Immigration Effect on Native Wages

Introduction

In 2013, 232 million people, or 3.2 percent of the world’s population were immigrants which is an increase from the past (UN 2013).^[1] There has always been a dispute between the philosophy of an on open door policy to the world and the fear of the economic and social impact that immigration might bring. The main questions in the debate regarding immigration policy are based in economic incentives. Major attention has been drawn to the possible adverse impact on the labour market outcomes immigration may cause. (Friedberg and Hunt, 1995)

Although research conducted on the economic impact of immigration has shown that the benefits outweigh the costs it is still opposed in some countries. The native population has become more concerned that immigrants may compete with native-born workers, which will either depress wages in the labour market or force the native workers into unemployment. (Brucker et al, 2014) Contrary to the these views, there are benefits to immigration, immigrants may complement some native factors in production, increasing the benefit in these factors and therefore help improve overall economic welfare.

To get a better understanding of the effects on wages, determining the profile of a migrant is important. There are a variety of sources of migration, for example citizens in Europe enjoy the freedom of free movement though most countries in the continent. However, in Canada and the United States managed labour migration plays a major role. During an interview with Dagens Nyheter, George Borjas states that immigration contributes to an uneven income distribution and that the policies that governments use to reduce income inequality are offset by international migration. The appropriate immigration policy for a country is dependent on what type of society the government wants to create. (DN Fokus, 2016) In order to implement labour market policies to achieve the desired society, it is important to study the actual impact of immigration on individuals with different skill sets in the Canadian labour market.

This study will focus on labour markets in this study to determine the overall effect that immigration has on wages. One of the most important determinants as to whether immigration is a benefit is the labour markets in the host country. Over the past 10 years migrants accounted for 47% of the increase in the workforce in the U.S. and 70% in Europe (OECD)^[2].The understanding of immigration impacts will help policies in education and employment such that they maximize the benefit for the host country. Several policies have been discussed in order to help integrate immigrants into the labour market. One policy discussed is to lower the entering wage for immigrants to prevent the expected higher burden on public finances caused by a higher unemployment rate among immigrants. (Hartog, 2000) An investigation into how immigrants will affect native born worker wages and unemployment in the labour market, one can predicts how policy reforms will affect the overall labour market within a country. According to Borjas, lowering initial wages for immigrants will result in native born workers being replaced by immigrants causing native born workers into a higher unemployment rate as it is less expensive to hire an immigrant. (DN Fokus, 2016)

Using panel data, this study investigates and interprets the estimated outcome of how wages for native born workers in the Canadian labour market respond to immigration in Canada. The research question that this study will attempt to answer is: How does immigration in the short and medium-run impact the wage level of native born workers in the Canadian labour market. The main findings, when controlling for age, unemployment, with differences between the year and municipalities in this study in the short-run are in line with the theory. The closer to a substitute that a native-born and foreign born workers are, the greater the adverse effect on the wages of native born workers when it is assumed that immigrants are low skilled. (Borjas, 1994) The effect on wages for high skilled native workers in the short run when assuming immigrants and natives as complements, is positive. This implies that the wages of high skilled natives increases as the share of immigrants increases. The effect on high skilled native born workers wages is positive even in the medium run and adverse for the low and medium skilled native workers. According to theory this is not an unexpected result as the wage is expected to be unaffected in the medium-run. (Borjas 1994) It is worth noting that this may be a result of errors in the assumption that immigrants are low skilled. Also the time-frame for the research may be too short of a time to see the expected effect in the mid-long run.

The impact of immigration may differ across countries which depends on difference in labour markets and welfare states. Policies such as minimum wage, unions with collective bargaining agreements and employee protection may affect how wages respond to changes in immigration. (Brucker at al., 2014) Given these differences and the complexity involved, this study will only analyze the Canadian labour market.

Economic Theory

Demand and Supply

Several theories try to explain how labour market outcomes and earnings of the native born are affected by labour supply shocks and other factors over time in the short and long run, respectively. Perhaps most famously, George Borjas in his 1994 paper tried to explain the impact of immigration on natives’ wages. In the paper, labour supply shocks referred to an increase in the workforce of the host country due to immigration. When immigrants enter the labour market labour supply increases.  An immigrant can be either a substitute or a complement for the native worker. If the immigrant is a substitute this means that they are equally productive, have the same skill set and this are competing for the same type of jobs.  Using a simple supply and demand model, an increase of substitutable workers will result in an increase in the supply of labour, causing wages to fall for workers with similar skills. However, when workers are complementary an increase in migrant labour can increase job opportunities and wages for native born workers thus immigrants and natives complement each other in the labour market. To illustrate this, consider that the low-skilled immigrant labour reduces the cost of production and increases the output of goods, this increased output increased the demand for more higher-skilled native workers.

Borjas (1994) first starts by assuming that immigrants and natives are perfect substitutes and that there is perfect competition in the labour market with capital held fixed in the short run. The outcome in this demand and supply model is that as immigrants enter the work force, the work force increases and the labour supply curve shifts out to the right. The result of this shift is an increase in the total employment and a reduction in wages. Borjas finds the decrease in the share of native workers willing to work for the lower level of earnings, an upward sloping labour supply curve. As the total employment level increases the employment level of native workers decreases. This implies that native workers do not find it worthwhile to work for the lower level of wages (Borjas, 1994).

Then, Borjas considers the alternative case, where immigrants are complements to native works rather than substitutes. In this case, the immigrants may differ in terms of productivity and education compared to native workers. As mentioned above, as complements the employment of an immigrant worker will help the wages and/or employment of a native-born worker. As complements, an increase in the number of immigrants will raise the marginal product of native workers and shift the demand curve for native workers to the right. The results imply an increase in native worker productivity and wage level (Borjas, 1994). Another finding is that native workers who did not previously find it profitable to work, enter the labour market since wages have increased, thus native worker employment increases in this case (Borjas, 1994).

Minimum Wage

Research conducted by Andri Chassamboulli in 2013 takes the basic migration theory and applies it to a market with a minimum wage. It states that an increase in immigration will negatively affect the wage level of low skilled native workers assuming that they are substitutes to immigrants. The results depend on two things, first on the presence of a minimum wage in the host country and second on the determination of the minimum wage which depends on the bargaining power of workers.  An increase in immigration results in the wage level decreasing in the host country, thus firms expect to pay lower wages, this lower cost allows them to offer more jobs which puts native workers in a better bargaining position. The higher availability of jobs which is a result of decreased wages puts native workers in a better bargaining position which helps put an upward pressure on the wage level. The native worker’s marginal product falls as well due to the high quantity of low skilled labour. The result is thus ambiguous. In the presence of a minimum wage, low skilled workers earned the minimum wage rather than a wage determined by bargaining power (Chassamboulli, 2013)

Long Run Effects

Another study by Sommerville and Sumption (2009) reviews a number of previous studies and summarize the results that the impact immigration has on the host country’s wage level which is relatively small. The results are explained by the theory that immigrants contribute to an increase in demand for domestic products and services that they consume themselves. This increase in demand leads to an increase in production and in many cases increases the demand for labour and long run economic growth that also puts an upward pressure on the wage level (Sommerville and Sumption, 2009). The study states that in a small open economy wages are expected to return to the pre-immigration level.

An important variable to determine is the length of short-run. Immigration is a flow, not a stock, the short-run is the time it takes for businesses to adjust their investment levels or in other words the time it takes for them to ‘absorb’ the increase in the labour supply. This also means that estimating the effects of what what impacts an influx of immigrants create, is an empirical exercise rather than a theoretical one (Sommerville and Sumption 2009).

The short-run results on the labour market from a labour supply shock differ from the long-run results. Using the assumption that immigrants and native workers are perfect substitutes, theory implies that when low-skilled workers immigrate to Canada it raises the level of labour supplied of low-skilled workers. This increase in labour supply negatively affects the wages and unemployment level for native workers. As for firms, the returns to capital increase because labour costs are now decreased. Given that firms are profit-maximizing they will invest in more capital and this leads to an increase in capital stock and shift the demand curve to the right and as a result, offset the negative impacts of the initial labour supply shock (Sommerville and Sumption 2009). After this, the rate of return to the capital stock decreases back to the normal long-run fixed level. The theory by Borjas implies that the price in the labour market, the wage, is constant in the long-run. This means that in the long-run immigrants have no effect on the native worker wages. However, the theory does not define the term of long-run so therefore it is not clear when the effects actually occur. Further the theory proposed by Borjas (1994) states that the negative wage impact on native worker wages weakens over time.

Review of Past Literature

To understand how immigration affects the native workers’ wages, we will refer to “The Effect of Immigration along the Distribution of Wages,” by Christian Dustmann, Tommasso Frattini and Ian P. Preston. This study analyzes the effect that immigration has on the wages of native workers along the full distribution of native wages giving us a clearer understanding of where the wages are affected. In the theoretical model they develop they use many types of capital as factors of production, as opposed to just two. They find that when the immigrants skills are different than that of the native labour force and when capital is elastic in supply, the effect [of what? ]on average wages of the native workers should be zero or even slightly positive (Dustmann, Tommasso, and Frattini, 2011). Although this holds true in theory, the authors believe not enough attention has been given on the effect of immigration when the capital is assumed to be fixed which leads to the surplus going mainly to capital owners.. The study focuses on this wage distribution to see where the effects are felt differently. They believe that immigration could possibly depress the wages of workers who are in segments of the labour market where there is a higher density of immigrants compared to native workers (Dustmann, Tommasso, and Frattini, 2011). This effect is determined through an empirical approach. In their theoretical framework they allow for many different skill types, rather than pre-allocating immigrants to skill groups as per their observed characteristics. This approach is beneficial as it doesn’t restrict where immigrants compete with natives giving a better overall picture of wages across the labour force. The results show that the estimated wage effects along the wage distribution are in line with the observed location of immigrants. This leads to the conclusion that although immigration depresses wages below the 20^th percentile, it contributes to wage growth above the 40^th percentile.(Dustmann, Tommasso, and Frattini, 2011).  In a theoretical model where capital supply is positive and due to the workers being at different parts of the wage distribution, they find that the average effects of immigration on wages are slightly positive. (Dustmann, Tommasso, and Frattini, 2011). The results they find are in line with what they expected given the density of immigrants across the distribution. So to conclude, immigration leads to a decrease in wages at parts of the distribution where the density of immigrants is higher than the density of natives. If the density of natives is higher than the density of immigrants then it leads to an increase in wages in certain parts along the distribution. On average, across the distribution of natives, immigration leads to a slight increase in average wages (Dustmann, Tommasso, and Frattini, 2011).

David Card (1990) in his research estimated the result on the labour market caused after an increase in immigration. This influx of immigration was a result of a significant number of Cubans immigrating to the United States in 1980, this immigration is referred to as the Mariel Boatlift. Card (1990) compares the results in two different scenarios, one where the Cubans immigrate to the United States and another scenario where they don’t. The results from his research imply that the influx of immigrants had no effect on the wage level for the low-skilled native workers. Card (1990) and other studies (Grossman, 1982; Borjas 1987; Card 1991) have been criticized for two main reasons. One is that immigrants are more likely to migrate to cities in which wages are higher so that the endogeneity of supply shocks induces a spurious positive correlation between immigration and wages (Borjas, 2015).  Two, both firms and workers respond to the labour supply shocks by resettling in areas that offer better opportunities, thereby diffusing the impact of immigration across the country’s labour market (Borjas, 2015). Borjas (2015) states in his study that Card’s Mariel study is invulnerable to these criticisms as the ‘Marielitos’ didn’t migrate so much for the wages as they did for the fact that Castro’s decision was sudden and also the Cuban-American’s who arranged the boat lived in South Florida. As for the second criticism, Card’s analysis is done in the short-run in which the firm and native workers did not have much time to adjust to the labour supply shock (Borjas, 2015)

Kugler and Yuksel (2008) also conducted a similar study using a natural experiment, they estimated the effect of native worker wages and previous immigrant wages after an influx of immigrants that resulted from Hurricane Mitch in 1998. Both studies have one major thing in common in that a significant number of the immigrants were among the less educated. Kugler and Yuksel (2008) results showed an increase in native worker wages as a result of the influx of immigrants. The different results from the two studies can be a attributed to the fact that the immigrants from the Kugler and Yuksel study quickly become legalized whereas the ones in Card’s study did not (Krugler and Yuksel, 2008).

Brucker at. Al. (2012) estimate the labour market effects of immigration by using micro data and investigate the effects on different skill levels for natives, determined by education, experience and national origin. The three countries used in the study are Germany, Denmark, and England and the assumption made is that all three countries have very different labour markets. These differences include unemployment benefits, different levels of minimum wages, and bargaining opportunities. With these differences the countries are expected to respond differently to labour supply shocks such as immigration. Brucker et. al. (2012) make the assumption that there is imperfection competition whereas, most studies assume perfect competition when studying the effect of immigration on wages in the labour market. By measuring the elasticities of the wage setting curves and the elasticities of substitution between different types of labour they solve for the wage effects of immigration for different groups. The study finds that since England has more wage flexibility the unemployment effect is not as large as it is in Germany and Denmark. This implies that wages respond more to a change in England’s labour market.

A study conducted in the Danish Labour market aimed to estimate the effect of an increase in low educated immigrants on equivalent natives. (Foged and Peri, 2014). The study focuses mainly on an influx of non-European immigrants. The study follows individuals over time and and captures the effect of immigrants that may spill over to other municipalities and regions through movement and mobility. This also allows them to see how their wage level is affected over time. In the study, municipality and individual based responses to immigration inflows over time are compared to determine whether the spillovers have a big role when estimating the impacts of immigration. The results in the study show that an increase in labour supply which consisted of low educated non-European immigrants resulted in a reallocation of the equally low educated natives to more ‘smart’ job positions (Foged and Peri, 2014). The wage levels either increased or were not affected at all.

In Norway, the study by Bratsberg et al. (2010) also estimates the effect on native wages as a result of a labour supply increase due to immigration. This study differentiates itself by splitting the immigrants share by origin. The change in the immigrant share works like an instrument for an increase in labour supply within skill groups and any wage adjustment will identify the slope of the labour demand curve. The different skill groups are defined by both education and work experience with four levels of education and eight levels of experience. In contrast to the results from Foged and Peri (2014) this study shows negative results on the native-born worker wages after an increase in the share of immigrants. The impact is more negative when the immigrants are born in other Nordic countries and less negative if the immigrants are from developing countries (Bratsberg et al., 2010)

Methodological Framework

To examine whether immigration affects native worker wages in Canada negatively this study will set up an ordinary least squares equation as follows:

1. Wage_cmt=β₁ + β₂ I_(t-n) + e , where:

Wage = wage

m= municipality

c = natives

t = time period t

β₁₌ constant, intercept

β₂ = the effect of the share of immigrants on native wages

I= share of immigrants

(t-n) = time period t minus number of years

e = error term

This model estimates how the share of immigrants in Canada relates to all the native worker wages in each municipality. In this regression, the impact of the effect is represented by β_2, the value and sign of the parameter will indicate how native wages are affected by immigrants. If β_2, is negative, this will imply that an increase in immigrants has a negative effect on wages. Whereas, if it’s positive, the effect on wages is positive, thus as share of immigrants increases, native wages increase.

This study investigates the impact of immigration on different skill groups in Canada. In order to investigate this impact, the native population is divided into three different skill groups as follows: low skilled, medium skilled and high skilled. Education is a measure of skill, therefore, three different education levels of natives is chosen to study the effect on each different skill group.

Immigration affects the wage level of highly educated natives differently when compared to natives with a low education level. An increase in immigrants will have an overall effect on the wage level for all natives but the level of education among natives determines the extent of that effect in the particular skill group. In the regression there is a possibility that the impact of variable varies across differing groups. The education level among native workers is integrating with immigration. In other words, the level of education modifies the effect that immigration has on native workers wage level.

In order to determine how different levels of immigration effect wages in different skill groups, an interaction term is created, I_shareeduc. To create this term, the dependent variable, natives’ wages and two independent variables, immigration and education level are used. To create the interaction term , the independent variables are multiplied. Then a regression would be run, this time consisting of the independent variable but also the interaction term, I_shareeduc. To further his, new interaction variables are generated that show share of immigrations and the different educations levels, I_shareloweduc, I_sharemideduc, and I_sharehigheduc, one variable for each education level. This gives us a new equation:

2. Wage_cmt = β₁+β₂I_(t-n)+β₃EDU_high_i(t-n)+β₄I_sharehigh_(t-n)+e

The new variable is generated as the wage for native workers may be affected by an interaction between the share of immigrants and the different levels of education. An interaction between the variables means that their effect on native workers’ wage may not just be additive but rather multiplicative (Gujarati and Potter, 2009). Given the sign of the parameter of the variable it can be determined whether the impact on native wages is negative or positive and also the extent to which the immigrants affect the different created education groups’ wages. Assuming that the interaction parameter is positive, the real effect that immigration has on native worker wages has strengthened because the certain level of education has an impact on the effect of share of immigrants which thereby strengthens the actual impact.

This study uses panel data instead of cross section data. One benefit of using panel data is that it is easier to identify and measure the effects that cannot be observed in cross sectional data. For example, if the share of immigrants over time for the same group is changed it allows for better results when looking at the effect that immigrants have on native worker wages. The data sets from panel data have less variability, less collinearity among variable, more degrees of freedom and efficiency. Cross section data is collected in the same point of time, the problem this creates is that the probability of measuring the impact of the typical immigrant in differing groups on native worker wages is plausible. This is because immigrants in different groups have different skills, therefore the empirical result based on this type of data cannot be used to estimate the extent of the labour market outcome in the future because future immigrants may consist of more or less skilled immigrants (Borjas and Tienda, 1987)

There are also limitations in using panel data, for instance the variables in the data can vary over time and across the cities. This studies purpose is to determine the impact on immigrants not the variation over municipalities and years this is why a fixed effect is added to the equation. Adding a fixed effect can be used to account for the individuality of the units. The intercept of observations may vary over time and municipalities, in order to prevent the intercept from varying, area and a fired year effect is added (Gujarati and Potter, 2009).

θj and μtrepresent the area fixed effects and year fixed effects, respectively.

3. Wage_cmt = β₁+β₂I_(t-n)+β₃EDU_high_i(t-n)+β₄I_sharehigh_(t-n)+

   θj+

   μt+e

The fixed effect removes invariance across municipalities and time in years. In addition two control variables are added to ensure that other factors that may influence the result by changing over time stay constant. The age earning profile for workers in the labour market is upward sloping and concave in the Human Capital Theory. Borjas 2013 states that older workers earn more since they invest less in human capital and they are collecting returns from investments they made previously. As a result this may lead to biased results in the regression since the wage workers earn may be affect by age. Therefore age is one of the control variables since the desired result is the impact of immigration rather than the effect of aging. However, there is no average age in the data set from the Labour Force Survey for natives in the different cities. Even though research has shown the immigrants average age is lower than natives (Migrationsinfo.se, 2016) in order to make the regression possible an assumption is made that immigrants do not affect the average age in the municipality. This assumption may result in a biased outcome. The other second control variable in the equation is unemployment among native workers. From macroeconomic theory, wages may be affected by the unemployment rate. After adding the control variables the following equation is formed:

4. Wage_cmt = β₁+β₂I_(t-n)+β₃EDU_high_i(t-n)+β₄I_sharehigh_(t-n)+

   θj+

   μt+

β₅age_cm(t-n) + β₆unemp_cm(t-n) + e

The parameters for age and unemployment will determine if the result before controlling for the variables is either over/underestimated. If the β₅ and β₆ are negative this implies that the previous result was overestimated. This means that the impact on the result variable is larger than the actual effect as age and unemployment are not taken into consideration when analyzing the effect of immigration on wages. A positive parameter implies an overestimation. As a result there would not be a differentiation between the impact of age and unemployment and impact of immigration on wages. The outcome would include the effect of not only immigration but also age and unemployment leading to a biased outcome.

Data

The data used in this study in order to investigate the research question is obtained from Statistics Canada. A Canadian agency that provides most of the countries public statistics. In order to analyze the impact on migration, data covering information on level of education for natives, average wage for natives, share of immigrants, average age and unemployment will be used from the different municipalities from years 2006 to 2015. Observations collected will be panel data, which gives the opportunity to study the outcome of resulting changes in the chosen variables over time.

The level of education measures, the level of education among natives in the Canada across the different municipalities. The variable is divided into three different categories; low education, medium education, and high education. Low education consists of natives with an education level, those who have not obtained a high school diploma; the individuals with an approved high school degree are in the category medium education while high education consists of natives who possess the highest level of education. Those with at least three years more than high school. Education is classified to make it possible to investigate how migration affects groups with different skill levels. The study is divided into a short run and medium-long run analysis thus the amount of observations in different terms will vary. The total amount of observations for the different education levels is 2752.

The observations in the data set used for the short-run consists of 2321 observations and the number of observations in the medium-long run are 1198. The short-run is defined as one year therefore the independent variables of year t-1 affect the outcome variable in year t. Thus, the number of observations in the short-run and medium-long run will differ. When analyzing the effect in the short run for the observations for the independent variable are from years 2006 to 2014 and years 2007 to 2015 for the dependent variable. The medium-long run is 5 years (year t-5 independent variable affects the outcome variable of year t). As the data used in the study is from 2006 to 2015 this limits the study in using the same amount of observations as used in the short run. The observations in the medium-long run are from the years 2006 to 2010 for the independent variable and for the dependent variable from years 2011 to 2015.

The average wage data for native workers is taken from CANSIM from Statistics Canada. It provides the study with data regarding the average wage of native works in different municipalities. Wages are measured at an annual level by occupation and economic region provided by the JVWS. The total number of observations is 2752, in the short-run from years 2007 to 2015 consists of 2321 observations. In the medium-long run there are few observations as mentioned above, 1198 from the years 2011 to 2015. Share of immigrants is one of the more important variables in the regression. Statistics Canada defines an immigrant as anyone born outside of Canada. The data set of immigrants contains a total of 2783 observations. In the short run 2321 observations are used in the regression and 1198 in the medium-long run.

There are two control variables in the equation, age and unemployment. The unemployment rate is the share of natives who are considered unemployed by Statistics Canada and the age is measured as average age in the municipalities. The dataset of age and unemployment consists of 2752 observations of which 1198 are used in the medium-long run and 2321observations in the short-run.

Table 1. Summary Statistics for municipalities years 2006 to 2015[3]

Variable	Obs	Mean	S.D.	Min.	Max.
Natives’ Wages	2752	338.21	67.25	189.31	771.37
Share Immigrants	2752	0.214	0.0613	0.02	0.53
Age	2752	41.1324	2.1158	32.3	47.5
Unemployment	2752	0.0745	0.0216	0.032	0.13
Low Edu	2752	0.2301	0.0466	0.0761	0.3963
Medium Edu	2752	0.5561	0.0523	0.2784	0.7343
High Edu	2752	0.2603	0.0782	0.1291	0.6867

The table above is a descriptive statistics table that summarizes the characteristics of the sample of the municipalities. From the results, the average way for a native born worker is $338.21. There is significant difference between the municipalities with the lowest wage, $189.31 and the highest wage $771.37. Similarly, the share of immigrants from the municipality with the lowest share of immigrants, only 2% of workers are born outside of Canada. Whereas, in the municipality with the highest share of immigrants, 53% of the workers are born outside of Canada. On average the Canadian population contains 21.4% of immigrants. The municipality with the youngest average age is 32.3 while the municipality with the highest population has an average population of 47.5.  The average age in Canada is about 41 years. Looking at the education levels in Canada the majority of native born workers are in the medium education level with 55.61%. About 23% of native born workers are in the low education level and 26% are in the high education level. There is a significantly large variation in the education levels across municipalities. This is evident by the standard deviation in the education variables and the range between the minimum and maximum variables. For instance, the standard deviation for natives in the low education level is about 4.66% it is about 7.82% for natives in the high education level. For the low education level the share of low educated natives is about 7.61% for the minimum value and the municipality with the highest share natives in the low education level is about 39.63%. The values for the medium and high education levels can be seen on the table.

The data set used in this study is not at a micro level which would make the outcome more reliable. The data is an aggregate from CANSIM. Since the study is conducted on a municipal level rather than national this allows for more observations in total and in theory allows for a more reliable outcome.

Results & Analysis

 

Short Run

The equations used in this analysis are discussed in the methodological framework. The assumptions made in the short-run analysis is that immigrants on average have a lower education level than natives and as a result are considered substitutes to low skilled native born workers. Also capital is held fixed in the short-run. The regression results are presented in the table below.

Table 2. Short-run results from equation (1).[4]

Variables	Natives’ wages
Share Immigrants	337.31*** (89.7)
Constant	281.76*** (9.12)
Area fixed effect	No
Year fixed effect	No
Observations	2321
R-squared	0.1891
Standard error in parentheses	***p<0.01,**p<0.05,*p<0.1

The first regression encompasses all natives and measures the impact on all native workers from an increase in the share of immigrants. The results in the table show a positive and significant share of immigrants of 337.31 at a 1% significance level. This implies that if the share of immigrants increases by 1% that native worker wages will increase by $337.31 Canadian dollars. The R² value is low which means there is not much correlation between the share of immigrants and native workers’ wages. This may be because the first regression model used has few variables by adding more variables it may be possible to increase the explanatory power of the model.

Next the regressions have the native born workers divided into three different skill groups determined by their respective education levels. The second table includes the fixed effects.

Table 3. Short-run results from equation (2) [5]

Variables	Natives’ wages
Share Immigrants	236.80*** (39.83)
High Education	349.6***(21.92)
Share Immigrants effect on high skilled natives	-140.31***(138.45)
Constant	209.26*** (1.42)
Area fixed effect	No
Year fixed effect	No
Observations	2321
R-squared	0.454
Standard error in parentheses	***p<0.01,**p<0.05,*p<0.1

 

 

 

 

Table 4. Short-run results from equation (3)

Variables	Natives’ wages
Share Immigrants	-162.43*** (24.72)
High Education	-49.74***(6.69)
Share Immigrants effect on high skilled natives	383.98***(47.42)
Constant	380.56*** (5.84)
Area fixed effect	Yes
Year fixed effect	Yes
Observations	2321
R-squared	0.993
Standard error in parentheses	***p<0.01,**p<0.05,*p<0.1

Equation (2) when measured on high skilled natives[6] shows a negative outcome of -104.31, which is not significant. The native born workers who are categorized as high-skilled will see their wages decrease by $104.31  Canadian dollars when the share of immigrants increases by 1%. The R² is much larger in the regression from equation (2) when compared to equation (1). This implies that the more of the variance in wages (dependent variable) is explained by the share of immigrants (independent variable).

In the next regression, equation (3) is used in which fixed effects are added to the regression. In this case the impact that the share of immigrants have on high-skilled native worker wages is positive at a 1% significance level.[7] This implies that native worker wages increase by $383.98 Canadian dollars when the share of immigrants increases by 1%. The R²after the fixed effects are added is even greater than the previous two regressions. A R² of 99.3% implies that almost all of the variance in the wages of high-skilled native workers is caused by changes in the share of immigrants. The next regression includes the two control variables, age and unemployment.

Table 5. Short-run results from equation (4) [8]

Variables	Natives’ wages
Share Immigrants	-227.31*** (24.46)
High Education	-35.12***(6.58)
Share Immigrants effect on high skilled natives	285.34***(46.63)
Age	-5.28***(3.112)
Unemployment	-79.33***(13.62)
Constant	209.26*** (1.42)
Area fixed effect	Yes
Year fixed effect	Yes
Observations	2321
R-squared	0.993
Standard error in parentheses	***p<0.01,**p<0.05,*p<0.1

 

The results once again show a positive effect between the share of immigrants and high skilled native workers, it does however, decrease slightly from the previous regression in equation (3) as the control variables are added. At a 1% significance level the parameter is significant, the result implies that a 1% increase in the share of immigrants increases native worker yearly wages by $285.34 Canadian dollars.[9] The control variable parameters, age and unemployment are both negative which leads to the conclusion that the previous regressions’ results are overestimated. Taking into consideration the theoretical model in Borjas (1994) the results so far are in line with the theory. The closer that immigrants and native born workers are to substitutes the larger the negative effect on wages in the short-run given the assumption that immigrants are considered low-skilled and complements to high-skilled native workers (Borjas, 1994).  Using this theoretical model to interpret the results means that when the share of immigrants increases, the native born workers who are considered high-skilled are given more job opportunities and wage increases as a result of immigrants entering the work force. The end result is native workers earn a higher wage. In the short-run the Canadian labour market does not have a chance to adjust for the immigration supply shock which is illustrated by the fact that wages for the low-skilled workers are unchanged. The R² value is significant and the same as in equation (3).

The results from the regression in table 5 imply that an increase in the share of immigrants in Canada results in wage increase for the high-skilled native workers. This means that immigrants and high-skilled native workers in Canada are complements and not substitutes. As mentioned above this is consistent with Borjas (1994) theory and other previous studies such as Kugler and Yusel (2008) and Fogel and Peri (2014) where the result is similar; the effect on native worker wages is positive.  However, Bratsberg et. al (2010) finds a negative relationship between an increase in the share of immigrants and native worker wages. Their study differs in that they define skill level by two terms, education level as well as work experience. This may be a reason for the differing results from the regression in this study. Also the countries in which the study is conducted are also different which may be another reason the results are opposite.

There is however a problem when analyzing the results of the regression. It is possible that immigrants in the sample are high-skilled and therefore would be considered substitutes to the high-skilled native workers, which would violate the theory in Borjas (1994). A explanation to this problem could be that immigration is an action of economic incentive, immigrants migrate for reasons such as achieving higher levels of utility than they achieved back home this may cause them to accept jobs there are overqualified for. The wage of a low-skilled job in Canada may be higher than the wage of the high-skilled job back in the immigrants’ home country. Also, another explanation when considering the short-run is that an immigrant does not have the opportunity to convert the foreign human capital into the Canadian labour market thus making the immigrant a complement to a native worker with the same amount of skill. It is not entirely clear to what degree immigrants and native workers are substitutes or complements. When the long-run impact is being studied, there is a possibility than an immigrant can become a substitute for high-skilled native workers as new skills are acquired and improved over time.

A limitation of the short-run analysis in this study is that with the data and regression only average effects are considered. Workers in the Canadian labour market do not have standardized labour. Workers in a labour market offer different unique sets of skills so a labour market for a professor is not the same as a labour market for a doctor. When analyzing the average effect the study is not able to capture the different effects across different labour markets.

Medium-Long Run

The assumption made when analyzing the outcome in the medium-long run is that capital is not fixed, whereas in the short run capital was fixed. This means that firms can now invest money in more capital and raise capital stock and thereby increase the demand for labour. The regression from equation (1) determined whether the share of immigrants had an impact on native worker wages regardless of the skill level within a municipality. The table below presents the results of the regression:

Table 6. Medium-Long run results from equation (1)[10]

Variables	Natives’ wages
Share Immigrants	334.04*** (20.26)
Constant	306.50*** (2.54)
Area fixed effect	No
Year fixed effect	No
Observations	1198
R-squared	0.175
Standard error in parentheses	***p<0.01,**p<0.05,*p<0.1

The outcome shows a positive and significantly positive share of immigrants of 334.04 at a 1% significance level. This implies that a 1% increase in the share of immigrants increases native worker wages by $334.04 Canadian dollars. R² which determines how much of the variance in the dependent variable is explained by the independent variable is only 17.5%. [11]The low R² suggests that this perhaps is not the ideal regression to use (Gujarati & Potter, 2009). As this first regression is broad and does not divide the population into skill groups and has limited variables the low R² value does not have much significance. When the native worker population is divided into skill groups according to their respective education levels, the results are vastly different than in the short-run. The tables below present the results.

Table 7.  Medium-Long run results from equation (2) [12]

Variables	Natives’ wages
Share Immigrants	228.40*** (52.03)
High Education	427.08***(28.22)
Share Immigrants effect on high skilled natives	-140.31***(188.1)
Constant	218.29*** (7.06)
Area fixed effect	No
Year fixed effect	No
Observations	1,198
R-squared	0.573
Standard error in parentheses	***p<0.01,**p<0.05,*p<0.1

Table 8. Medium-Long run results from equation (3)[13]

Variables	Natives’ wages
Share Immigrants	-344.15*** (52.42)
High Education	-14.93***(49.18)
Share Immigrants effect on high skilled natives	1,041.33***(151.27)
Constant	401.87*** (16.46)
Area fixed effect	Yes
Year fixed effect	Yes
Observations	1,198
R-squared	0.996
Standard error in parentheses	***p<0.01,**p<0.05,*p<0.1

 

The regression from equation (3) is presented in Table 7 which shows that the effect of immigration on high-skilled native worker wages is -140.31 this means that a 1% increase in the share of immigrants decreases high-skilled native worker wages by $140.31 Canadian dollars, again this outcome is not substantial. However, as estimated the effect of immigration on native wages is higher when investigating different skill groups as seen by the increase in R². The increase in R² means that more of the variance in native worker wages is explained by the changes in share of immigrants. Table 8 shows the regression from equation (3) which includes the fixed effects. In this case the result for high-skilled native worker wages is vastly different from the previous regressions. The results indicate that a 1% increase in the share of immigrants increases high-skilled native worker wages by $1,041.33 Canadian dollars. The fixed effects added to the regression which are the area fixed effect and the year fixed effect result in a higher R².[14] This implies that 99.6% of the variance in high-skilled native worker wages is explained by the share of immigrants. The standard deviation of 151.27  is at a reasonable level. The final regression is run with the same two control variables as before, age and unemployment.

Table 9. Medium-Long run results from equation (4)[15]

Variables	Natives’ wages
Share Immigrants	-350.22(53.82)
High Education	-29.14***(49.12)
Share Immigrants effect on high skilled natives	957.31***(152.49)
Age	-3.32***(0.99)
Unemployment	5.97(18.8)
Constant	536.92*** (43.86)
Area fixed effect	Yes
Year fixed effect	Yes
Observations	1198
R-squared	0.996
Standard error in parentheses	***p<0.01,**p<0.05,*p<0.1

 

The share of immigrants’ effect on high skilled natives is still large as seen from the results from equation (4). At a 1% significance level it has a value of 957.31 which implies that a 1% increase in the share of immigrants will result in high-skilled native worker wages increasing by $957.31 Canadian dollars. The two control variables in this regression have opposite signs, the age parameter is negative whereas the unemployment parameter is positive. The unemployment parameter is larger than the negative parameter of age therefore the previous results are underestimated. As in the previous regression in equation (3) the R²is a significantly large value and the exact same value.

The regression from equation (4), the results of which are presented in table 9 imply that an increase in the labour supply positively affects native worker wages who are considered high-skilled. When assuming that immigrants and natives are perfect substitutes this result is not consistent with the theory. The assumption made in the medium-long run analysis is that immigrants are low-skilled and the concentration is the effect on high –skilled native wages. [16]The results when considering this implies that immigrants and native-workers are complements. The outcome asper the theory and the outcome from the regression are consistent. The conclusion is that even in the medium-long run, immigrants and natives are complements.

The results in this study are similar to the ones in Brucker et. Al (2014) which studied the impact of immigration on native worker wages using three different skill levels in three different countries. In the study its mentioned that the differences in the results may be due to differences in policies and labour markets in the three countries. Likewise, the results in this study may be affected on the labour market conditions and policies in Canada. The most significant factor may be the labour market conditions in Canada as some workers are in unions. This prevents firms from paying a wage they want to in some cases, if there were no unions the results may have been different. Previous studies such as (Grossman 1982 and Borjas & Tienda 1987) find a small negative effect when studying the effect of immigration on low-skilled native workers. The results for the same regression are presented in the appendix of this study.

In accordance to theory, when a supply shock increases the return to capital, firms will react by investing more capital as the return to capital is high. The demand curve reacts by shifting back and the end result is that there will be no effect on native worker wages. The results in this study are not consistent with the theory. The results show that the impact of low-skilled immigrants on high-skilled native worker wages increases sharply from the short-run to the medium-long run. This may be because the period of medium-long run which is 5 years is not a long enough of a period to study long run effects and thus the outcome in the medium-long run is unanticipated. As mentioned before, immigration is a flow and not a stock event and the short-run is the time it takes for businesses to adjust their investment levels to the increased labour supply (Somervill and Sumption, 2009).  This can mean that when investigating the outcomes in the medium-long run they may in fact still be part of the ‘short-run’. Perhaps, five years is a better time period to study the effects of increased share of immigrants on native worker wages. The additional time allows for native workers to possibly be promoted or take on higher paying jobs with the increased level of experience. When considering the effect of the labour supply shock on low skilled native workers the parameter gets more negative in the medium-long run from the short-run. Also it takes time for immigrants to integrate into a new country so the time period chosen can also give unexpected results. There are certain barriers to entry new immigrants may face in a new labour market such as language barriers this also effects how much time it takes to see the expected outcome.

For the purposes of this study the analysis is done on a short and medium-long run basis without taking the above factors into consideration. One explanation why the outcome is not in line with the economic theory is that immigrants are considered low skilled. However, this is not always the case in the real world, immigrants come with a range of education levels and skill sets. It can be assumed that these immigrants add to the human capital stock in the labour market and help strengthen firm’s productivity and as a result help increase the native worker wages in the medium-long run which is driven by the increased productivity. Wage growth is driven by productivity in the long-run which might help explain the increased impact of the share of immigrants over time.

Immigration affects certain native workers more than others. Thus it is possible that natives who are most affected, this case low-skilled and medium skilled workers, migrate to different municipalities where they earn higher wages. Like immigration affects the labour supply cure, migration by natives within a country will affect the labour supply curve as well. In this study, low -skilled and medium-skilled native workers are most affected by immigration. If they decide to migrate within Canada they will affect the wages of all native workers irrespective of their skill level, similarly to the way immigrants do (Borjas et. Al., 1996). There are no barriers to migrate within Canada so therefore this is another factor that may affect the outcome.

Summary & Conclusion

The regressions conducted in this study have determined the effect that immigration has on native worker wages. More specifically, how the wages for native workers who are in different skill levels respond to changes in the share of immigrants. The outcomes using equation (4) which is considered the ‘best’ regression in this study produces results that are consistent with theory in the short-run. For low-skilled and medium-skilled workers it is the opposite case, the result of immigration on native worker wages for these two skill levels is negative. The results in medium-long run are not consistent with what theory states. Native workers in the high-skilled level see an increase in their wages with an increase in the share of immigrants whereas, low and medium-skilled workers see a decrease in their wages. The factors that cause this outcome are many. Different labour markets have different policies and conditions that may play a role in affecting the outcome and thus the theory may not hold in every labour marke

Taking the results of this study into context, if Canada were to see an influx of immigration, highly educated native workers would expect to benefit from this while the lower educated native workers would expect a negative result. The results seen in this study may explain why some people are opposed to immigration as it negatively impacts citizens of the host country. However, it also positively impacts the more educated individuals within a country in terms of wage.

 

^[1] http://www.un.org/en/development/desa/news/population/number-of-international-migrants-rises.html

^[2] https://www.oecd.org/migration/OECD%20Migration%20Policy%20Debates%20Numero%202.pdf

[3] Statistics collected from Statistics Canada

[4] Source: Statistics Canada and calculations

[5] Source: Statistics Canada and calculations

[6] Appendix table 3a & 3b for low & medium skilled results

[7] Appendix table 4a and 4b for low & medium skilled results

[8] Source: Statistics Canada and calculations

[9] Appendix table 5a and 5b for low & medium skilled results

[10] Source: Statistics Canada  and calculations

[11] Appendix table 7a and 7b for low & medium skilled results

[12] Source: Statistics Canada and calculations

[13] Source: Statistics Canada and calculations

[14] Appendix table 8a and 8b for low & medium skilled results

[15]Source: Statistics Canada

[16] Appendix table 9a and 9b for low & medium skilled results