The impact of foreign direct investment on racial income inequality: evidence from a panel of US states

(1)

The impact of foreign direct investment on racial income inequality:

evidence from a panel of US states

University of Groningen Faculty of Economics and Business MSc Economic Development and Globalization

Name: Aniek Diphoorn Student Number: 2203898 E-mail: a.r.diphoorn@student.rug.nl

Date paper: 05-01-2021

(2)

Abstract

This thesis analyzes the impact of foreign direct investment (FDI) on racial income inequality, and if this impact is dependent on higher black employment. It covers an interesting gap in research, since almost no attention is paid to the impact of FDI on racial income inequality. A fixed effects panel regression of a balanced dataset of 46 US states with the years ranging from 2010 to 2017 show that inward FDI has a decreasing effect on racial income inequality. When introducing control variables this impact remains significant and robust. Furthermore, separate regressions check if there are differences between states and/or regions and this turns out to be the case. The results provide evidence for the ‘discrimination theory’ and ‘outsiders network advantage’, suggesting that inward FDI reduces discrimination against black employees and creates opportunities for the previously excluded group (i.e. black employees).

(3)

1 Introduction

In the last few years ‘Black Lives Matter’ became a very actual topic again. “Like the later stages of the Civil Rights movement in the 1960s, Black Lives Matter is interested in general class inequalities as well as specifically racial inequalities” (Chernega, 2016 p.240). Title VII of the Civil Rights Act tried to eliminate racial income inequality by prohibiting wage and employment discrimination based on race (Critzer, 1998). Unfortunately, according to Gordils, et al. (2020), the racial income gap in the United States as of today still exists. The difference is larger if we look at income by race and ethnicity than if we look at the differences in income between men and women (Fairlie, 2018; McCall, 2001). Black Americans earn less than White Americans, with a median income of 45.438 US dollars compared to a median income of 72.204 US dollars in 2019, a difference of 26.766 US dollars which is more than half of what Black Americans earn in the first place (United States Census Bureau, 2019). The following two theories try to explain racial income inequality. First, there are differences in human capital, like educational differences, between black and white that explain a large part of the racial income gap (Bjerk, 2007). Second, blacks are more often than whites in the lower end of the labor market since employers have a tendency to employ blacks for these ‘blue-collar’ jobs (Semyonov & Lewin-Epstein, 2009).

(5)

In the last couple of years, the United States is the largest recipient of foreign direct investment (FDI) inflows in the world with 261 billion US dollars in 2019, followed by China with 156 billion US dollars (OECD, 2020). It is not only a leader in attracting FDI, the United States is also one of the countries with the largest amount of foreign-based multinational enterprises (MNEs) (Chintrakarn, et al. 2012). The United States is an attractive economy for foreign investment since it is considered a low-risk economy. FDI inflows come with major advantages like higher productivity growth, more exports, and the creation of jobs for the American worker (Kornecki & Ekanayake, 2012). The United States provides an interesting empirical context for testing the impact of FDI on racial income inequality for 3 reasons. First, the United States, as discussed, is one of the largest recipients of FDI. Second, since race is a prominent category in the US and the differences in income between Blacks and Whites are persistent, this could indicate that due to this ‘racial’ income inequality the overall income inequality in the US is higher than in other developed countries (Gordils, et al. 2020; Nielsen & Alderson, 1997). Third, the United States is divided into 50 states which gives the opportunity to also control for heterogeneity and see if there are other factors influencing this relationship. This is supported by the paper of Chintrakarn et al. (2012) where large differences in the results of the impact of FDI on income inequality in the different states of the US were found.

(6)

since black workers get paid less than white workers (Becker, 1971; Huffman & Cohen, 2004; Kerr & Walsh, 2014).

Therefore, the main question in this paper is; “What is the impact of FDI on racial income

inequality in US states, and is this effect influenced by higher black employment?”

The data for racial income inequality is constructed using the ratio of median incomes of black and white workers. The results indicate that an increase in FDI leads to lower racial income inequality in the United States. This relationship is both statistically significant and robust across several specifications. If the variable FDI increases with 1 percentage point, racial income inequality will decrease with 2.45 percentage points. Just as in the research of Chintrakarn et al. (2012) the results differ for different states. When splitting the States into Southern states and all other states, a significant effect is found for the Southern states where an increase in FDI leads to lower racial income inequality.

The outline of the rest of this research is as follows. Section 2 discusses the theoretical explanations for the impact of FDI on racial income inequality. Section 3 discusses and describes the data and the empirical model. Section 4 reports the results as well as introducing a robustness test. The last section 5 provides a discussion about the results. It also discusses possible limitations and suggestions for further research in this underexplored field, after that the paper will be concluded.

2 Literature review

Due to the scarce literature on the specific relationship between FDI and racial income inequality this paper first discusses the overall relationship between FDI and income inequality. When income inequality is discussed the focus is primarily on the personal income distribution and more specifically salaries and wages.

2.1 FDI and income inequality

(7)

relationship is relatively scarce. Sylwester (2005) studies a group of less developed countries during 1970-1989 and finds no evidence that FDI leads to an increase in income inequality in these countries. Blonigen & Slaughter (2001) looked at the effect of inward FDI on wage inequality between skilled and unskilled workers in the United States during 1977-1994, but also found no evidence that there is a significant relationship between FDI and wage inequality. The studies that find a negative relationship between FDI and income inequality are not abundantly available either. A study on Mexico’s 32 states between 1990 and 2000 found that for all states, an increase of FDI inflows led to a decrease in income inequality (Jensen & Rosas, 2007). When we look at mixed results, Figini & Görg (2011) found by studying more than 100 developing and developed countries, that in developing countries inward FDI stock led to an increase in wage inequality. Basu & Guariglia (2007) find the same result for developing countries, which is mainly due to low human capital. In developed countries, however, inward FDI stock led to a decrease in wage inequality (Figini & Görg, 2011). This difference between developed and developing countries is also found by Gopinath & Chen (2003). According to them, the increase in income inequality in developing countries is due to the different effect on skilled and unskilled workers. This finding is confirmed by multiple within-country studies that found that an increase in FDI increases the income gap between skilled and unskilled workers, which increases the overall income inequality (Aitken & Harrison, 1999; Feenstra & Hanson, 1997; Mah, 2002; Chen, et al., 2011). Multiple cross-national studies also found that an increase in FDI increases income inequality, but made no distinction between skilled and unskilled workers (Alderson & Nielsen, 1999; Reuveny & Li, 2003; Choi, 2006; Herzer, et al., 2014)

There is a scarcity of research on the impact of foreign direct investment on income inequality in Europe. Herzer & Nunnenkamp (2013) did research in 8 European countries and found a positive relationship between inward FDI and income inequality in the short-run, but a negative relationship between the two in the long-run. Another study on 10 countries, from Central and Eastern Europe, found that FDI increases income inequality in the short-run, but this relationship diminishes with the increase of GDP per capita and education levels (Mihaylova, 2015).

(8)

the manufacturing and the primary sector. A reason for these various effects of FDI on income inequality in different sectors can be that workers, who work in sectors that receive foreign direct investment, get higher average earnings (Mahler, et al., 1999).

Another reason for the effect of FDI on income inequality is that multiple research found that multinationals offer higher wages than domestic firms and foreign ownership can increase wages with 20-40% on average (Chen et al, 2011; Velde & Morrissey, 2003; Lipsey & Sjoholm, 2004; Jensen & Rosas, 2007). According to Lipsey & Sjoholm (2004) this could be due to the fact that these foreign-owned firms have a preference for more educated workers. The downside is that due to the presence of multinationals the wage level in domestic enterprises goes down and it discourages wages in these domestic enterprises to increase, which will increase income inequality even more (Chen et al. 2011). Still governments see it as a positive way to increase the wages for their employees, for example Setzler & Tintelnot (2019) find that the wages of a worker in the U.S. increases with 7 percent due to a foreign multinational firm.

Chintrakarn et al. (2012) found that the influence of FDI on income inequality in the US can differ by state. In 27 states FDI led to a decrease in income inequality, in 21 states FDI led to an increase in income inequality.1_{When they took the US as a whole, they find a significant}

relationship that FDI led to a decrease in income inequality in the long run. A great example that the big picture can mask the underlying differences. Coughlin et al. (1991) find that there also is a difference in the amount of foreign direct investment in different states; the bigger the state the more foreign direct investment transactions. This in turn leads to differences in per capita income across states. There is also an opposite effect, if a state has higher wages to begin with there is less amount of FDI in this state (Coughlin et al., 1991). Not only does FDI have an influence on income, inward FDI can also lead to higher US employment (Lucyna & Ekanayake, 2016).

2.1 High- and low-skilled

There is an extensive amount of research about the effects of FDI on income equality, either on a multiple country base or by looking at one country. Many of these studies focus on the earlier work of the international trade theory, the Stolper-Samuelson theorem. (Feenstra & Hanson, 1997; Figini & Görg, 2011; Jensen & Rosas, 2007; Mah, 2002; Mahler, et al., 1999; Mihaylova, 2015; Reuveny & Li, 2003) It predicts that in, usually developing, countries with large

(9)

endowments of -low-skilled- labor (wages) relative to capital (profits), FDI and free trade will increase the demand for labor and therefore it will increase wages. This in turn leads to a decrease in income inequality due to the fact that the returns to labor increased and the returns to capital decreased (Stolper & Samuelson, 1941). FDI often increases the use of more technology, which asks for more high-skilled labor. Looking at developed economies, like the United States, this theory predicts that FDI will lead to a higher demand of high-skilled labor. This higher demand leads to even higher wages for high-skilled labor and the difference in income with low-skilled labor will increase, leading to higher income inequality (Mihaylova, 2015). These different perceptions are not only there between countries but can also apply to different regions within a country (Feenstra & Hanson, 1997).

We can conclude that inward FDI increases the demand for high skilled labor, since inward FDI in developed countries is more often in industries with a larger amount of high skilled labor. One thing that is not clear from this line of reasoning is if blacks benefit as much from this increase in high skilled labor as whites. The educational differences between blacks and whites are a well-known source of racial income inequality as of today (Bjerk, 2007), even though Semyonov & Lewin-Epstein (2009) found that the large number of black workers in low-status and low-paying jobs has declined and their educational levels have increased. In 2001, The No Child Left Behind (NCLB) act tried to narrow the gap of academic achievement between black and white, but apparently not enough. “The achievement gap is frequently described as the often-observed academic differences between successful white American students and racial minority students” (Smith et al., 2020, p.630). Black and white usually live in different school districts and a lot of these districts, are racially segregated. The school districts for blacks (often) have lower educational resources, less qualified teachers, higher crime rates, and less job opportunities. This places black students at an academic disadvantage to succeed and graduate (Smith et al., 2020). Hoover & Yaya (2010, p.81) state this as: “Whites and Blacks systematically find different employment opportunities and sources of income that leads to dual economies and persistent income inequality”. Nielsen & Alderson (1997) named these dual economies ‘racial dualism’, it compares the incomes of black and white on average resulting in the amount of inequality, this in turn has a significant effect on racial income inequality.

(10)

skills2_{, racial income inequality in the whole labor market is more present in ‘blue-collar’ jobs.}

A reason is that academic skills in ‘white-collar’ jobs are considered more important and therefore there is less reason to discriminate in this sector. There are reasons why discrimination does occur in the ‘blue-collar’ sector. First, it is less costly in this sector to hire a white worker over a more skilled black worker. Meaning that the loss of productivity of hiring a less skilled worker is lower in the ‘blue-collar’ sector than in the ‘white-collar’ sector. Second, when hiring someone for a job in the ‘blue-collar’ sector it is less likely that a firm will spend a sufficient amount of money to obtain specific information about the academic skills of a potential employee. In this sector the bias that comes with race about the academic skill of the employee can give the employer enough ‘free’ information. Since it is important to invest in the assessment of academic skills in the ‘white-collar’ sector this statistical discrimination only/more often occurs in the ‘blue-collar’ sector (Bjerk, 2007).

The research by Gelan et al. (2007) also finds a difference in the employment opportunities between high and low skilled employees, but this impact is amplified with an increase of inward FDI. The employment opportunities for black workers go up with an increase in FDI, especially in high paying jobs. When industries only have domestic owners, it results in the opposite, the employment opportunities go down in high paying jobs for blacks (Gelan et al., 2007). We can conclude that having foreign owners provides benefits for the previously excluded group (e.g. blacks). In a study by Siegel et al. (2019) this effect is explained by means of what is called the ‘outsiders network advantage’. Multinationals hire the well-qualified excluded group, which in turn increases their performance. The multinational does not have the same social bias that the domestic owned firm (unknowingly/knowingly) has and can therefore hire the talented workers out of the excluded group. In this way the multinational can outcompete the domestic firm by operating outside the domestic social network. This not only creates an advantage for the multinational but also new possibilities for the multinational itself and for the excluded group. Another finding of this study is that multinationals use the opportunity to hire the excluded group more often than firms that are domestically owned. The competitive advantage that this creates for the multinational indicates that the domestic firm has to change something about its bias in order to keep up with the multinational. Slowly the market will participate in less discrimination practices, simply because it is not profitable (Siegel et al., 2019). These discrimination practices will be discussed next.

2_{Note: In the paper by Bjerk (2007) academic skills are not measured as academic achievement but by means of}

(11)

2.3 Discrimination

Just as with the ‘outsiders network advantage’ where a previously excluded group is hired to gain profit opportunities for a firm that hires this group, the discrimination theory by Becker (1971) in general explains the same. This theory entails that firms that discriminate, even though they have to pay more for their hiring preferences, only hire people that are from the favored demographic group. This indicates that firms that do not discriminate and hire the previously excluded group can gain higher profits in comparison with the firm that discriminates (Siegel et al., 2019). In a world with perfect competition this would lead to the disappearance of firms that discriminate, since capital would flow to the more profitable nondiscriminatory firms (Fan et al., 2017). When a firm however has some kind of market power, like rules enforced by the government about foreign entry or a monopoly position, the competition from other firms does not matter that much and these firms are more likely to discriminate (Siegel et al., 2019). In oligopolistic labor markets, employees that are looking for a job have far less options, and this gives the employer the opportunity to engage in discrimination. When a country is able to increase its inward FDI the position of the firms in an oligopolistic labor market are weakened since there are more possible employers that like to hire the well-qualified employees. These findings suggest that more competition, especially if these competing firms come from outside the US, can create better job and income opportunities for the previously excluded group (Gelan et al, 2007). Again, the discrimination theory by Becker (1971) can explain this mechanism when there is an increase in FDI or international trade. More competition through FDI ensures that discrimination will decrease, because it is getting more difficult to prosecute discrimination practices. More investment from abroad leads to a changing business environment where it is no longer possible for firms to pay minorities not according to their marginal productivity. Through foreign investment there will be shifts in market power and increased competition and the wage levels of minorities will increase. With FDI the amount of racial income inequality can therefore decrease (Becker, 1971; Johnson-Lans, & Jones, 2017).

Hypothesis 1: “Higher levels of FDI in a US state has a negative effect on racial income

inequality”

(12)

this is that black employees looking for a job, first try to get one at a firm that is less discriminating and only thereafter go to the relatively more discriminating firms. When there is a higher number of black workers, there is a higher chance that these black workers have to work for a discriminating firm, leading to higher income differences between black and white (Kerr & Walsh, 2014). There is a higher concentration of black people in the South of the United States than in other regions. Even though employment inequality is lower in the South, income inequality is higher in the South (Critzer, 1998; Hoover & Yaya, 2010). There are 2 studies that studied the effect of the amount of black people or black workers in different metropolitan statistical areas (MSA) on income inequality. One study takes the whole black population and the other just takes the black full-time workers. Both, find that a high amount of black people/workers income inequality increases (Huffman & Cohen, 2004; Kerr & Walsh, 2014). One earlier study finds a contradicting relationship where income inequality was decreasing in the South and increasing in other regions (Bishop, et al., 1994). Huffman & Cohen (2004) explain this by means of two mechanisms. First; with a higher black population in a particular area it may lead to more black workers in jobs that are mostly done by black workers. Second; black workers get paid less than white workers, a higher number of black workers can therefore increase income inequality. They found that these mechanisms are more present in a higher black population. Moreover, black workers are more often excluded from better jobs and the inequality in wages between black and white is higher when the black population is higher (Huffman & Cohen, 2004). These mechanisms are reinforced by the average income of the white population which is higher when the black population is higher. This results in even higher income inequality in regions with a higher black population (Deaton & Lubotsky, 2003).

Hypothesis 2: “Higher black employment (due to the presence of a higher black labor force)

in a US state has a positive effect on racial income inequality”

(13)

profitable for the multinationals to act against the existing social bias of the domestic firms and with the presence of a higher black labor force, the multinational has more well-qualified employees to choose from that are overlooked by the domestic firm (Hoover & Yaya, 2010; Siegel et al., 2019).

Hypothesis 3: “Higher levels of FDI in interaction with higher black employment in a US state

have a negative effect on racial income inequality”

3 Methodology and Data

The main analysis in this research is a fixed effect regression analysis of foreign direct investment and black employment on racial income inequality. Thereafter, an interaction term is created with the variable’s foreign direct investment and black employment to see whether they have a combined effect on racial income inequality. The regression analysis is carried out in a sample of 46 states over 7 years. This research only covers 7 years since the data for the two main variables, racial income inequality and foreign direct investment, is not available for any further years. Below, the control variables and descriptive statistics of the variables will be discussed. In the next sections the empirical model will be explained and the analysis is conducted.

3.1 Data

Dependent & Explanatory Variables

(14)

Following Ford, et al. (2008), we use data on employment in foreign-owned affiliates to construct an inward stock measure for FDI by state. We use stocks instead of flows because stocks are more effective to capture the long-run effect of FDI and foreign control of domestic production (Figini & Görg, 2006). This data is obtained from; Foreign Direct Investment in the

United States, Majority-Owned Affiliates, Employment and Compensation of Employees, Employment of Affiliates, State by Industry of Affiliate, published by the Bureau of Economic

Analysis. The variable FDI is constructed as the ratio of employment in foreign-owned affiliates to total employment in that state. The data for total employment is obtained from the same database as black employment and will be explained next.

Another explanatory variable is the number of black people that are employed by state as a percentage of total employment. This data is obtained from the US bureau of labor statistics from the employment status of the civilian noninstitutional population in states by sex, race,

Hispanic or Latino ethnicity, marital status, and detailed age. The data used is under the

heading ‘black or African American’ and is available for the years 2003-2019. In this database there are a few missing observations for ‘black or African American’, some of these missing observations could be obtained from the database of the United States Census Bureau.

Control Variables

To see if there are variables, besides FDI and black employment, that have an influence on racial income inequality, the following control variables are included. These variables are also included to try and overcome the problem of omitted variable bias as much as possible. The belief is that when some of these relevant control variables are omitted this will lead to biased results. In research it is impossible to know all the relevant control variables, but when the most obvious and sufficient relevant variables are included it reduces the threat from omitted variable bias (Clarke, 2005).

(15)

controlling for GSP can lead to a biased estimate, since Setzler & Tintelnot (2019) found that there are larger wage premiums paid by foreign multinationals if they have higher GDP per capita.

- Unemployment. It is expected that the unemployment rate has a positive relationship with racial income inequality. Blacks are more often unemployed compared to Whites (Gordils, et al., 2020). The average unemployment rate for whites was 3.5%, which was close to the national average of 3.9%, and the average unemployment rate for blacks was 6.5% (US Bureau of Labor Statistics, 2018). If employers prefer to hire whites for certain jobs rather than blacks (Jaret, et al. 2003), this can lead to higher racial income inequality between blacks and whites. The unemployment rate is measured as the amount of people that is unemployed by state as a percentage of the total labor force in that state. It is obtained from the US Bureau of Labor Statistics and is available for the years 2003-2019. Not controlling for unemployment can lead to a biased estimate, since higher unemployment rates attract relatively more FDI (Kornecki & Ekanayake, 2012)

- Education. Blacks have in general lower formal education, in 2019 22% of Whites had a Bachelor degree and 9% a Master degree, with Blacks only 15% and 7% respectively. The levels for secondary achievement state that 28% of Whites have a secondary degree with 33% of Blacks (United States Census Bureau, 2019). Income is likely to rise with education, but in research by Semyonov & Lewin-Epstein (2009) was found that Blacks receive even higher income returns on their educational achievement than Whites. It is therefore expected that education has a negative relationship with racial income inequality. There is no data on educational achievement on state level by race, so education is measured as the total number of people with a secondary degree of working age population by state as a percentage of the total population in that state. Education data is obtained from the Kids Count Data Center (Education attainment of working age population 25 to 64) and is available for the years 2000-2019. Education is included since with an increase in FDI there is higher demand for skilled workers, and the wage premium for skilled workers tends to increase (Gelan et al., 2007; Johnson-Lans & Jones, 2017).

(16)

income inequality (Johnson-Lans & Jones, 2017). On the other hand, openness to trade can cause more competitiveness, which can discourage discrimination practices and therefore lead to a decrease in racial income inequality (Johnson-Lans, & Jones, 2017). At this point it is therefore not certain what the expected sign of the relationship with trade openness to racial income inequality should be. Trade openness is measured as the total value of imports and exports by state to GSP (Figini & Görg, 2011). The data is obtained from USA Trade Online by the US Census Bureau and is available for 2002-2019. Openness to trade can counteract the effect of FDI increasing or decreasing income inequality since this is also a main contribution to this effect (Figini & Görg, 2011).

- State Tech and Science Index (STSI). This index measures R&D inputs, entrepreneurship & risk capital, human capital investment, scientific & technological workforce, and technology concentration & dynamism on state level. In previous research a positive relationship with STSI on wages and income is found (Klowden et al., 2018). It is expected that this index has a positive relationship with racial income inequality due to the fact that there is growth of high technology employment, that leads to bigger divergence of high- and low-income industrial sectors and regions in the US (Leicht & Jenkins, 2017). This index is obtained from Milken Institute and is available for the years 2002-2018 with a gap every other year. To get an index for every year the average of the years 2010 & 2012 is calculated for the year 2011 and so on.

3.2 Descriptive Statistics

(17)

that these variables are not normally distributed.3_{Interesting to see is that in this dataset, on}

average, only 4.3% of total employment is in foreign-owned affiliates and that on average only 10.2% of people being employed is black.

Table 1: Descriptive statistics

Variable Description Observations Mean Standard deviation

Inc Racial income inequality 368 0.6321655 0.0978801

FDI Employment in foreign-owned

affiliates to total employment 368 4.297719 1.180659

Black Black employment to total

employment 368 10.21539 8.573041

Educ Ratio of people with a secondary

degree to total population

368 52.82256 7.097055

GSP GDP per capita in current US$ by

State 368 52200.22 10472.73

STSI State Technology and Science

Index 368 53.08444 14.63783

Trade Total import and export by state to

GSP 368 18.94661 9.62472

Unemp Unemployment to total

employment 368 6.455082 2.190948

3.3 Empirical model

Panel analysis is used to analyze the effects of FDI and black employment on racial income inequality over different US states and time. The sample consists of 46 states over 7 years between 2010-2017, without gaps. Since we are interested in effects over time states with only one or none observations were dropped from the dataset. Following Critzer (1998) the States with a ratio for racial income inequality above 1 were also dropped, since these are considered as outliers. The States Idaho, Montana, Vermont & Wyoming are therefore dropped from the analysis.4_{The District of Columbia and Puerto Rico were never included, since there is hardly}

any data and these are officially not considered as states. This leads to a balanced panel of 368 observations.

As discussed in the descriptive statistics the histograms of the variables black, trade and GSP show that they are not normally distributed. To solve this normality issue, these variables are transformed in natural logarithms (Herzet et al, 2014). When using non-normal variables, it increases the chance of producing negatively skewed errors. By using the logarithms of these variables, the overall fit of the model will be improved. No observations were dropped, when taking logarithms of the variables black, trade and GSP.

3_{The histograms for the variables black, trade and GSP can be found in Appendix A2}

(18)

The variable FDI is introduced with a lag of one year, to solve a potential endogeneity problem. In previous research lagged variables of FDI are also used to decrease potential endogeneity and causality issues (Figini & Görg, 2011; Herzer et al.; Herzer & Nunnenkamp, 2013; Tsai, 1995). 46 observations are dropped from the analysis due to creating the lag of one year for FDI. To find out whether to use a fixed or random effects panel model, the Hausman test is carried out (Appendix A3). This test gives a significant result indicating that the error terms are not correlated, and confirming that a fixed effects model is preferred over a random effects model. The fixed effects panel model with robust standard errors allows controlling for time-invariant variables, unobserved heterogeneity across states and uses year fixed effects (Mihaylova, 2015). The characteristics of an entity may impact or bias the predictor or outcome variables. By controlling for these time-invariant characteristics with a fixed effect model the impact or bias will be removed. There are also always variables you cannot observe or measure, and with a fixed effects panel model we can control for these unobserved heterogeneity variables that do not change over time.

The model to determine the effect of FDI and black employment on racial income inequality is depicted below:

INCit = ß0 + ß1FDIit-1 + ß2lnBLACKit + ß3lnGSPit + ß4UNEMPit + ß5EDUCit + ß6lnTRADEit +

ß7STSIit+δi + δt + uit

For our third hypothesis the combined effect of FDI and black employment is introduced into the model:

INCit = ß0 + ß1FDIit-1 + ß2lnBLACKit + ß3 (FDIit-1 x lnBLACKit)+ ß4lnGSPit + ß5UNEMPit +

ß6EDUCit + ß7lnTRADEit + ß8STSIit+δi +δt + uit

In both models all the variables are measured per state i in time t, except for the variable FDI

which is measured with a lag of one-year it-1. INC stands for racial income inequality, FDI for

(19)

where highly correlated variables could potentially lead to issues of multicollinearity. Some of the pair-wise correlations show high multicollinearity, such as GSP and education, STSI and education and GSP and STSI. To confirm that there are no issues concerning these highly correlated variables a VIF test is performed, the results are shown in Appendix A5. Using a threshold of 4, it seems that no variables have issues with multicollinearity, so therefore no variables are excluded from the model.

Next the Breusch-Pagan (Appendix A6) and the Wald test (Appendix A7) are used to test for heteroscedasticity. The p-value that corresponds to the Chi-square is 0.0004 in the Breusch-Pagan test. Since this value is less than 0.05, we can reject the null hypothesis and conclude that heteroscedasticity is present in the data. The Wald test can be used in order to test for heteroscedasticity in a fixed effect model and again it is confirmed that heteroscedasticity is present. The problem of heteroskedasticity is solved by using a fixed effect estimator with robust standard errors. Serial correlation can be another problem in panel data with long time series, where the present value effects the future value of a variable. Since this research only covers 7 years no serial correlation problems are expected. After carrying out the Woolridge test (Appendix A8), we fail to reject the null hypothesis (H0: no serial correlation) and can conclude that this data indeed does not have first-order autocorrelation.

4 Results

In the following sections the analysis and results will be discussed, after that a robustness test is carried out.

4.1 Analysis

(20)

Johnson-Lans & Jones (2017) where foreign direct investment also decreases racial income inequality. Apparently, the increase of FDI is racial income inequality decreasing in both a developing and developed country. It seems that it also does not matter if the difference in the number of black residents between the two studied countries shows large disparities. On average, 74.6% of the people in South Africa are Black and in the United States this percentage lays only at 13%, on average. Of next main interest is the independent variable for black employment, which is added from model 2 onwards. In contrary with our expectations this variable shows a positive sign, what gives an indication that higher black employment is racial income inequality decreasing. However, this effect is not statistically significant so no main conclusions can be drawn. Each model adds a control variable to show the strength of the influence of this particular variable. Both coefficients remain with the same sign and significance throughout the models, making this finding very robust to controls. Only when the control variable unemployment is added in model 7 the main variable of interest FDI decreases slightly in significance. This result can be explained by the fact that FDI is measured as employment in foreign-owned affiliates which can be related to unemployment overall.

When looking at the control variables, the only one which is (highly) significant is GSP per capita. This variable has a negative sign throughout all models, indicating that wealthier states have higher levels of racial income inequality. Setzler & Tintelnot (2019) found that there are higher wage premiums paid by foreign multinationals if they have a higher GDP, our findings suggest that these wage premiums go to white employees. All other control variables have no significant correlation with racial income inequality. Surprisingly the variable education has a negative sign, indicating that more education is racial income inequality increasing, which goes against our expectations. The variable STSI is also not significant but has a positive sign which is in line with our expectations.

(21)

Table 2 Fixed effect model

Significant at *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in brackets

R-squared is the within R-squared

In Table 3 our third hypothesis is tested following Gelan et al (2007) to see if there is a combined effect of higher black employment and higher levels of foreign direct investment. None of the main variables of interest show any sign of significance. However, interesting to mention is that the sign for black employment, when introducing the interaction variable goes from a positive sign (in Table 2) to a negative sign (in Table 3). The interaction term measuring the combined effect of black employment and foreign direct investment has a positive sign, indicating that this combined effect is racial income inequality decreasing. However, since this variable is not significant there is no sufficient evidence that there is a relationship with the combined effect of foreign direct investment and black employment on racial income inequality. Therefore, there is no evidence to support our third hypothesis. The sign for our main variable foreign direct investment has not changed but became a lot smaller, but again there is no significant support. The sign for trade openness changed from positive (in Table 2) to negative, indicating in this model that states more open to trade were racial income inequality increasing.

Dependent variable: (Model 1) (Model 2) (Model 3) (Model 4) (Model 5) (Model 6) (Model 7)

Racial income inequality FDI 0.0230*** 0.0240*** 0.0277*** 0.0275*** 0.0260*** 0.0252*** 0.0245*** (0.00824) (0.00870) (0.00859) (0.00859) (0.00827) (0.00846) (0.00852) Black employment 0.0175 0.0155 0.0155 0.0110 0.00747 0.0163 (0.0413) (0.0399) (0.0401) (0.0397) (0.0400) (0.0450) GSP -0.212*** -0.215*** -0.211*** -0.223*** -0.291*** (0.0698) (0.0709) (0.0708) (0.0694) (0.0997) Education -0.000619 -0.000813 -0.000883 -0.00148 (0.00302) (0.00301) (0.00296) (0.00324) STSI 0.0174 0.0164 0.00973 (0.0371) (0.0362) (0.0355) Trade openness 0.00170 0.00152 (0.00130) (0.00139) Unemployment -0.00613 (0.00711) Constant 0.532*** 0.496*** 2.768*** 2.839*** 2.768*** 2.825*** 3.659*** (0.0316) (0.0932) (0.737) (0.793) (0.825) (0.808) (1.217) Observations 322 322 322 322 322 322 322 R-squared 0.035 0.036 0.052 0.052 0.053 0.057 0.062 States 46 46 46 46 46 46 46 Year FE

(22)

Table 3 Fixed effect model with interaction variable

Dependent variable: (Model 1) (Model 2)

Racial income inequality

FDI 0.00644 0.00964

(0.0186) (0.0199)

Black employment -0.0126 -0.01000

(0.0413) (0.0506)

FDI * Black employment 0.00854 0.00740

(0.00677) (0.00781) GSP -0.284*** (0.10030) Education -0.00191 (0.00349) STSI 0.00167 (0.00143) Trade openness 0.00470 (0.0364) Unemployment -0.00512 (0.00684) Constant 0.555*** 3.648*** (0.0996) (1.2342) Observations 322 322 R-squared 0.040 0.064 States

Year FE YES 46 YES 46

State FE YES YES

R-squared is the within R-squared

To give some more explanation about the interaction term in Table 3, figures 1 and 2 are included.

Figure 1 Adjusted Predictions

.58 .6 .62 .64 .66 .68 Pre d ict e d ra ci a l in co me i n e q u a lit y -.2 .2 .6 1 1.4 1.8 2.2 2.6 3 3.4

black employment (in logs) FDI = 1% FDI = 4% FDI = 8%

(23)

Figure 1 shows the effect of black employment on racial income inequality when FDI is at a certain percentage. We can conclude that with a higher amount of FDI the same negative effect is found of black employment on racial income inequality.

Figure 2 Predictive Margins

In figure 2 the predictive margins are shown for the model in Table 3, it shows the linear prediction of the racial income inequality ratios for values of FDI for each level of black employment.5

Following Chintrakarn et al. (2012) it is important to note that the above models measure the effect of foreign direct investment on racial income inequality as an average of the individual state effects. In the research by Chintrakarn et al. (2012) the effect of FDI on income inequality is studied in the United States as a whole but also allowing for possible heterogeneity effects across states. Important differences across States, such as the black/white population ratios, are not taking into account in the above models. In order to look at these heterogeneity effects, the coefficients of FDI on racial income inequality for each state are presented in Table 4. These coefficients are estimated by performing a separate regression for each US state without fixed effects since this effect disappears when regressing each state separately. Table 4 indeed shows that there is a lot of heterogeneity across states, with the coefficients range from -0.41 in New

(24)

Hampshire to +0.24 in West Virginia. If we split the States in a positive and negative effect of FDI on racial income inequality, we see that 16 states show an increase in racial income inequality with an increase in FDI, with 3 coefficients being statistically significant. 30 states show a decrease in racial income inequality with an increase in FDI, with 6 coefficients being statistically significant. Overall, we can conclude that the racial income inequality decreasing effect is the more dominant effect. These results are in line with the results in Table 2, where on average there is a decreasing effect on racial income inequality for the US states.

Table 4 State Estimates, the impact of FDI on Racial income inequality

State FDI t-Stat State FDI t-Stat

New Hampshire

-0.408516 -4.02*** South Carolina 0.007441 0.33

Utah -0.112493 -0.51 Indiana 0.0086406 1.5

Massachusetts -0.08309 -2.78** Wisconsin 0.0089719 0.34

New Mexico -0.068697 -0.36 Mississippi 0.0096905 0.57

Nevada -0.049016 -0.174 Michigan 0.0152837 2.79*

Oregon -0.045035 -0.71 North Carolina 0.0172806 1.54

California -0.036705 -2.11* Ohio 0.0198573 0.99

Rhode Island -0.032001 -0.71 Pennsylvania 0.0246432 0.61

Iowa -0.028022 -1.14 Kentucky 0.0308283 1.98

Kansas -0.024526 -0.9 Arkansas 0.0320045 1.2

South Dakota -0.016354 -0.12 Washington 0.0322272 1.54

Arizona -0.014903 -0.57 Georgia 0.0346795 2.41*

Illinois -0.012902 -1.55 Nebraska 0.0372015 1.67

New York -0.007376 0.45 Florida 0.0378112 6.37***

New Jersey -0.006897 -0.84 Hawaii 0.0418872 0.43

Louisiana -0.004565 -0.31 Texas 0.0472839 4.43***

Oklahoma 0.0018086 0.04 Delaware 0.0492309 0.88

North Dakota 0.0020204 0.02 Minnesota 0.0557239 1.29

Tennessee 0.0024151 3.44** Colorado 0.0621467 2.46*

Virginia 0.0024776 0.24 Maine 0.0629611 0.32

Maryland 0.0028533 0.09 Alaska 0.0676022 0.32

Alabama 0.0037055 0.26 Connecticut 0.1205886 0.99

Missouri 0.0057371 0.18 West Virginia 0.2355708 1.62

Significant at *** p<0.01, ** p<0.05, * p<0.1

(25)

explanation for the positive and negative effects could be that the States with a significant negative effect have, on average, lower black employment (1.24%, 6.94% & 5.57%) and the States with a significant positive effect have, on average, higher black employment (16.07%, 11.06%, 29.11%, 14.95%, 11.64%) with only Colorado (3.78%) being an exception. In Figure 3 the effect of FDI on racial income inequality is graphically shown, where the dark colors represent a racial income inequality increasing effect of FDI and the lighter colors represent a racial income inequality decreasing effect of FDI. We can conclude that the increasing and decreasing effects of FDI on racial income inequality are not centered around a particular area and that states with an increasing or decreasing effect even coexist.

Figure 3 State Estimates, the Effect of FDI on Racial income inequality

4.2 Robustness check

(26)

the South than in other regions. Following these findings and to test if our findings hold, the States are divided in two groups, the Southern States6_{and all other states.}

The bar charts in figure 4 & 5 show that the mean values of FDI and racial income inequality are quite similar in the Southern and other states. When we look at the minimum and maximum values of both variables, we see some differences between the Southern and other states. The maximum racial income inequality ratio in figure 5 is almost 1 in other states and around 0.85 in Southern states. The minimum ratio of employment in foreign-owned affiliates in other states is around 1.8% and in Southern states this is 2.2%.

Figure 4 FDI in Southern and other states Figure 5 Racial income inequality ratios

In Table 5, 4 models are presented, model 1 and 2 are for the Southern States, model 3 and 4 are for all other states. Model 1 and 3 represent the first estimation equation and in model 2 and 4 the interaction term of FDI and black employment is added. Interpretation of the models need to be made with caution since the observations in model 1 and 2 dropped to 112. An interesting observation is that the coefficient for FDI is significant in the Southern states and insignificant for all other states. An increase in FDI will decrease racial income inequality in Southern States. To test whether this coefficient in Southern states is really different from all other states, a dummy for the Southern states is generated and a t-test is conducted to test whether the coefficients of the two independent samples are statistically different. The results of this t-test are shown in Appendix A9. At the 5% significance level, there is sufficient evidence to support the claim that the coefficients of the two independent samples are statistically different.

6_{Southern States are Alabama, Arkansas, Delaware, Florida, Georgia, Kentucky, Louisiana, Maryland,}

Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia and West Virginia.

0 2 4 6 8 0 1 R a ti o o f e mp lo yme n t in f o re ig n -o w n e d a ffil ia te s / to ta l e mp lo yme n t 0 = other states 1 = southern states

min FDI max FDI

mean FDI 0 .2 .4 .6 .8 1 0 1 R a ci a l in co me i n e q u a lit y ra ti o s 0 = other states 1 = southern states

(27)

Compared to our first model in Table 2, the coefficient almost doubled in model 1 in Table 5. These findings are somewhat surprising since the state estimates did not show a positive significant relationship for all southern states. Louisiana even shows a negative relationship where FDI would lead to an increase in racial income inequality if this effect was significant. There are a few possible explanations for these higher and more significant results in the South. First, there is a larger amount of black people in the South which could implicate that chances that one of these people are hired is larger. Second, there could be less discrimination in the South due to an increase in FDI, decreasing racial income inequality. Third, the decreasing effect of FDI on racial income inequality can be attributed to the fact that FDI adds more jobs for people that are high-skilled and it can be that there are more high-skilled black workers in these regions. Last, another possibility is that there could be more MNEs hiring black workers in the South, confirming the theory of the ‘outsiders network advantage’. It can be that MNEs find more profitable business opportunities in the Southern states.

Table 5 The effect of FDI on racial income inequality in Southern States and all other states

Dependent variable: (Model 1) (Model 2) (Model 3) (Model 4)

Racial income inequality

FDI 0.0406*** 0.0983*** 0.0201 0.0214

(0.0134) (0.0323) (0.0157) (0.0265)

Black employment -0.143 -0.0430 0.0515 0.0544

(0.0826) (0.0776) (0.0513) (0.0645)

FDI * Black employment -0.0221 -0.000786

(0.0144) (0.0099) Education -0.00371 -0.00354 -0.00359 -0.00360 (0.00264) (0.00252) (0.00654) (0.00651) GSP -0.05339 -0.02680 -0.30152* -0.30219* (0.16727) (0.17498) (0.14021) (0.14048) STSI 0.00147 0.00185 0.00232 0.00231 (0.00128) (0.00137) (0.00198) (0.00198) Unemployment 0.00520 0.00527 -0.01045 -0.01056 (0.00875) (0.00885) (0.00985) (0.01013) Trade openness -0.0440 -0.0455 0.0234 0.0239 (0.0528) (0.0489) (0.0379) (0.0397) Constant 1.659 1.0903 3.810* 3.814* (1.899) (2.042) (1.809) (1.809) Observations 112 112 210 210 R-squared 0.289 0.305 0.078 0.078 States 16 16 30 30

Year FE YES YES YES YES

State FE YES YES YES YES

(28)

5 Conclusion

5.1 Discussion

Most importantly, the paper finds support for hypothesis 1. These results are very significant and robust to controls. In a robustness check is found that hypothesis 1 is particularly significant in the Southern US states. Therefore, the main findings are that, especially in Southern US states, an increase in FDI is related to a decrease in racial income inequality. The hypothesis is supported with a coefficient of 0.0245, and for Southern US states with a coefficient of 0.0406. The interpretation of this coefficient is that with a 1 percentage point increase in FDI racial income inequality will decrease with 2.45 percentage points, for Southern US states this means a decrease of 4.06 percentage points. Johnson-Lans & Jones (2017) found the same (although bigger) effect, where FDI led to a decrease in racial income inequality in South Africa. This paper is the first to explain a relationship between FDI and racial income inequality in a developed country and to explain the different effects of FDI on racial income inequality in different regions. The paper argues that this effect can be explained through the effects of the ‘outsiders network advantage’ and the discrimination theory by Becker (1971).

The relation between higher black employment and an increase in racial income inequality (hypothesis 2) was found in several studies (Becker, 1971; Deaton & Lubotsky, 2003; Huffman & Cohen, 2014; Kerr & Walsh, 2014). This study did not find such a relationship and if the results were significant, they would explain the opposite, higher black employment would lead to a decrease in racial income inequality. Since our dataset is rather small and uses very recent years, we suspect that the relationship found in previous research is due to a larger dataset or is something from before the year 2000. In previous research the 1980 and 1990 U.S. Census or different data sets between the years 1980 and 1990 were used (Deaton & Lubotsky, 2003; Huffman & Cohen, 2014). The most recent U.S. Census used are those from 1990 and 2000 in the research by Kerr & Walsh (2014), which is almost 20 years ago.

(29)

5.2 Limitations & Future Research

This paper has several limitations which will be discussed next. First and foremost is the lack of data on median income to define racial income inequality, which is the dependent variable in this research. Sufficient data on median income before the year 2010 was very hard to find. Since income does not change that much over the years, it is recommended to find data for more years (if possible) or try to define another measure for racial income inequality. Gordils, et al. (2020) created a gap score using the income difference between black and white in a particular area, where high scores indicate that whites earn more than blacks. Hoover & Yaya (2010) used inequality measures like Gini and Theil coefficients to define racial income inequality. These inequality measures were not available for the different US states. Another important variable in our research is the independent variable FDI, and there are several ways to define this variable. It was a challenge to find sufficient amount of data for this variable. Taking another measure for FDI could have had an influence on the findings in this paper. For example, Chintrakarn, et al. (2012) took the gross book value of property, plant, and equipment of affiliates in all industries. This could potentially lead to different results.

Second, a variable that could have had an influence but is not used in this paper is gender. Income differences between men and women are still present, taking the research by Critzer (1998) where income ratios show that black men, on average, earn more than black and white women. We suspect that separating the income difference between men and women would lead to different outcomes. Another variable that could have had an influence but is not included in this paper is age. Income is changing several times when one gets older, and in the research by Hoover & Yaya (2010) is found that average age had a negative effect on income inequality for the white population. We have not taken these two variables into account since there is no data available for median income separated by gender or age.

(30)

independent variable. Finding an IV that meets these conditions can be challenging and in research with an interaction term this term even needs its own IV (Ebbes, et al., 2017).

Fourth, after conducting a Pesaran’s test of cross-sectional independence, for which the results are shown in Appendix A10, the null hypothesis of no cross-sectional independence can be rejected. The average absolute correlation is 0.357, which is considered a very high value (de Hoyos & Sarifidis, 2006). The interactions between different states are not taken into account in the model in this research. Possible consequences are that the estimated standard errors are biased or the fixed effect coefficients are biased and inconsistent. For future research the CCE mean group estimator by Perasan (2006) can be used. “This estimator is consistent in the presence of cross-sectional dependencies that potentially arise from multiple unobserved common factors and involves augmenting the basic regression with cross-sectional averages of the dependent and explanatory variables as proxies for the unobserved common factors” (Chintrakarn et al., 2012, p.794). There are two main advantages when using this estimator, it allows for different responses per US state to the communal time effects as shown by the coefficients that are specific for each state on the cross-sectional averages of the variables. The second advantage is that it produces estimates that are consistent even when the variables are correlated with the communal factors (Chintrakarn et al., 2012).

5.3 Conclusion

This paper examines the impact of FDI on racial income inequality in US states. It also examines if the impact of FDI on racial income inequality differs in different states and/or regions in the US. Another examination is the impact of black employment on racial income inequality and the interaction effect of black employment and FDI on racial income inequality. We used a panel of 46 states for the period 2010 to 2017 to study this impact.

In conclusion, inward FDI leads to a decrease in racial income inequality taken the US as a whole. Different effects are found for separate regressions for different states where in some states FDI led to an increase in racial income inequality and in some states FDI led to a decrease in racial income inequality. Separating the States in Southern states and all other states, only a significant decreasing effect is found in the Southern states. No support is found regarding the effect of black employment or the interaction effect on racial income inequality.

(31)

(32)

References

Aitken, B. J., & Harrison, A. E. (1999). Do domestic firms benefit from direct foreign investment? Evidence from Venezuela. The American Economic Review, 89(3), 605-618

Alderson, A.S., & Nielsen, F. (1999). Income inequality, development, and dependence: a reconsideration. American Sociological Review, 64(4), 606-631

Basu, P., & Guariglia, A. (2007). Foreign direct investment, inequality, and growth. Journal

of Macroeconomics, 29(4), 824-839

Becker, G. (1971). The Economics of Discrimination. Chicago: University of Chicago Press. Bishop, J. A., Formby, J. P., & Thistle, P. D. (1994). Convergence and divergence of regional

income distributions and welfare. The Review of Economics and Statistics, 76(2), 228- 235

Bjerk, D. (2007). The differing nature of black-white wage in equality across occupational sectors. Journal of Human Resources, 42(2), 398-434

Blonigen, B. A., & Slaughter, M. J. (2001). Foreign-affiliate activity and U.S. skill upgrading.

The Review of Economics and Statistics, 83(2), 362-376

Bogliaccini, J. A., & Egan, P. J. W. (2017). Foreign direct investment and inequality in developing countries: does sector matter? Economics & Politics, 29(3), 209-236 Bureau of Economic Analysis (2017) Employment and Compensation of Employees.

Retrieved from https://www.bea.gov/international/di1fdiop

Chen, Z., Ge, Y., & Lai, H. (2011). Foreign direct investment and wage inequality: evidence from China, World Development, 39(8), 1322-1332

Chernega, J. (2016). Black lives matter: racialized policing in the United States. Comparative

American Studies, 14(3-4), 234-245

Chintrakarn, P., Herzer, D., & Nunnenkamp, P. (2012). FDI and income inequality from a panel of U.S. states. Economic Inquiry, 50(3), 788-801

Choi, C. (2006). Does foreign direct investment affect domestic income inequality? Applied

Economics Letters, 13(2), 811-814

Clarke, K. A. (2005). The Phantom Menace: Omitted variable bias in econometric research.

Conflict Management and Peace Science, 22(4), 341-352

Coughlin, C. C., Terza, J. V., & Arromdee, V. (1991). State characteristics and the location of foreign direct investment within the United States. The review of Economics and

Statistics, 73(4), 675-683

Critzer, J. W. (1998). Racial and gender income inequality in the American States. Race and

Society, 1(2), 159-176

Deaton, A., & Lubotsky, D. (2003). Mortality, inequality, and race in American cities and states. Social Science and Medicine, 56(6), 1139-1153

Ebbes, P., Papies, D., & van Heerde, H. J. (2017) Dealing with endogeneity: a nontechnical guide for marketing researchers. Handbook of Market Research, 1-37

Fairlie, R. (2018). Racial inequality in business ownership and income. Oxford Review of

Economic Policy, 34(4), 597-614

(33)

soft & hard skills and racial wage gap. Economic Inquiry, 55(2), 1032-1053 Feenstra, R. C., & Hanson, G. H. (1997). Foreign direct investment and relative wages:

evidence from Mexico’s maquiladoras, Journal of International Economics, 42(3-4), 371-393

Figini, P., & Görg, H. (2011). Does foreign direct investment affect wage inequality? An empirical investigation. The World Economy, 34(9), 1455-1475

Gelan, A., Fealing, K. H., & Peoples, J. (2007). Inward foreign direct investment and racial employment patterns in US manufacturing. The American Economic Review, 97(2), 378-382

Gopinath, M., & Chen, W. (2003). Foreign direct investment and wages: a cross-country analysis. Journal of International Trade & Economic Development, 12(3), 285-309 Gordils, J., Elliot, A. J., Jamieson, J. P., & Sommet, N. (2020). Racial income inequality,

perceptions of competition, and negative interracial outcomes. Social Psychological

and Personality Science, 11(1), 74-87

Herzer, D., Huhne, P., & Nunnenkamp, P. (2014). FDI and income inequality-evidence from Latin American economies. Review of Development Economcis, 18(4), 778-793 Herzer, D., & Nunnenkamp, P. (2013). Inward and outward FDI and income inequality:

evidence form Europe. Review of World Economics, 149(2), 395-422

Hoover, G. A., & Yaya, M.E. (2010). Racial/Ethnic differences in income inequality across US regions. The Review of Black Political Economy, 37(2), 79-114

de Hoyos, R. E., & Sarafidis, V. (2006). Testing for cross-sectional dependence in panel-data models. The Stata Journal, 6(4), 482-496

Huffman, M. L., & Cohen, P. N. (2004). Racial wage inequality: job segregation and

devaluation across US labor markets. American Journal of Sociology, 109(4), 902-936 Jaret, C., Reid, L. W., & Adelman, R. M. (2003). Black-white income inequality and

metropolitan socioeconomic structure. Journal of Urban Affairs, 25, 305-334 Jensen, N. M., & Rosas, G. (2007). Foreign direct investment and income inequality in

Mexico, 1990-2000. International Organization, 61(3), 467-487

Johnsons-Lans, S., & Jones, P. (2017). Foreign direct investment and racial wage inequality: Evidence from South Africa. In: Johnsons-Lans, S. (eds) Wage inequality in Africa. Global Perspectives on Wealth and Distribution. Palgrave Macmillan

Kerr, C. & Walsh, R. (2014). Racial wage disparity in US cities. Race and Social Problems, 6(4), 305-327

Kids Count Data Center (2020) Educational attainment of working age population 25 to 64 in

the United States. Retrieved from

https://datacenter.kidscount.org/data/tables/6295-

educational-attainment-of-working-age-population-25-to-64#detailed/1/any/false/1729,37,871,870,573,869,36,868,867,133/1311,1304,1264,12 65,1309/13092,13093

Klowden, K., Lee, J., & Ratnatunga, M. (2018). State Technology and Science Index. Milken

Institute. www.milken.org

Leicht, K.T., & Jenkins, J.C. (2017). State investments in high-technology job growth. Social

Science Research, 65(1), 30-46

(34)

Lucyna, K., & Ekanayake, E. M. (2016). State based factors affecting inward FDI

employment in the U.S. economy, International Journal of Research in Business and

Social Science, 1(1), 1-7

Mah, J. S. (2002). The impact of globalization on income distribution: the Korean experience,

Applied Economics Letters, 9(15), 1007-1009

Mahler, V. A., Jesuit, D. K., & Roscoe, D. D. (1999). Exploring the impact of trade and investment on income inequality: a cross-national sectoral analysis of the developed countries. Comparative Political Studies, 32(3), 363-395

McCall, L. (2001). Sources of racial wage inequality in metropolitan labor markets: racial, ethnic, and gender differences. American Sociological Review. 66(4), 520-541

Mihaylova, S. (2015). Foreign direct investment and income inequality in Central and Eastern Europe. Theoretical and Applied Economics, 22(2), 23-42

Nielsen, F., & Alderson, A. S. (1997). The Kuznets Curve and the great U-turn: income inequality in U.S. counties, 1970 to 1990. American Sociological Review, 62(1), 12-33 Nunnenkamp, P., Schweickert, R., & Wiebelt, M. (2007). Distributional effects of FDI: how

the interaction of FDI and economic policy affects poor households in Bolivia.

Development Policy Review, 25(4), 429-450

Pesaran, M. H. (2006). Estimation and inference in large heterogeneous panels with a multifactor error structure. Econometrics, 74, 967-1012

Reuveny, R., & Li, Q. (2003). Economic openness, democracy, and income inequality: an empirical analysis. Comparative Political Studies, 36(5), 575-601

Semyonov, M., & Lewin-Epstein, N. (2009). The declining racial earnings’ gap in United States: multi-level analysis of males’ earnings, 1960-2000. Social Science Research, 38(2), 296-311

Setzler, B., & Tintelnot. F. (2019). The effect of foreign multinationals on workers and firms In the United States. NBER Working Paper Series, 26149

Siegel, J., Pyun, L., & Cheon, B. Y. (2019). Multinational Firms, Labor Market

Discrimination, and the Capture of Outsider’s Advantage by Exploiting the Social Divide. Administrative Science Quarterly, 64(2), 370-397

Smith, A. K., Black, S., & Hooper, L. M. (2020). Metacognitive Knowledge, Skills, and Awareness: A possible solution to enhancing academic achievement in African American Adolescents. Urban Education, 55(4), 625-639

Stolper, W., & Samuelson, P. A. (1941). “Protection and Real Wages,” Review of Economic

Studies, 9, 58-73

Suanes, M. (2016). Foreign direct investment and income inequality in Latin America: a sectoral analysis. Cepal Review, 118, 45-61

Sylwester, K. (2005). Foreign direct investment, growth and income inequality in less developed countries. International Review of Applied Economics, 19(3), 289-300 Tsai, P-L. (1995). Foreign direct investment and income inequality: further evidence. World

Development, 23(3), 469-483

United States Bureau of Labor Statistics (2020) employment status of the civilian

noninstitutional population in states by sex, race, Hispanic or Latino ethnicity, marital status, and detailed age. Retrieved from https://www.bls.gov/lau/ex14tables.htm

(35)

https://www.bea.gov/data/gdp/gdp-state

United States Census Bureau (2020) Income and Poverty in the United States: 2019.

Retrieved from https://www.census.gov/library/publications/2020/demo/p60-270.html United States Census Bureau (2019) National Population Totals and Components of Change:

2010-2019. Retrieved from

https://www.census.gov/data/tables/time-/demo/popest/2010s-national-total.html

United States Census Bureau (2019) USA Trade Online. Retrieved from https://usatrade.census.gov/

OECD (2020). FDI in figures, April 2020. Retrieved from

(36)

Appendix

Table A1 State racial income inequality ratios

State Ratio State Ratio

Alabama 0.58 Missouri 0.63

Alaska 0.68 Nebraska 0.55

Arizona 0.75 Nevada 0.67

Arkansas 0.61 New Hampshire 0.73

California 0.67 New Jersey 0.60

Colorado 0.69 New Mexico 0.81

Connecticut 0.56 New York 0.63

Delaware 0.71 North Carolina 0.63

Florida 0.69 North Dakota 0.49

Georgia 0.65 Ohio 0.54

Hawaii 0.90 Oklahoma 0.62

Illinois 0.54 Oregon 0.61

Indiana 0.59 Pennsylvania 0.58

Iowa 0.51 Rhode Island 0.57

Kansas 0.61 South Carolina 0.57

Kentucky 0.65 South Dakota 0.61

Louisiana 0.51 Tennessee 0.68

Maine 0.51 Texas 0.70

Maryland 0.71 Utah 0.64

Massachusetts 0.59 Virginia 0.63

Michigan 0.57 Washington 0.68

Minnesota 0.48 West Virginia 0.70

Mississippi 0.54 Wisconsin 0.49

(37)

Table A3 Hausman test

Chi-square 15.20

Prob>chi2 0.0188

Table A4 Correlation matrix

Variable INC FDI BLACK EDUC GSP STSI TRADE UNEMP FDIBLACK

Racial income inequality 1.0000 FDI 0.0475 1.0000 Black employment -0.0667 0.2725 1.0000 Education 0.1286 -0.1207 0.2187 1.0000 GSP -0.0187 0.3123 -0.0910 -0.6350 1.0000 STSI 0.1122 0.1763 -0.0431 -0.7367 0.5395 1.0000 Trade openness -0.2315 0.3352 0.4234 0.2425 -0.1500 -0.0534 1.0000 Unemployment 0.0572 -0.0414 0.2771 0.3464 -0.3586 -0.0374 0.3471 1.0000 FDI * Black employment -0.0419 0.6244 0.8892 0.0899 0.0878 0.0789 0.4678 0.1895 1.0000

Table A5 VIF test

Variable VIF 1/VIF

Education 3.21 0.311146 STSI 2.64 0.378815 GSP 2.01 0.496901 Unemployment 1.50 0.667872 Log trade 1.48 0.673481 FDI lagged 1.38 0.722585

Log black employment 1.32 0.757480

Mean VIF 1.94

Table A6 Breusch-Pagan test

(38)

Table A7 Wald test

Chi2(46) 25860.82

Prob > chi2 0.0000

Table A8 Woolridge test

F (1,45) 1.733

Prob > F 0.1947

Table A9 T-Test

F (2,318) 4.07

Prob > F 0.0180

Table A10 Pesaran’s cross-sectional independence test

The impact of foreign direct investment on racial income inequality: evidence from a panel of US states