton, T. J Explaining trends in UK immigration. Journal of Population Economics 18: Hoover, G. A., and M. E. Yaya. 2010a. Gender differences in income inequality among immigrant populations to the United States. Social Science Journal 47: Hoover, G. A., and M. E. Yaya. 2010b. Racial and Ethnic Differences in Income Inequality across U.S. Regions. Review of Black Political Economy 37(2): Hoover, G. A., and M. E. Yaya Racial and Ethnic Income Inequality Responses to a Government Maintenance Program in the US. Public Finance Review, 39(3), 2011, Jousseeuw, P. J., and A. M. Leroy Robust regression and outlier detection. San Francisco: Wiley. Kuznets, S Economic growth and income inequality. American Economic Review 34(1):1 28. Lee, S., and J. O Leary Determinants of income inequality in U.S. nonmetropolitan tourism- and recreation-dependent communities. Journal of Travel Research 46: Liebig, T., and A. Sousa-Poza Migration, self-selection and income inequality: An international analysis. Kyklos 57:

28 McCall, Leslie Explaining Levels of Within-Group Wage Inequality in U.S. Labor Markets. Demography 37(4): Meng, Xin and R. G. Gregory Intermarriage and the Economic Assimilation of Immigrants. Journal of Labor Economics 23(1): Moore, E. G., and M. A. Pacey Changing income inequality and immigration in Canada, Canadian Public Policy 1: Nielsen, F., and A. S. Alderson The Kuznets curve and the great U-turn: Income inequality in U.S. counties, 1970 to American Sociological Review 62(1): Nolan, Brian Income Inequality and Public Policy. The Economic and Social Review 40(4): Odedokun, M. O., and J. I. Round Determinants of income inequality and its effects on economic growth: Evidence from African countries. African Development Review 16(2): Sana, Mariano Immigrants and Natives in U.S. Science and Engineering Occupations, Demography 47(3): Siebert, Horst Commentary: Economic consequences of income inequality. Symposium of the Federal Reserve Bank of Kansas City on Income Inequality: Issues and Policy Options, Stark, O Inequality and migration: A behavioral link. Economic Letters 91: Sylwester, K A note on geography, institutions, and income inequality. Economic Letters 85: UN Department of Economic and Social Affairs Population Division International Migration New York. UN Publication (ST/ESA/SER.A/256). 27

29 Vanhoudt, P An assessment of the macroeconomic determinants of inequality. Applied Economics 32(7): Winkelmann, L. and Winkelmann, R Does Inequality Harm the Middle Class? Kyklos 63:

30 TABLES: Table 1.a: Sample Statistics of Natives and Immigrants in the U.S. Sample Size % Median Income Education Natives 1,634, $25, Immigrants 275, $19, Table 1.b: Educational Attainment of Natives and Immigrants, percent of the sample Less than HS Less than College Some College College Advanced Natives 9.37% 28.23% 34.17% 18.40% 9.82% Immigrants 28.19% 21.61% 21.27% 16.93% 12.01% Table 2.a: Income Inequality Metrics of Natives and Immigrants Gini Theil Atkinson eps=0.50 Natives Immigrants Table 2.b: Education Inequality Metrics of Natives and Immigrants Gini Theil Atkinson eps=0.50 Natives Immigrants Table 3.a: Sample Statistics of Immigrants Sample Size % Median Income Education Immigrants Latin America 140, % $16, Asia 85, % $24, Europe 31, % $30, Africa 10, % $22, North America 5, % $35, Oceania 1, % $26, Table 3.b: Educational Attainment of Natives and Immigrants, percent of the sample Less than HS Less than College Some College College Advanced Immigrants Latin America 44.05% 25.46% 18.72% 7.87% 3.90% Asia 13.18% 15.36% 21.77% 28.71% 20.97% Europe 8.99% 22.93% 26.35% 21.02% 20.71% Africa 10.80% 19.12% 29.84% 23.96% 16.29% North America 5.59% 16.87% 29.72% 27.01% 20.81% Oceania 11.23% 24.06% 30.78% 20.36% 13.57% 29

31 Table 4.a: Income Inequality Metrics of Natives and Immigrants Gini Theil Atkinson eps=0.50 Immigrants Latin America Asia Europe Africa North America Oceania Table 4.b: Education Inequality Metrics of Natives and Immigrants Gini Theil Atkinson eps=0.50 Immigrants Latin America Asia Europe Africa North America Oceania Table 5: Descriptive Statistics for Dependent Variables Count Mean St. Dev. Min Max Income Gini Coefficient Theil Index Atkinson Index, eps= Education Gini Coefficient Theil Index Atkinson Index, eps= Table 6: Descriptive Statistics of Explanatory Variables (Determinants of Inequality) Number Mean St. Dev. Min. Max. Median Personal Income 152 $24, $7, $8, $48, Average Education Level Percentage of Females Average Number of Years Language other than English Unemployment Rate Fraction of permanent visa holders

32 TABLE 7: Pairwise Correlations between Explanatory Variable Years Income Education % Female Spent Language Income % Unemp. % Naturalized OECD Political Stability Education * % Female Years Spent * Language * * Unemployment * * * % Naturalized * * OECD * * * * * * Political * * * * * * Stability Number on the top show the correlation coefficient and on the bottom shows the p-values. Coefficients with * show statistical significance at 95%. 31

33 Table 8: Determinants of Income Inequality Gini Coeff. Gini Coeff. Theil Index Theil Index Atkinson Atkinson Median Income *** *** *** *** *** *** 2.38e e e e e e-06 Median Income^2 1.90e-10*** 2.02e-10*** 4.16e-10*** 4.56e-10*** 2.08e-10*** 2.08e-10*** 3.79e e e e e e-11 Education *** *** *** *** *** *** % Females *** *** *** *** *** *** # of Years Spent *** ** *** * *** Language Unemployment % Naturalized OECD *** ** *** Political Stability Latin America Asia * Europe Africa North America Dropped Dropped Dropped Dropped Dropped Dropped Oceania Constant *** *** *** *** *** *** Observations R-squared Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 32

34 Table 9: Determinants of Education Inequality Gini Coeff. Gini Coeff. Theil Index Theil Index Atkinson Atkinson Median Income -2.04e e e e e e e e e e e e-06 Median Income^2 2.51e e e-10* 8.56e e-11** 7.03e e e e e e e-11 Education *** *** *** *** *** *** % Females *** * * # of Years Spent Language *** *** *** *** *** *** Unemployment * * ** * % Naturalized ** ** ** OECD *** ** ** Political Stability * Latin America Asia * * ** ** ** Europe Africa North America Oceania *** *** *** *** *** *** Constant *** *** *** *** *** *** / Observations R-squared Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1 33

35 FIGURES: Figure 1: Right-skewed Income Distribution of Natives and Immigrants in United States Figure 2: Gini Coefficient for Income in Descending Order, By Country of Origin 34

36 Figure 3: Scatter Diagram of Gini Coefficient of Immigrant Groups, Median Income 35