Taylor Chappie

1. INTRODUCTION

      The United States is known as a melting pot and immigration has caused the diversity of our country. As of 2014, 13.3 percent of the total population was made up of immigrants and between 2013 and 2014 the foreign-born population increased by 2.5 percent. Many immigrants come to this country looking to pursue the “American Dream”, but is their pursuit having a positive or negative impact on the native population that already lives here? It is apparent that immigration has always been an important part of this nation’s history, but within the last century, researchers have taken a closer look at the actual effects of immigration on native wages, the labor market, and the economy.

1.1 Background Legislation

     The first immigration policy was enacted in 1870, which specified who could become a citizen. The law limited that privilege to free whites of “good moral character”. Since then, many laws have been enacted and changed to fit the needs of the country. In 1965, the Immigration and Nationality Act created a system that favored family reunification and skilled immigrants, rather than meeting a quota. Another important law came into effect in 1986. The Immigration Reform and Control Act granted legalization to millions of unauthorized immigrants who met certain conditions. The most recent policy change was in 2012, which allowed young adults, who were in the country illegally, to apply for deportation relief and a work permit.

     Many studies have conflicting results, which makes it hard to understand what policies should be put in place and which shouldn’t. For instance, Islam and Khan (2015) conclude that in the short run immigration does not have a negative effect on native wages, but in the long run there are negative effects due to the surplus of labor. Enchautegui’s (1995) results indicate that native wages are not negatively impacted by immigration and, in fact, can be positively affected in some areas with medium to high immigration levels. With so many studies that contradict each other, how is anyone supposed to know the true effects of immigration on state wages?

     This paper analyzes the change in real mean annual wages of each state with a range of independent variables for 2013 with the objective of getting an understanding of the current effects of immigration. Since the data set is so recent, it takes into account all the major policies that have been put in place. The results of the analysis could help states decide if there is a need for policy reform.

1.2 Outline of Paper

     This paper attempts to determine if immigration plays a significant role in determining the wages that are earned in each state. The second section will give insight into previous literature on the effects of immigration. In the third section, we will discuss the data, variables, descriptive statistics, and expected signs in our research. Methodology of the paper will be included in the fourth section. The fifth section will discuss regression results and provide interpretation of those results. In the final section, conclusions and extensions of the research will be discussed.

1. LITERATURE REVIEW

     Immigration and its effects has always been a popular topic to study. One of the most prevalent concerns when talking about immigration is its effect on the labor market. The literature focuses heavily on the economy, wages, and jobs when analyzing immigration. Each of the following studies examines these categories using multiple models and techniques.

     In one study, Enchautegui (1995) focuses on the effects of immigration on male hourly wages in the United States. The study takes into account multiple variables such as education, experience, industry of employment, size of the labor market, and region. The time frame of the study is 1980 to 1990. The data was collected from the U.S. Censuses of Population and Housing. Enchautegui (1995) uses ordinary least squares and instrumental variable techniques to estimate the equations. The results of the study suggest that immigration has no negative effect on native male hourly wages and actually positively affected wages in some areas with medium to high levels of immigrants. Enchautegui (1995) concludes that there is no reason to believe that the United States’ economy cannot absorb more immigration and that immigration does not put the native workers at risk.

     SrungBoonmee (2013) conducts an analysis of how U.S. native workers’ wages are affected by the varying levels of education of immigrants in the work force. The study uses U.S. census data from 1980-2000 and restricts its estimates to capable working adult males (IPUMS). An instrumental variable approach along with OLS estimation was used throughout the study. Weekly wages were calculated by dividing annual wages by number of weeks worked and then were deflated using the CPI. The results indicate that there was a 3-4 percent decrease in wages of native-born citizens when the highest level of education was high school and there was a 1-percent increase in local immigrants with similar education levels. The study also indicates that there was an increase in wages for native-born citizens, 4-percent for high school graduates and 8.1-percent for high school dropouts, when a 1-percent increase in immigrants with some college education occurred.

     In another study, Kim and Sakamoto (2013) investigates the difference that can occur when taking a spatial approach versus an occupational approach to analyze the impact of immigration on native workers’ wages. They used the Current Population Survey Outgoing Rotation Group (CPS-ORG) data from 1994-2006. They argue that the spatial approach does not accurately reflect the impact of immigrants on native workers’ wages; instead they suggested that a cross-sectional comparison with occupational-specific variables would yield more reliable results. The results of the study show that a 10-percent increase in low-skilled immigrants would reduce hourly wages in low-skilled occupations by 1.1-percent. Their results also indicate that high-skilled immigrants had a positive effect on wages, but with no significance.

     Islam and Khan (2015) investigate the relationship between GDP per capita, immigration, and real wage rates in the USA. They used a Johansen-Juselius co-integration technique to test for long-run relationships and a Granger causality test with a vector error-correction model (VECM) to account for short-run dynamics. Data was compiled from 1948 to 2002 using separate sources for real GDP per capita (Angus Maddison website), immigration (2007 Yearbook of Immigration Statistics), and average weekly earnings (Bureau of Labor Statistics). VECM estimates revealed causality between immigration and GDP per capita. When GDP per capita grows, the inflow of immigration will rise with it until it reaches a point where GDP per capita will stop and immigration rates will then fall as well. The short-run results suggest that GDP causes immigration and not vice-versa. Islam and Khan (2015) concludes that immigration does not affect average weekly earnings in the short run. There are, however negative effects on average weekly earnings in the long run due to increases in GDP per capita and immigration that eventually cause earnings to decline because of the excess labor supply.

3. DATA

     We obtained our data from the United States Census Bureau’s American Community Survey and the United States Bureau of Labor Statistics website. From the American Community Survey, we collected demographic data on the total population, native population, and foreign-born population. Our sample includes all 50 states and the District of Columbia for the years 2013 and 2014. We calculated the percent change from 2013 to 2014 for all of our variables. We collected mean annual wages and labor force participation rates for the year 2013 and 2014 from the Bureau Labor of Statistics website.

3.1 Variables and Hypothesis

     The variables, their descriptions, and expected signs can be found in Table 1. STWAGECHG is our dependent variable, which is the percent change in mean annual wages for each state.

We grouped our independent variables into four different categories: education, industry, labor, and population.                                                                                                                                                  In the population category, FBPOPCHG represents the percent change in the foreign born population in each state. POPCHG is the percent change in the total population of each state. FBPOPCHG and POPCHG are both predicted to have negative coefficients. Our negative prediction is influenced by the idea presented in Islam and Khan (2015), that when the population increases, the labor supply naturally increases as well, causing wages to decrease.

     The education category includes HSGRADCHG and LESSHSCHG. HSGRADCHG represents the percent change in percent of high school graduates based on each state’s population. LESSHSCHG describes the percent change in the percent of each state’s population with less than a high school diploma. LESSHSCHG has a negative expected sign and HSGRADCHG has a positive expected sign. These assumptions follow SrungBoonmee (2013), who suggests that less education has a negative effect on wages.                                                                                                    The labor category includes LFRTCHG and FBLFRTCHG. LFRTCHG is the percent change in the labor force participation rate of each state. FBLFRTCHG is the percent change in the foreign-born labor force participation rate of each state. LFRTCHG and FBLFRTCHG both have a positive expected sign because a higher labor force participation likely means wages are constant or increasing.

     The industry category includes AGRICCHG, MANUFCHG, SERVCHG, AGRIC2013, and SERV2013. AGRICCHG is the percent change in the percent of agriculture employment, MANUFCHG is the percent change in the percent of manufacturing employment, and SERVCHG is the percent change in the percent of service employment. AGRIC2013 represents the percent of agriculture employment in the year 2013. SERV2013 represents the percent of service employment in 2013. All of these variable’s expected signs are ambiguous because it is hard to know their impact on state wages.

3.2 Descriptive Statistics

            The descriptive statistics can be found in Table 2. Our dependent variable, STWAGECHG, has a mean of 0.018 with a standard deviation of 0.007. The range consisted of a minimum at -0.004 held by Nevada and a maximum of 0.045 held by North Dakota. FBPOPCHG has a mean of 0.016 with a standard deviation of 0.016. For FBPOPCHG the minimum is -0.014 held by Alabama and the maximum is 0.083 held by North Dakota.

3.3 Data Limitations

            The United States Census Bureau’s American Community Survey uses estimation techniques to collect its data. Our data was taken from the ACS’s 5-year based estimates. Although it is known to contain an acceptable amount of precision, we believe it is important to note the use of these estimates in our data. It is also important to note that our data focus’ on a short-term analysis. Immigration is a topic that differs year to year and in order to obtain a better insight into its effects, a larger study over many years would be more beneficial.    

4. ECONOMETRIC MODEL

      This study utilizes ordinary least squares (OLS) regression analysis to investigate the change in state wages from the effects of immigration. The overall econometric model is as follows:

           

 is the constant,  are the estimated coefficients of the independent variables, and  represents the stochastic error.

4.1 Econometric Procedures

            During our analysis, we tested for heteroskedasticity using the White test to verify that our models were not violating the Classical Assumption V. The results of the tests showed that our model was homoskedastic. We also performed an F-test on the industry group to test for joint significance, and determined that they were not jointly significant. We determined that using the percent change in the industry variables may not be the best use for our model. The variables had a lack of significance, which could be due to the small amount of change in industry sectors over a single year. Therefore, in model 2 we used the industry variables for the year 2013 to better reflect the current standings of the market. We also decided to drop manufacturing due to the decline of the manufacturing industries in the United States. Dropping that variable provided a better fit for our model.

5. RESULTS

     The results of Model 1, which can be found in Table 3, indicates that FBPOPCHG is significant at the 1% level, POPCHG and LESSHIGHCHG is significant at the 5% level, and LFRTCHG is significant at the 10% level. The model’s  is 0.612 and the adjusted  is 0.538. The small difference between the two values indicates that we have relatively few, if any, irrelevant variables in the model.

     LESSHSCHG has a significant negative impact on mean annual state wages. This shows that when a higher percent of the population has less than a high school diploma that wages in that area are expected to be lower. In other words, low education levels have a negative effect on wages, which is consistent with SrungBoonmee’s (2013) results. On average, for every 1% increase in the population with less than a high school diploma there is a 0.17% decrease in mean annual wages, ceteris paribus.

     POPCHG and FBPOPCHG have significant positive impacts on mean annual state wages. For every 1% increase in the total population a 0.37% increase in mean annual wages can be expected to occur, ceteris paribus. Similar results are shown with every 1% increase in foreign-born total population an expected 0.22% increase in mean annual wages should occur, ceteris paribus. These findings are consistent with much of the literature (Islam and Khan, 2015; SrungBoonme, 2013; Enchautegui, 1995). Our results indicate that immigration can actually be a helpful tool in increasing wages. 

     The results of Model 2 can be located in Table 3. The results indicate that POPCHG and SERV2013 are significant at the 1% level, FBPOPCHG, LESSHIGHCHG, and AGRIC2013 are significant at the 5% level and FBLFRTCHG is significant at the 10% level. The model’s  is 0.722 and the adjusted  is 0.684. The higher adjusted in model 2 indicates that it is a better fit than model 1 (adjusted = 0.538).

     The variables SERV2013 and LESSHIGHCHG both have significant negative impacts on the annual mean wages. LESSHIGHCHG is robust throughout model 1 and model 2 and indicates that less education results in lower wages, which still holds a consistent to the literature (SrungBoonme, 2013). For SERV2013, a 1% increase in the amount of employment in the service industry will result in a 0.17% decrease in mean annual wages, ceteris paribus.

FBPOPCHG, POPCHG, and AGRIC2013 are positively significant coefficients. FBPOPCHG and POPCHG are robust throughout model 1 and model 2, however FBPOPCHG drops significance from 1% to 5% and POPCHG gains significance from the 5% to 1%. For the variable AGRIC2013, a 1% increase in the percent of employment in the agriculture industry will result in a 0.06% increase in wages, ceteris paribus.

6. CONCLUSIONS

            Our results on the foreign-born population change, FBPOPCHG, are consistent with those of other studies (Islam and Khan, 2015; SrungBoonmee, 2013; Enchautegui, 1995). The foreign-born population change variable was used to measure the amount of immigration during 2013. Our results indicate that during 2013 in the United States an increase in the foreign-born population increased mean annual wages. Therefore, concluding that immigration has a positive effect on state mean annual wages. Total population change, POPCHG, also increased mean annual wages. One reason for these results relates to the research done by Kim and Sakamoto (2013), who concluded that immigration can be a positive determinant on wages when the labor markets allow for it, by having an increased demand for workers. In our study, it is possible that in 2013 the labor market was in a state of demand where a population increase, including a foreign-born population increase, would cause wages to increase.

6.1 Policy Reforms

     Our results also indicate that there is no need for more stringent policy reform, but possibly more lenient policy reform. Since immigration positively impacts state wages, states could allow for an increase in immigration. Due to this study only being for one year, it might be helpful to further this study before making any permanent reforms.

7. FURTHER RESEARCH

     This study could be taken further to investigate the current impact of immigration using multiple years or even decades. Due to lack of time, our study was conducted for only one year, which could cause omitted variable bias and skew the results. It is hard to understand the full effect of immigration in just one year, therefore taking a look at the longer period of time could give a better understanding of the topic. Research could also be taken to understand the hypothesis of Kim and Sakamoto (2013), which concludes that immigration can be positive during times when workers are in higher demand.