University of Amsterdam
How does immigration affect the wages of natives?
Evidence from Norway and Sweden
Abstract: This paper investigates whether immigration affects local wages by looking at the impact of Bosnian refugees on the earnings of natives in Norway and Sweden from 1993 till 1997. Norway and Sweden, similar in many economic aspects, took large number of Bosnian refugees when the Bosnian war broke out in 1992. This paper examines the impact of the number of Bosnian refugees on earnings of native people from different occupations for Sweden and Norway. The result shows that although the intake level of Bosnian refugees of Sweden is larger than the intake level of Norway, the effect of immigration on native people’s earnings is almost the same for both countries. Therefore, the conclusion of the research is that immigration does not affect local people’s wages.
Yiyan Li
Student number: 10227288
Table of content
1 Introduction ... 3
2 Literature review ... 4
2.1) A summary of previous economic studies ... 4
2.2) Conventional economic theories for the impact of immigration on the host country
and their limitations ... 5
2.2.1) Closed economy model: effect of immigration depends on the relationship
between local labor and foreign labor ... 5
2.2.2) the Heckscher-Ohlin model: effect of immigration depends on the size of
immigration ... 6
2.3) A common difficulty in the research: reverse causality ... 6
2.4) The solution to the difficulty: focus on exogenous immigration shock ... 7
3 Methodology and data ... 8
3.1) Data ... 8
3.2) Methodology ... 13
4 Results and analysis ... 14
4.1) Descriptive statistics ... 14
4.2) Discussion of results and analysis ... 17
5 Conclusions ... 19
6 Limitations ... 19
Appendix ... 21
Bibliography ... 22
1 Introduction
Immigration is a controversial issue in the industrialized nations of the world. Several studies have been conducted on the impact of immigration on the host countries labor market. Most of them concern traditional immigration countries, such as the United States, Canada and Australia (Borjas, 1994, p.201). In Europe, scholars have focused on the position of immigrants themselves and they pay most attention to the poor position of low skilled immigrants from Mediterranean countries who were attracted to the booming European economies (Hartog & Zorlu, 2002, p.114). Many studies have focused on the United Kingdom and Germany but some research has also been conducted on the French, Dutch, Swedish and Norwegian labor markets.
The findings reported by the various researches have been inconsistent. Many studies have concluded that immigration has a negative influence on local people’s wages while others have found that immigration has a positive effect or almost no effect. A common problem for these studies is that their research methods suffer a great deal from the problem of reverse causality, namely, the high level of wages can attract more immigrants while immigrants might affect local people’s wages. A good way to solve this econometric problem is to focus on a certain time period when people are forced to immigrate into other countries not because they are attracted to high local wages. An example of this is the Bosnian war, which broke out in 1992 and ended in 1995 (Kondylis, 2009, p.235). United Nations Refugee Agency recorded the number of Bosnian refugees taken by each country. Both Sweden and Norway admitted a large number of Bosnian refugees from 1993 till 1997. However, Sweden experienced much larger Bosnian immigration shock. Since Sweden and Norway have a lot of economic and geographic similarities (Østbye &Westerlund, 2010, p.50), this research will specifically examine the influence of Bosnian refugees on Swedish and Norwegian labor markets and compare the differences of the impact. If the results show that the influence of Bosnian immigration shock on native Swedish and Norwegian’s wages is significantly different, then the conclusion is that
immigration has impact on local people’s wages. Otherwise, it can be concluded that the effect of immigration has no effects on local wages.
The structure of the following text is as follows. Section 2 briefly summarizes previous studies regarding the influence of the immigrants on the host country and the economic theory concerning the impact of immigration. It will also point out the shortcoming of some of previous scholars’
research methods and an approach to improve on them. Section 3 contains the description of the data and the model used for the empirical analysis. Section 4 moves on to the results of the
econometric analysis. Section 5 gives the conclusion of the research and section 6 elaborates on the limitations of this study.
2 Literature review
2.1) A summary of previous economic studies
Many scholars have found a negative relationship between immigration and local people’s wages. For example, Goldin (1994) exploited U.S. cross-city variation by using several sources of data from 1890 to 1923. She matched the city-level wage data with information on immigrant density, which was taken from the nearest Censuses of Population. Then she studied the effect of changes in the number of foreign-born as a fraction of city population on changes in wages for various industries and occupations. The most common result was that a 1 percentage point increase in the fraction of the foreign-born population reduced local wages by 1.0–1.6 percent, which was quite substantial.
Many studies suggest that immigration has almost no influence on local wages. Manacorda, Wadsworth and Manning (2012, p.129) conducted a research in the UK. They presented a stylized model of labor demand disaggregated by skill and age which allowed them to estimate the impact of changes in the supply of educated labor on the wage structure from the mid-1970s to the mid-2000s. Although immigration has primarily reduced the wages of immigrants, particularly of university educated immigrants, it has little discernable effect on the wages of the native-born people. For some skill groups, one percent increase of immigration reduced native people’s wage by a maximum 0.89 percent, while for other groups the effect is almost zero.
Some studies have found positive relationships. Addison and Worswick (2002) studied the influence of immigration on native people’s wage in Australia from 1982 till 1996. They divided each of Australia’s six states into eight broad occupational labor markets and regressed the mean of native’s weekly earning on the proportion of recent immigrants in each labor market. Results showed that 10 percent increase in immigrant’s presence would lead to a 1.21 percent increase in wage levels. The conclusion was that immigrants had a positive impact on the real wages of native Australians. The
conclusion remained unchanged when the specific influence of immigration on less educated or young Australian workers was investigated.
The studies listed above found are typical examples. People found positive, negative or no relationships between the number of immigrants and local wages. Table 1 in the Appendix briefly summarizes more studies on this issue and their results. However, some of these studies suffer from econometric problems, which will be discussed in the third section of the literature review part.
2.2) Conventional economic theories for the impact of immigration on the host country and their limitations
2.2.1) Closed economy model: effect of immigration depends on the relationship between local labor and foreign labor
Rachel and Hunt (1995, p.6) consider a closed economy where production takes place using three factors- capital, skilled labor and unskilled labor. Capital and skilled labor are complements. Unskilled labor and capital are imperfect substitutes and unskilled labor and skilled labor are also imperfect substitutes. If two factors are perfect substitutes with each other, the increase in the amount in one of them will lower the price of the other. If two factors are imperfect substitutes with each other, the increase in the amount in one of them will have an ambiguous impact on the price of the other one, namely the price of the other factor can be increased or decreased. If two factors are complements with each other, the increase in the amount of one of them will raise the price of the other factor.
Following the above principle, if immigration of unskilled workers occurs, the wage of unskilled workers will fall, and the effect on the return to capital and the skilled wage can be positive or negative. If the immigrants are skilled, they will lower the skilled wage and the impact on the
unskilled wage can be positive or negative. However, the fall in the skilled wage and the rise in skilled employment will lead to increased demand for the complementary factor, capital, and hence an increase in the return to capital.
Although in the closed-economy model, the factor of capital is taken into account, since this 5
research is about the labor market, the analysis should focus on the impact of immigration on the wage of skilled labor and unskilled labor. The straightforward message from the above simple analysis is: skilled immigration could benefit unskilled native workers and might benefit or hurt skilled native workers, and unskilled immigration hurts unskilled and may benefit or hurt skilled native labor (Kahanec, M., Zaiceva A. & Klaus F. Z., 2009, p.5). This theory explains the diversity of the research findings. The type of immigrating labor and its relationship with local labor can determine if the research will find a positive effect or a negative effect.
However, this theory also has some flaws. According to Bonin, the model does not take into account broader positive effects of immigration. Knowledge and skills transfer or improved allocations of production factors (mainly human capital) which result from technological progress should also be viewed as factors in production on top of unskilled labor, skilled labor and capital (2009, p.9).
2.2.2) the Heckscher-Ohlin model: effect of immigration depends on the size of immigration
As Rachel and Hunt (1995) pointed out, the Heckscher-Ohlin model also helps to determine the effect of immigration on local people. It assumes that countries have different endowments of factors of production and they will specialize in producing certain types of goods, instead of each producing all goods. Thus, countries with a large labor endowment will produce a more labor-intensive mix of goods than countries with a large capital endowment. And the resulting cross-country differences in wages could then generate migration. In this case, the impact of a labor influx will then depend upon its size: a large enough inflow will force the country to produce a more labor-intensive mix of products, which will lower the wage (and increase the return to capital). If the inflow is small, immigration will not affect wages, as the country will increase production of its relatively labor-intensive goods and sell more of those goods on the world market, and thus factor price equalization will be achieved through trade.
2.3) A common difficulty in the research: reverse causality
According to Addison and Worswick (2005, p. 5), a common model for studies in section 2.1 is: 6
Wjt=βXjt+ γIMMjt+ εjt
where Wjt takes the log of the mean value of weekly earnings in labor market j in year t, Xjt is a vector
of native labor market’s characteristics, IMMjt is the proportion of recent immigrants in each labor
market and εjt is the residual term. The parameter γ measures the impact of immigrants on the mean
wages in the labor market.
One of the most serious problems that these studies face is reverse causality. People leave their hometown and settle in other countries where they expect to earn higher wages. The high expected wages arise from the great potential of development of their destination countries. However, the value of expected wages is hard to estimate. Suppose in period t immigration starts. We regress the actual local wages in period in t, t+1, t+2, t+3… on immigrants share. Because of the great
development of local economy, local wages will rise a lot after period t. Due to the reason that local labor market takes more and more immigrants, the conclusion of the research is that immigration has positive impact on local wages, but this is wrong. The problem is that the local wage in period t+1 is supposed to be caused by the immigrant inflow in t. However, the local wage in t+1 and the immigrant inflow in t are strongly correlated by the expected wage in t for t+1. So the variance in local wages could be explained by the variance in immigration and also vice versa. Therefore, the research method suffers from reverse causality.
What we should do is to regress the expected wages in the following periods t, t+1, t+2, t+3 on the immigration share but not the actual local wages. If expected wages of immigrants can be obtained, the real impact of immigration on local people's wage can be measured. Focusing on exogenous immigration shock is a good strategy to solve the problem that values of expected wages are hard to obtain.
2.4) The solution to the difficulty: focus on exogenous immigration shock
As explained in section 2.3, since the expected wage is hard to obtain, it is better to shift the focus of study to certain historical events when people leave their home country not because they are attracted to the high expected wages. The idea of this approach is to focus on exogenous immigration shock when different numbers of immigrants hit different labor markets and then measure the effect on local wages (Stock & Watson, 2012, p.530). According to Bonin, it is a good way to solve the
problem of reverse causality (2005, p.2), since it leaves out the factor that people immigrate to foreign countries because they expect to earn higher wages there. A typical study is done by
Grossmann (1982, p. 596). It analyzed the impact of the Mariel boatlift from Cuba to the metropolitan area of Miami. The event was precipitated by a sharp downturn in the Cuban economy which later led to internal tensions on the island and a bid by up to 10,000 Cubans to gain asylum in
the Peruvian embassy. Although this exogenous shock raised the local workforce dramatically, Card did not find evidence for a significant adverse effect on natives’ employment and wages (1990, p. 257).
The advantage of this method is that data of immigration and local wages is easy to gather. Therefore, this method will be applied in the following research.
3 Methodology and data 3.1) Data
To solve the reverse causality problem, this paper examines the impact of immigration shock- Bosnian war on the labor market of Sweden and Norway. The Bosnian War was an
international armed conflict. It took place in Bosnia and Herzegovina between 6 April 1992 and 14 December 1995 (Sumantra, 2009, p.124). As recorded by Jefferson (1999), the war came about as a result of the breakup of Yugoslavia. The Slovenians and Croatians withdrew from the Socialist Federal Republic of Yugoslavia in 1991. Following the secessions, the multi-ethnic Socialist Republic of Bosnia and Herzegovina, inhabited by Muslim Bosniaks (44 percent), Orthodox Serbs (31 percent)
and Catholic Croats (17 percent), passed a referendum of independence on 29 February 1992. However, the result was rejected by the political representatives of the Bosnian Serbs, who had established their own territory. The Bosnian Serbs, supported by the Serbian government of Slobodan Milošević and the Yugoslav People's Army (JNA), put armed forces inside the Republic of Bosnia and Herzegovina to secure Serbian territory. Then war soon broke out across the country, accompanied by the ethnic cleansing of the Muslim Bosniak and the Croat population.
The Bosnian refugees were taken by many European countries, such as Norway and Sweden (Birgitta, Anders & David, 2001, p.1).The reason that Norway and Sweden are chosen as study
subjects is that these two countries have many similarities. They are located close to each other and their languages are alike. They formed a union between 1814 and 1905 (Holmøyvik, 2005, p.135). Although the union dissolved in the 1905 (Holmøyvik, 2005, p.135), Norway and Sweden would still have many similarities in economic and political aspects. In the following research, factors such as inflation, exchange rate against dollar and world GDP growth are controlled. Therefore, if there is any difference in wage changes of these two countries, the difference could be explained by the different intake level of Bosnian refugees to a large extent.
First of all, the intake level of Bosnian refugees will be calculated. The intake level is defined as the number of Bosnian refugees divided by the total population of the country in that year. The number of Bosnian refugees is extracted from the report of United Nations refugee agency1and the
total population is picked out from the website of World Bank data2. Then the intake level can be
determined. The following tables summarize the intake level of Bosnian refugees for both Norway and Sweden.
Table 1 intake level for Norway
Year 1993 1994 1995 1996 1997
Number of Bosnian refugees 25110 2649 1059 262 742
Total population 8718600 8780700 8826900 884100 8846100 Intake level 0.00288001 0.0030168 0.0001110 0.0000296 0.00008389
1 United Nations Refugee Agency was established on December 14, 1950 by the United Nations General Assembly. Its primary purpose is to safeguard the rights and well-being of refugees. It also keeps record of the number and nationality of refugees each country takes. Here the immigration shock is interpreted as the number of asylum seekers each year. The report can be found in each country’s report on the website of United Nations Refugee Agency http://www.unhcr.org/
2 See World Bank data http://data.worldbank.org/
Table 2 intake level for Sweden
To see the relationship between the intake level of refugees of Norway and Sweden, the intake level of Sweden is divided by the intake level of Norway of each year. The following chart summarizes the results from calculation.
Table 3 the relationship between Norway and Bosnian refugee intake level
The conclusion from the Table 3 is that Sweden always has much higher intake level of Bosnian refugees. The next step is to get the wages for each country for year 1993 till 1997. The reason why this paper examines the impact of Bosnian on local wages after year 1993 but not before is that after the war broke out in 1992, Bosnian refugees gradually settled in Norway and Sweden and started to influence local wages. The reason why this research paper does not look into years after 1997 is that the local wages from 1998 on for Sweden in Swedish yearbook are summarized in a different way compared to how they were summarized before 1997. The wages of two low skilled industries hotel and construction workers (refers to unskilled workers) and two highly skilled industries banking and insurance are chosen. The reason for choosing various industries is that data of average wage of both countries are hard to obtain. The only available information is the average wage for each industry.
Year 1993 1994 1995 1996 1997
Number of Bosnian refugees 7051 201 106 73 90
Total population 4312000 4336600 4359200 4381300 4405200 Intake level 0.0016352 0.0000464 0.0000243 0.0000167 0.0000204
Year 1993 1994 1995 1996 1997
Results 1.7612 6.5088 4.9339 1.7786 4.1055
Also, testing more industries can make the conclusion more persuasive. The wages for Norway and Sweden are taken from the yearbook 1.
Table 4 the wage for each industry in Norway (in dollars)
Year 1993 1994 1995 1996 1997
Unskilled private construction workers2 12.65 14.06 15.61 15.83 14.59
Hotels and restaurants2 1772.67 2009.84 2204.75 2281.29 2089.62
Banks2 2446.01 2820.12 3162.66 3247.36 2997.95
Insurance2 2797.61 3284.17 3571.20 3677.95 3407.92
Table 5 the wage for each industry in Sweden (in dollars)
Year 1993 1994 1995 1996 1997
Unskilled private construction workers2 11.14 10.89 12.26 13.64 12.33 Hotels and restaurants2 2033.78 2134.11 2429.31 2816.72 2388.74
Banks2 2496.86 2561.35 2920.41 3312.99 2964.66
Insurance2 2304.10 2464.92 3052.52 3414.33 3147.91
1 The information for wages of Norway can be found on the Norwegian yearbook published on www.ssb.no. The information for wages of Sweden can be found on the Swedish yearbook published on http://www.scb.se/en/. 2 The wages reported on both Norwegian and Swedish yearbooks are categorized into male and female. Since gender is not a factor of consideration here, the wages summarized in the tables above are the average of male and female earnings. The wages for unskilled construction workers are hourly wages for both countries, the wage for remaining four industries are monthly wages.
Besides immigration, there are other factors that might affect local wages. Examples are the strength of local currencies against dollar (namely the exchange rate of the local currency to dollar), world GDP growth and local inflation rate. The following table summarizes the information about these variables. The exchange rate, world GDP and inflation rate for both Norway and Sweden can be found on the website of World Bank data1.
Table 6 exchange rate of local currency to dollar
Year 1993 1994 1995 1996 1997
Exchange rate (Norway Kroner to dollar) 7.52 6.76 6.32 6.44 7.32 Exchange rate (Swedish Kroner to dollar) 7.8 7.71 7.13 6.7 7.64
Table 7 World GDP (unit: in thousands of billions of dollars)
Year 1993 1994 1995 1996 1997
World GDP 2.58662 2.76105 3.04545 3.10798 3.0989
Table 8 inflation rate for both Norway and Sweden
Year 1993 1994 1995 1996 1997
Inflation rate in Norway 2.27% 1.40% 2.46% 1.26% 2.58% Inflation rate in Sweden 4.65% 2.20% 2.53% 0.47% 0.52%
1 See World Bank data http://data.worldbank.org/
3.2) Methodology
The model used in this study is based on a study by Addison and Worswick (see section 2.3): Wjtc= αjc+ βjcIMMjtc + γtcEXCtc +θtcINFtc+ λtcGDPt + εjc
( j=1 represents construction industry; j=2 represents industry of hotels and restaurants; j=3 represents banking industry; j=4 represents insurance industry; c=1 represents Norway; c=2, represents Sweden)
In the regression model, Wjtc takes the log of the average value of wages in labor market j in year
t of country c, αjc is the constant term, IMMjtc is the proportion of recent immigrants in each labor
market, namely the intake level calculated in Table 1 and Table 2, and εjc is the residual term. The
parameter βjc measures the impact of immigrants on the mean wages in the labor market. EXCtc is the
exchange rate of local currency of country c in year t. INFtc is the inflation rate of country c in year t
and GDPt represents the global GDP in year t. γtc , θtc and λtc measures the impact of these factors on
wages respectively.
The reason of taking natural logarithm is that a change in one unit of Bosnian immigration share is associated with a β% change in the local wage. Therefore, β measures the change of local wage with regard to 1% change of immigration share (Stock& Watson, 2012, p.314).
Firstly, the control variables of exchange rate, world GDP and inflation are left out. The wage of each industry is regressed against the intake level of the country and then the comparison between the coefficients β of the same industry for these two countries is made. t-test is applied to test if the coefficients β for Norway is different from the coefficients β for Sweden. If the coefficient β does not differ much from that of the other country, the conclusion is that immigration does not affect local people’s wages. The reason is that although the intake level of Sweden and Norway are different, the influence on local wages (measured by β) for Sweden does not differ much from the influence for Norway. If the difference is very huge, then the conclusion is that immigration has impacts on local wages. In the following steps, each time one or two or three variables out of inflation rate, exchange rate against dollar, world GDP are put in the regression model. The value of β is again compared between two countries.
4 Results and analysis 4.1) Descriptive statistics
First of all, all the control variables are left out. The dependent variable is the wage of each industry, and the independent variable is the intake level of Norway and Sweden. The following tables summarize the regression results.
Table 9 regression results for Norway
Coefficients Standard error t value p-value Construction industry -106.85 38.17 -2.80 0.068 hotel and restaurants -119.05 38.18 -3.12 0.053 Banking -138.75 42.28 -3.28 0.046 Insurance -136.83 33.79 -4.05 0.027
Table 10 regression results for Sweden
Coefficients Standard error t value p-value Construction industry -39.28 35.65 -1.10 0.351 hotel and restaurants -70.39 43.34 -1.62 0.203
Banking -63.48 39.97 -1.59 0.210
Insurance -103.38 51.62 2.00 0.139
In the next steps, control variables are included. The following tables summarize the value of β for different industries for each country. In model 1, exchange rate is included. Model 2 includes the inflation rate. Model 3 includes the world GDP. Model 4 includes exchange rate and inflation. Model 5 includes exchange rate and world GDP. Model 6 includes inflation and world GDP. The first line in each grid summarizes the value of the coefficient β and the number in the brackets is the value of the standard error for the coefficient β. p-value in the table means the significance level for the coefficient
of IMMjtc. The reason why p-value of IMMjtc is important is that this paper examines the impact of
immigration on local economies. The smaller the p-value is, the more persuasive the value of β is and the more precise the impact is measured. Model 7 is omitted because the regression does not show the standard value of β in the results overview. The reason might be that the relationship between wages, intake level and all three control variables may not be linear and the relationship cannot be described by the regression model presented in section 3.2.
Table 11 Regression results for β of construction industry
Regression models
(1) (2) (3) (4) (5) (6)
Results for Norway
β -65.39 (47.58) -106.95 (48.21) -41.81 (54.32) -63.41 (62.60) 2.12 (7.97) -21.53 (76.46) p-value 0.30 0.16 0.52 0.50 0.835 0.83
Results for Sweden
β -6.94 (28.36) 22.44 (70.67) 45.07 (39.24) 38.09 (26.18) 38.09 (26.18) 54.79 (60.90) p-value 0.83 0.78 0.37 0.38 0.38 0.53 15
Table 12 Regression results for β of hotel and restaurant industry
Table 13 Regression results for β of banking industry Regression models
(1) (2) (3) (4) (5) (6)
Results for Norway
β -80.69 (50.30) -117.59 (48.03) -52.32 (52.73) -79.57 (69.92) -11.58 (26.47) -25.28 (63.29) p-value 0.25 0.13 0.43 0.46 0.74 0.76
Results for Sweden
β -27.63 (26.25) -2.22 (89.67) 20.39 (62.31) 20.43 (23.20) 8.89 (35.46) 32.12 (98.78) p-value 0.40 0.98 0.78 0.54 0.84 0.80 Regression models (1) (2) (3) (4) (5) (6)
Results for Norway
β -101.45 (59.54) -139.74 (53.32) -57.28 (50.02) -99.25 (79.67) -17.39 (15.33) -35.37 (66.20) p-value 0.23 0.12 0.371 0.431 0.460 0.69
Results for Sweden
β -27.93 (33.19) 9.22 (77.10) 31.22 (43.78) 27.51 (42.40) 23.78 (33.33) 44.61 (66.09) p-value 0.49 0.92 0.550 0.63 0.61 0.62 16
Table 14 Regression results for β of insurance industry
4.2) Discussion of results and analysis
First of all, all the control variables are left out and the pure impact of immigration on local wages is analyzed. Regression results presented in Table 9 and Table 10 are examined firstly.
t-test is applied to examine if the coefficient of the two countries are significantly different from each other. The formula for t statistic is:
tβtest= (βtest-β0)/s.e.(βtest)
where β0 is a non-random, known constant. Here, it refers to the value of β for each industry of
Sweden. s.e.(βtest) is the standard error of the estimator βtest . βtest is the parameter for each industry
of Norway. The significance level α is chosen as 5%.
Applying the results obtained from Table 9 and Table 10 to the formula of t statistic, the t-value for four industries can be calculated. The t statistics are listed in Table 11.
Regression models
(1) (2) (3) (4) (5) (6)
Results for Norway
β -103.35 (44.89) -135.79 (42.57) -76.99 (45.89) -102.27 (62.17) -41.41 (22.19) -53.78 (55.67) p-value 0.15 0.09 0.24 0.35 0.31 0.51
Results for Sweden
β -63.44 (52.41) -14.35 (102.54) 31.52 (30.32) 5.96 (93.82) 26.26 (21.94) 40.08 (46.32) p-value 0.35 0.90 0.41 0.960 0.12 0.55 17
Table 15 t-value of each industry
The null hypothesis is H0: βj1=βj2 and the alternative hypothesis is H1: βj1≠ βj2. Since this is a
two-tailed test, to get the critical region, the value of tα/2 should be calculated: t0.025=1.96. If t<-1.96 or
t>1.96, then the null hypothesis should be rejected. If this is not the case, then H0 should not be
rejected.
In this case, since all the values of t do not fall into the critical region, the null hypothesis should not be rejected. Therefore, there is not enough evidence to conclude that immigration has impact on local wages.
Next, control variables are included to see if the conclusion is still the same. As recorded in Table 11, 12, 13 and 14, almost all the p-values are far much larger than the pre-set significance level of 5%. p-value, also called the significance probability, is the probability of observing a value of β which is as different from the estimate which is calculated (Stock & Watson, 2012, p.189). The large p-value indicates that when these control variables are added, the value of β that is generated from regression is not trustworthy. A possible reason for this is that there is hardly any effect of immigrants on the wages of natives for both Norway and Sweden in the short run (this study only looks at a time period of five years). Another possible reason is that the relationship between wages, intake level and control variables cannot be described by the model presented in section 3.2. However, the discussion of which model fits the relationship between the wages, intake level and control variables is beyond the scope of this paper.
Due to the large p-value recorded in Table 11, 12, 13 and 14, it is not wise to conduct t-test for regression models which take into account control variables, because values of β generated in Table 11, 12, 13 and 14 are not reliable. However, the large p-value confirms the conclusion that there is
t-value Unskilled private construction workers -1.77
Hotels and restaurants -1.27
Banks -1.78
Insurance -0.99
hardly any effect of immigrants on the wages of natives for both Norway and Sweden, at least in the short run. Therefore, the conclusion that the Bosnian refugees do not influence local wages seems robust.
5 Conclusions
This paper examines the impact of immigration on local wages by looking at a certain historical event that Bosnian refugees immigrated into Sweden and Norway because of Bosnian war. The conclusion of this research is that even though the intake level for Bosnian refugees differs across the two countries from year 1993 till year 1997, the change in the wage level for both countries does not differ a lot. Therefore, immigration does not affect local wages.
The finding for this paper can be explained by both closed economy model and the Heckscher-Ohlin model. According to closed economy model, skilled immigration could benefit unskilled native workers and might benefit or hurt skilled native workers, and unskilled immigration hurts unskilled and may benefit or hurt skilled native labor. For industries such as construction and hotel which require low skills, highly-skilled Bosnian immigrants will benefit native low-skilled workers and low-skilled Bosnian immigrants might hurt low-skilled workers. These two impacts offset each other and the local wages for low skilled industries do not change much. For industries such as banking and insurance which require high skills, highly-skilled Bosnian immigrants could hurt native high-skilled workers and low-skilled Bosnian immigrants might benefit low-skilled workers. These two impacts offset each other and the local wages for low skilled industries do not change much.
According to the Heckscher-Ohlin model, if the inflow is small, immigration will not affect wages. Both Table 1 and Table 2 show that the Bosnian immigration shock is not that huge compared to total population. Therefore, Bosnian immigration does not influence local wages.
6 Limitations
This research has several drawbacks. First of all, the data of the wage level of Norway and Sweden are not easy to gather. The main source of wage level is the yearbook. In 1998, the wage summary of Norway statistics yearbook is changed and the relevant information about unskilled construction workers is lost. Ideally, the research should include at least ten years length. However,
because of the change of yearbook, the research fails to include the year 1998 and from then on. Secondly, although Norway and Sweden have many similarities, they are different in many aspects. A better approach is better done at city level. For example, if the data for the number of Bosnian refugees of each city within Sweden is available, the same research can be conducted in Sweden using the same method. Thirdly, according to Bonin (2005, p.6), if capital or native workers respond to the immigration shock by relocating, the effects of a local supply shock from international migration could spread beyond the locality. For example, native workers can react to a deterioration of local
employment opportunities by moving to other labor markets. In this case, the measured impact of the immigrant shock will be rather small since immigration eventually affects all local labor markets. This paper fails to find variables which can control the relocating factor. Therefore, the conclusion of this paper is still doubtful.
Appendix
Table 1 overview of research results Source: Borjas (1994), pp.1697
study Impact of immigrants on Dependent variable Effect on local people’s wage(in percentage) from
one percent increase of immigrants
Goldin (1994) Native American Hourly wage Around -1%
Manacorda, Wadsworth and Manning (2012)
British Hourly wage -0.89%
Addison and Worswick (2002) Austrilian Monthly earning 0.121%
Altonji and Card (1991) Less skilled natives (US) Weekly wages -1.2%
Bean Lowell and Taylor(1988) Native Mexican men
Black men
Annual earnings Annual earnings
-0.005% to 0.05% -0.003% to 0.06%
Borjas (1990) US White native men
US Black native men
Annual earning Annual earning
-0.01% -0.02%
Grossman (1982) All Canadian natives Annual earnings -0.02%
LaLonde and Topel (1991) Young Black Natives
Young Hispanic Natives
Annual earnings Annual earnings
-0.06% -0.01%
Bibliography
Addison T. & Worswick C. (2002). The impact of immigration on the earnings of natives: Evidence from Australian micro data. The economic record, 78(1), 68-78.
Altonji J.G. & Card D. (1991). The Effects of Immigration on the Labor Market Outcomes of Less-skilled Natives. University of Chicago Press.
Birgitta A., Anders H, David I. (2001). Effects of war and organized violence on children: A study of Bosnian refugees in Sweden. American journal of Orthopsychiatry, 71(1).
Bonin, H. (2005) Wage and Employment Effects of Immigration to Germany: Evidence from a Skill Group Approach. IZA Discussion Paper, 1875.
Borjas, G.J. (1994). The economics of immigration. Journal of Economic Literature, 32(4), 201-234. Borjas G.J., Grogger J. & Hanson G.H. (2008). Imperfect substitution between immigrants and natives:
a reappraisal.
Borjas G. J. (1990). Friends or strangers: The impact of immigrants on the U.S. economy. New York: Basic Books.
Card, D. (1990). The Impact of the Mariel Boatlift on the Miami Labor Market. Industrial and Labor Relations Review, 43, 245–257.
Goldin, C. (1994).The Political Economy of Immigration Restriction in the United States,
1890–1921. The Regulated Economy: A Historical Approach to Political Economy, 223–257. Grossmann, J. B. (1982). The Substitutability of Natives and Immigrants in Production.
Review of Economics and Statistics, 54, 596–603.
Holmøyvik E.(2005) . The theory of sovereignty and the Swedish-Norwegian union of 1814. Journal of the History of International Law , 7, 137–156.
Jefferson K. W. (1999). The Bosnian War Crimes Trial Simulation: Teaching Students about the Fuzziness of WorldPolitics and International Law. Political Science and Politics, 32 (3), 588-592. Kahanec, M., Zaiceva A. and Klaus F. Z. (2009). Lessons from Migration after EU Enlargement. IZA
Discussion Paper, 4230.
Kondylis F. (2009). Conflict displacement and labor market outcomes in post-war Bosnia and Herzegovina. Journal of Development Economics, 93, 235–248.
Lalonde R.&Topel, R. (1992).Labor Market Adjustments to Increased Immigration, 167–199. 22
Longhi S., Nijkamp P. &Poot J. (2005). A meta-analytic assessment of the effect of immigration on wages. Journal of economic surveys, 19(3).
Melitz M.J., Obstfeld M. &Krugman P. R. (2012). International Economics Theory & Policy. Pearson Education, ninth edition.
Manacorda M., Wadsworth J. & Manning A. (2012). The impact of immigration on the structure of wages: theory and evidence from Britain. Journal of the European Economic Association, 10(1), 120–151.
Rachel M.F. & Hunt J. (1995). The impact of immigrants on host country wages, employment and growth. Journal of Economic Perspectives, 9(2).
Sumantra B. (2009). Contested lands: Israel-Palestine, Kashmir, Bosnia, Cyprus, and Sri Lanka. Harvard University Press.
Stock J.H. &Watson M.M. (2012). Introduction to Econometrics. Pearson Education, third edition. Taylor J. E. (1987). Undocumented Mexico-U.S. Migration and the Returns to Households
in Rural Mexico.69(3), pp. 626–38.
Zorlu A. and Hartog J.(2002). The effect of immigration on wages in three European countries. Journal of Population Economics, 18, 113-151.
ØstbyeS. & Westerlund O. (2010). Industrial structure, regional productivity and convergence: The case of Norway and Sweden. European Urban and Regional Studies, 18(1), 47–61.
