Case Qoutaiba Boeshaar, B.A. Washington, DC April 15, 2016

Transcription

1 DO EUROPEANS PERCEIVED INCOME LEVELS AFFECT THEIR ATTITUDES TOWARDS IMMIGRANTS? A REGRESSION ANALYSIS OF INDIVIDUAL SELF- REPORTED COMFORT IN INCOME VIS A VIS ATTITUDES TOWARDS IMMIGRATION USING THE EUROPEAN SOCIAL SURVEY A Thesis submitted to the Faculty of the Graduate School of Arts and Sciences of Georgetown University in partial fulfillment of the requirements for the degree of Master of Arts in Public Policy By Case Qoutaiba Boeshaar, B.A. Washington, DC April 15, 2016

2 Copyright 2016 by Case Qoutaiba Boeshaar All Rights Reserved ii

3 DO EUROPEANS PERCEIVED INCOME LEVELS AFFECT THEIR ATTITUDES TOWARDS IMMIGRANTS? A REGRESSION ANALYSIS OF INDIVIDUAL SELF- REPORTED COMFORT IN INCOME VIS A VIS ATTITUDES TOWARDS IMMIGRATION USING THE EUROPEAN SOCIAL SURVEY Case Qoutaiba Boeshaar, B.A. Thesis Advisor: Andreas Kern, Ph.D. ABSTRACT This paper explores individual economic determinants of attitudes towards immigration in 15 European countries, using the 2014 round of the European Social Survey. Most literature tends to focus on economic data when analyzing attitudes towards immigration. While this can be useful, it may miss the fact that individuals attitudes are often shaped by their perceived, rather than actual, economic status. This paper aims to fill this gap by examining respondents self-assessed comfort in their household incomes as a determinant of their attitudes towards immigration. My findings show that there is a statistically significant relationship between how comfortable a respondent is with his/her own household income and how he/she perceives immigrants. Additional robustness checks were conducted, in addition to an instrumental variable analysis, to strengthen the validity of these findings. Given the recent influx of immigration, the largest since World War II, European governments could benefit to better understand their citizens attitudes towards immigration. It could be beneficial to develop more integrative policy that is inclusive of both the local citizens and immigrants. iii

4 This paper would have not been possible without the help of my advisor, Andreas Kern, who was supportive throughout the entire process. Thanks to my family for their full support in all aspects of grad school. Thanks to Aaron Albert for his guidance with Stata. Thanks to Tara Siegel for her help and support in editing the paper. And a thanks to Zeina giving me more work in removing cat hair from my clothes. I am indebted to all of you! Case Qoutaiba Boeshaar iv

5 TABLE OF CONTENTS Introduction... 1 Theory... 3 Empirical Analysis... 6 Regression Results Robustness Checks Concluding Remarks Works Cited Appendix v

6 LIST OF FIGURES AND TABLES Table 1: Voting support for radical right wing parties in Europe... 1 Table 2: Respondents by country... 9 Table 3: Baseline regression using Reject Factor Table 4: Variation in the independent variables Table 5: Plausibility tests Table 6: Stratification Table 7: Summary statistics Table 8: Variable definitions Figure 1: Attitudes towards immigration by country Table 9: Ordered Probit and Instrumental Variable vi

7 INTRODUCTION The European Union is experiencing an unprecedented wave of mass immigration and refugees from the Middle East, Africa, and Central Asia. In 2015, an estimated one million migrants have entered Europe; since the beginning of 2016, over 173,728 people crossed into Europe by sea (UNHCR 2016). Asylum claims totaled 1,321,560 in 2015, with more than 476,000 for Germany alone ( Asylum statistics 2016). This dramatic increase in migration to Europe has coincided with a rise in the popularity of right wing parties. Europeans with negative views of immigrants tend to vote for these parties, whose main platforms and rhetoric tend to be fiercely anti-immigrant (Brug, Fennema, and Tillie 2000). Native-born citizens often scapegoat immigrants, blaming them for social tensions, increased crime, job competition, and draining local social services. Since the early 2000s, European right wing parties have made sweeping gains in local elections (Lublin 2015). Table 1 demonstrates this below. Table 1: Voting support for radical right wing parties in Europe Country Last Election Year Last Election Votes Polling in 2015 Increase Since Last Election Austria Belgium France Italy Netherlands Norway Sweden Mean Source: Anti-immigrant parties on the rise internationally (Lublin, 2015) 1

8 Evidence shows that immigration is a becoming an increasingly important topic for European voters, which is why it is vital to analyze this issue. The current refugee crisis in Europe has allowed right wing parties to flourish across the continent. In Germany and France, for example, right wing parties such Alternative fuer Deutschland and Front National have gained substantial public support (Amann et al. 2016). Given these incompatible recent trends Europe s increasingly anti-migrant politics and its unprecedented migration crisis it is critical to understand divers of European citizens attitudes towards immigrants. The focus of previous research on attitudes towards immigration tends to focus on hard economic data. While this is certainly a useful analysis, it may miss the fact that individuals attitudes are often shaped by their perceived, rather than actual, economic status. Previous research by Stockemer (2015) found that individual perceptions on immigrants, rather than the actual percentages of immigrants in a country, was associated with how likely Europeans were to vote for anti-immigrant right wing parties. Building on this research, I focus on perception-based income and other individual level characteristics that may shape individuals attitudes towards immigration on a cross-country level in Europe. Specifically, this study asks if respondents who are less confident in their household income are more likely to reject immigrants. I employ an ordinary least squares (OLS) model as my baseline, and conduct robustness checks using a Probit, additional OLS, and instrumental variable analysis to check my findings. Using the European Social Survey s 2014 wave, I find that individuals confidence in their household income significantly impacts their tolerance for immigration, with less confidence leading to more anti-immigrant sentiments. I rerun my model using a two-stage least squares (2SLS) setup 2

9 using an individual s height as an instrument for comfort in household income and find identical results. I also find that a person s confidence in his/her household income has a relationship with how positively they view homosexuals and minority groups, with less confidence leading to less tolerance towards these groups. From a policy perspective, these results suggest that addressing socioeconomic issues and integrative policy may help reduce social tensions during episodes of mass immigration and integration into Europe. THEORY The refugee crisis has created immense challenges of integrating an unprecedented number of non-white, non-christian people into European societies. The growing negative sentiment against refugees is a common phenomenon. A 2015 IMF report on international migration states that, in the short term, immigration incurred negative costs for receiving countries. However, the same report also emphasizes that when immigrants and refugees are allowed to contribute economically, there can be net economic benefits to immigration. This is especially true in Europe, which is currently experiencing an aging population which is leading to a decline in the workforce (IMF 2015). Other research agrees with the IMF report, demonstrating that migrants can be beneficial to their new host country s economy when given the opportunity to work. A study by Maystadt and Verwimp (2009) on refugees from Burundi and Rwanda show that refugees provided net economic benefits for their new host country, Tanzania. A World Bank report by Carpio and Wagner (2015) looked at the effect of wages in Turkey from the influx of Syrian migrants and 3

10 refugees and found a gradual wage increase across the board in the country. A recent Brookings article by Cali and Sekkarie (2015) dispels myths of refugees harming their host country s economies. Cali and Sekkarie find that when refugees are allowed the opportunity, they can provide net benefits to fiscal contributions of their new host country. Nonetheless, attitudes towards refugees and immigrants remain a controversial topic in Europe. What determines individuals attitudes towards immigration? Extensive research has examined attitudes towards immigration both internationally and in Europe. Some of the literature considers economic factors. Scheve and Slaughter (2001) use individual-level data and find that less skilled individuals with a weaker position in the labor market are more likely to oppose immigration. Similarly, Mayda (2006) finds that low-skilled individuals in low per capita GDP countries have more negative attitudes towards immigration. Bridges and Mateut (2014) find that natives in countries with higher ratios of social security expenditure to total GDP view immigrants of a different race as a greater threat to their country s social welfare state. Meuleman, Davidov, and Billiet (2009) find that respondents from countries that experienced decreasing unemployment rates during a five year period demonstrated more positive attitudes towards immigration a. Paas and Halapuu (2012) find that people who belong to higher income deciles tend to have more favorable views towards immigrants. Beyond these factors, non-economic factors also shape individuals attitudes towards immigration. Card, Dustmann, and Preston (2012) find that older, less-educated people view a However, Meuleman et al. (2009) found no relationship on attitudes towards immigration when using real GDP growth as their indicator. They concluded that is possible that GDP was an ineffective measure of an increase in wealth because GDP does not calculate how equitable distribution is among the population. 4

11 immigration more negatively due to concerns about changes in the ethnic composition of their countries. Hainmueller and Hiscox (2007) find that higher levels of education correlate with more favorable views on immigration. Paas et al. (2012) find that younger, better-educated people are more likely to have more tolerant attitudes towards immigrants. Substantial research focuses on observed economic and non-economic data. For example, Paas et al. (2012) use income deciles in their economic analysis when determining attitudes towards immigration. These factors can be very valuable, yet it is also important to consider how individuals perceptions impact their attitudes. Some researchers have found perceptions to be influential. Stockemer (2015), for example, found a correlation with how threatening Europeans perceived immigrants to be and how likely they were to vote for right wing parties, irrespective of the actual immigration percentages in their own country. Given the observed relevance of both individual-level perceptions and economic factors in determining attitudes towards immigration, I opt to investigate individuals perceived economic realities. Specifically, I ask: do respondents self-reported comfort in their own household income affect their attitudes towards immigration. I hypothesize that respondents who are less confident in their household income are more likely to reject immigrants. Based on previous research, it would be reasonable to expect that respondents who answer that they are not confident in their household income to be less tolerant towards immigration. This may be due to this group of respondents having lower skill or education levels (Scheve and Slaughter 2001, Mayda 2006, Paas and Halapuu 2012). These respondents may consider immigrants a source of increased competitiveness in the labor market, and a burden to national 5

12 social welfare programs. In the next step, I will test my hypothesis using an OLS model and additional robustness checks. EMPIRICAL ANALYSIS To test my hypothesis, I run several regressions in order to isolate the effect of perceived income on attitudes towards immigrants. First, I run an OLS regression using Reject Factor as the dependent variable and Income Feeling as the independent variable. I add controls for age, education, gender, political beliefs, and born in country, in addition to time and country fixed effects. (Summary statistics of all variables used in the baseline regression can be found in Table 7.) Second, I run robustness checks using self-reported health, happiness, income decile, and test for the relationship Income Feeling has on attitudes towards Gypsies, Jews, Muslims, and homosexuals. Additional robustness checks consider country-level subsets that are either former Eastern Bloc, economically hard hit, or have unemployment rates above 7.5%. These robustness tests serve to check for possible measurement error or omitted variable bias. Third, I rerun my model in order to account for a potential bias arising from a potential reverse causal mechanism at play: I re-estimate my model using an ordered Probit model and a 2 SLS approach. In the next section, I outline my baseline regression, including the construction of my key dependent and independent variables and my modeling setup. My OLS baseline regression model is as follows: 6

13 γ (Reject Factor) = β0 + B1IncomeFeeling + β2(education) + β3(age) + β4(gender) + β5(born in Country) + β6(left Right Scale) + β7(interview Month) + β8(country) + μ Reject factor: An index variable is created using simple latent factoring of three immigrationrelated survey questions. This methodology is similar to the one employed by Meuleman et al. (2009) and Paas et al. (2012) when creating the immigration index for the dependent variable named Reject factor. Combining three immigration related questions for the dependent variable allows for a more precise aggregate measure covering multiple dimensions of ethnicity, income levels, and race. The main dependent variable is an index of three immigration survey questions, created using factor analysis. The three questions ask respondents how many immigrants they are willing to accept: 1) Allow many/few immigrants from poorer countries outside Europe; 2) Allow many/few immigrants of same race/ethnic group as majority; and 3) Allow many/few immigrants of different race/ethnic group from majority. The four possible responses to this question are: Allow many to come and live here, allow some, allow a few, or allow none. Higher values indicate a higher probability of a respondent wanting to reject immigrants. Income Feeling: Independent variable showing respondents self-reported comfort in their current household's income. The question asks respondents to rate their Feeling about household's income nowadays. There are four possible answers to this question: living comfortably, coping, difficult, or very on present income. Higher values indicate a respondent s belief that they are less able to cope at their present income. 7

14 Second, for robustness of these results, I run additional tests in order to strengthen the validity of my results, in addition to an ordered Probit model to address potential estimation bias. Third, I use Height as an instrumental variable on Income Feeling to minimize the risk of reverse causality. Education is a control variable that is commonly used in immigration analyses as a measure of individual skill level. Age, gender, and being born in country are used in the analysis of Paas et al. (2012) for determining attitudes towards immigration. Left right scale has been included in this analysis due to the growing influence of anti-immigrant right wing parties in Europe as detailed in the introduction section of this paper. Self-reported political beliefs could be a helpful indicator of individual determinants of attitudes towards immigration. Country and time fixed effects have been used in analyses including Mayda (2006) and Meuleman et al. (2009). Attitudes towards immigrants have been found to vary on a country by country basis, and so its inclusion is vital to finding differences in immigration attitudes within the European Union. A description of all variables used in the baseline regression and the robustness checks can be found in the Appendix under Table 8. The data for this setup comes from the 7th round of the European Social Survey. This dataset records survey responses conducted in the year 2014 in 15 European nations. The full list, along with the breakdown of the number of respondents by country, can be found below in Table 2. 8

15 Table 2: Respondents by country Country Number Cumulative Total Austria 1, Belgium 1, Czech Republic 1, Denmark 1, Estonia 1, Finland 2, France 1, Germany 2, Ireland 2, Netherlands 1, Norway 1, Poland 1, Slovenia 1, Sweden 1, Switzerland 1, Total 26, The survey has both population weights and design weights. Design weights account for within country sampling features, while population weights scale countries to account for different sample/population ratios (ESS, 2016). This study combines both weights into a single weight variable for cross country analysis since all 15 nations are used in the analysis. In further iterations of the model, I also include an interview date fixed effect to determine to control for possible events or incidents during specific months. In addition, I include country dummies on the right hand side of my equation. The reason for the inclusion of country effects is that prior 9

16 research b has found significant cross-country differences and given the heterogeneity among European countries, I want to make sure that specific country outliers are not driving my results. Please refer to Appendix Figure 1 for a depiction of a color coded relationship between countries and the probability of rejecting immigrants. My results show significant cross-country variation in attitudes towards immigration. Possible explanations for this may include the countries specific ethnic compositions, economic conditions, policies, cultures, and immigration histories. All of these factors could affect results. For example, Mayda (2006) found that respondents from wealthier nations had more positive attitudes towards immigrants when they were better educated, which could partially explain the more tolerant attitudes found in countries such as Germany, Norway, and Sweden. This paper will not go into full detail of cross-country differences, but they are important to consider and could be worth investigating further. REGRESSION RESULTS OLS regression results demonstrate that confidence in household income plays a significant role in determining attitudes towards immigration (see Table 3 below). Across Models 1 through 4, there is a consistent pattern between the Reject Factor and Income Feeling variables. Income Feeling remains highly statistically significant at conventional levels (p<0.1 and below) in all models, even when adding controls. As comfort in a household s income decreases, respondents become more likely to reject immigrants, holding all else constant. This b For a similar argument, see Meuleman et al. (2009) 10

17 demonstrates that the less economically secure people feel, the less tolerant they will be of immigrants. Table 3: Baseline regression using Reject Factor (1) (2) (3) (4) VARIABLES Model 1 Model 2 Model 3 Model 4 Coping on income 0.303*** 0.300*** 0.298*** 0.234*** (0.0201) (0.0203) (0.0203) (0.0197) Tough on income 0.586*** 0.583*** 0.579*** 0.472*** (0.0349) (0.0352) (0.0350) (0.0339) Very tough on income 0.756*** 0.754*** 0.754*** 0.579*** (0.0753) (0.0756) (0.0755) (0.0722) Education *** ( ) Age *** ( ) Gender (0.0186) Born in Country *** (0.0303) Left Right Scale *** ( ) Interview Month FE ( ) ( ) ( ) Country FE * *** ( ) ( ) Constant *** *** *** 0.377*** (0.0148) (0.0279) (0.0312) (0.0668) Observations 26,550 26,547 26,547 24,504 R-squared Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Model 2 shows the baseline regression with only time fixed effects included from the Interview Month variable. Disaggregating by month reveals that the months of January, 11

18 February, and June correlate with the highest increase in the likeliness of rejecting immigrants. The months of August, September, and October show a decrease in the likeliness of rejecting immigrants. The values are positive, but low, for the months of November and December. However, Interview Month is not significant at conventional levels (p<0.1 and below); thus, its results should be approached with caution. Model 3 includes both time and country fixed effects, yielding only slightly different coefficients than the preceding models. This indicates that even controlling country specific effects does not significantly weaken the postulated relationship. In fact, these findings lend additional support for my hypothesis. Model 4 includes time and country fixed effects, in additions to the full set of controls. The controls, Education, Age, Born in Country, and Left Right Scale are significant at conventional levels (p<0.01 and below) whereas Gender is not significant. Higher age is associated with an increase in the probability of rejecting immigrants holding all else constant. The findings on age are consistent with Paas et al. (2012) who found that younger respondents are likely to be more tolerant towards immigration. A possible explanation may be contact theory, where individuals who spend more time abroad develop more tolerant attitudes towards immigration, or if younger generations have more exposure to multi ethnic societies due to increased immigration. Higher levels of education is associated with respondents being less likely to reject immigrants. The findings on education are consistent with Hainmueller et al. (2007) and Paas et al. (2012) where higher education has a positive relationship with attitudes towards immigration. Education is often used as an indicator of skill, and previous research has shown that more skilled individuals are on average less threated by 12

19 immigrants than their less skilled peers. The gender variable is not significant at conventional levels, which is in line with findings from Paas et al. (2012). Respondents who are native born are correlated with a higher probability of rejecting immigrants. These findings are again consistent with Paas et al. (2012) who found that non-native-born respondents were more tolerant towards immigration; they attributed a possible explanation to contact theory (additional exposure to foreigners makes a person more tolerant towards them). Respondents who selfidentified as having right wing beliefs were more likely to reject immigrants. As right wing parties in Europe campaign heavily on anti-immigrant platforms, it is reasonable to assume that people with more right wing beliefs are going to harbor more negative views towards immigrants. My results show that there is a significant relationship between income feeling and how tolerant a person is towards immigration. ROBUSTNESS CHECKS The results above appear significant and consistent. Nonetheless, OLS regressions tend to be susceptible to omitted variable bias and measurement error. To address these issues, I conduct additional robustness checks to assess the strength of my findings. Below, Table 4 below shows results from tests run with different independent variables. Model 1, included for comparison, is the baseline regression of Reject Factor on Income Feeling. 13

20 Table 4: Variation in the independent variables (1) (2) (3) (4) VARIABLES Model 1 Model 2 Model 3 Model 4 Income Feeling 0.223*** (0.0136) Income Decile *** ( ) Happiness *** ( ) Health *** (0.0122) Controls Y Y Y Y Constant *** *** *** *** (0.0702) (0.0742) (0.0688) (0.0688) Observations 24,504 20,197 24,452 24,493 R-squared Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Model 2 replaces the independent variable with Income Decile. A one-unit increase in Income Decile is associated with lower income for a respondent on a decile scale. This means that the higher the value for Income Decile, the poorer the respondent. Income Decile has a positive relationship with Reject Factor, and is significant at conventional levels (p<0.01), holding all else constant. The results are consistent between Models 1 and 2 and show a similar relationship, namely: respondents who are less confident in their household income or are lower on the income decile are associated with a higher likelihood of rejecting immigrants. These findings are in line with Paas et al. (2012) and Bridges et al. (2014), who both found that individuals with higher incomes are more likely to be tolerant towards immigrants. 14

21 Models 3 and 4 replace the independent variable with the variables Happiness and Health, respectively. Happiness and Health are self-assessed by respondents. Higher values of Happiness and Health reflect lower levels of respondents wellbeing. The results for Models 3 and 4 are very similar, with both showing a positive relationship between the independent variable and Reject Factor. This means that respondents with lower levels of self-reported wellbeing are associated with a higher probability of rejecting immigrants. Both Happiness and Healthy are significant at the conventional levels (p<0.01 and below), holding all else constant. These results are consistent with previous research, which has shown that there is a positive relationship between income and wellbeing (Rojas, 2011). Therefore, if Income Feeling and Income Decile are associated with higher levels of respondents wanting to reject immigrants, we would expect a similar relationship for respondents who are less healthy or happy. This paper conducted additional plausibility tests to see if Income Feeling has a relationship with respondents views towards homosexuals or minorities. If lower household income comfort levels coincide with respondents decreased tolerance of immigrants, would the same correlation be seen towards other groups as well? Table 5 answers this question with regression results using variables that ask respondents about whether or not they believe homosexuals should be allowed freedom, and whether or not Muslims, Jews, or Gypsies should be allowed into the respondents country. 15

22 Table 5: Plausibility tests (1) (2) (3) (4) VARIABLES LGBT Rights c Allow Muslims Allow Jews Allow Gypsies Coping on income 0.256*** 0.260*** 0.213*** 0.208*** (0.0211) (0.0189) (0.0170) (0.0200) Tough on income 0.387*** 0.432*** 0.412*** 0.344*** (0.0409) (0.0296) (0.0289) (0.0302) Very tough on income 0.384*** 0.607*** 0.529*** 0.484*** (0.0728) (0.0625) (0.0669) (0.0658) Controls Y Y Y Y Constant 1.109*** 1.874*** 1.811*** 2.062*** (0.0820) (0.0626) (0.0576) (0.0661) Observations 24,504 24,158 24,073 24,112 R-squared Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 The dependent variable has been changed in the analysis, but the relationship of higher values corresponding to rejection remains the same from previous analyses using Reject Factor. The same controls are used as before. All results for income feeling were significant at conventional levels. The findings show that as respondents are less comfortable with their household income, they become less tolerant of homosexuals, Muslims, Jews, and Gypsies. Surprisingly, the gender variable has become significant with a positive sign for the Freedom for homosexuals and allowing Muslims questions, and at the p<0.05 level for allowing Gypsies. These results show that women are more tolerant towards minorities and homosexuals compared to their male counterparts. These findings strengthen the hypothesis that how comfortable an c Whether or not the respondent believes that gays and lesbians should be free to live life as they wish. 16

23 individual is with their household income has a relationship with how tolerant they are towards immigrants, homosexuals, and minorities. Selection bias could result from outliers within the data. This is why I have conducted additional robustness checks using subsamples of different countries using specific criteria. The results can be found below in table 6 below. VARIABLES Countries Former Eastern Bloc Countries Table 6: Stratification (1) (2) (3) (4) Hardest hit economically Unemployment above 7.5% Coping on income 0.234*** 0.201*** 0.229*** 0.216*** (0.0197) (0.0200) (0.0197) (0.0194) Tough on income 0.473*** 0.423*** 0.464*** 0.429*** (0.0340) (0.0350) (0.0341) (0.0333) Very tough on income 0.580*** 0.553*** 0.575*** 0.572*** (0.0724) (0.0723) (0.0724) (0.0726) Country *** ( ) Former Eastern Bloc 0.206*** (0.0268) Hardest hit economically 0.240*** (0.0196) Unemployment above 7.5% 0.297*** (0.0219) Controls Y Y Y Y Constant *** *** *** *** (0.0670) (0.0671) (0.0657) (0.0647) Observations 24,504 24,504 24,504 24,504 R-squared Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 17

24 The first subsample is former Eastern Bloc countries, which are defined as countries which were part of the former Soviet Union before its dissolution in Former Eastern Bloc countries in this dataset are the Czech Republic, Estonia, Poland, and Slovenia. The second subsample includes nations which were hardest hit economically in the financial crisis of (Fischer 2015). The nations which are considered the hardest hit in this dataset are Estonia, Ireland, and Slovenia. The last subsample includes European countries in this dataset with unemployment rates above 7.5%. The rates used are from data gathered by EuroStat ( Unemployment Rates 2016). All three subsamples are correlated with less tolerant attitudes towards immigration. Possible explanations include that Eastern bloc countries are less experienced with immigration compared to their Western counterparts, and that countries with less than optimal economic conditions creates a climate that is less positive towards additional immigration. In order to address potential estimation bias from applying an OLS model, I re-ran my baseline model using an ordered Probit model specification. The findings were identical in terms of the significance and relationship (see Table 9 of the Appendix). A potential additional threat towards the validity of my results is the possibility of reverse causality. It might be the case that more intolerant citizens are also less educated and thus more vulnerable to income shocks. In order to eliminate these effects, I re-run my model using a 2SLS setup. As an instrumental variable, I choose the individual s height. The reason for selecting height is that previous research indicates a positive relationship between income and height d. However, it is less likely d Findings by Judge and Cable (2012) showed that every additional inch above average height would lead to more individual earnings per year. 18

25 that height has an impact on attitudes towards immigration and thus should not be correlated with initial error term. The results remain identical when compared to my base results both in terms of significance, directionality, and relationship (see Table 9 of the Appendix). CONCLUDING REMARKS The continuing refugee crisis and rise of right wing parties in Europe render immigration an ongoing policy concern. As lawmakers at the national and European Union level seek solutions to the crisis, it will be valuable to understand citizens perceptions and drivers of citizens attitudes towards immigration. Relying exclusively on economic data may miss another dimension: the reality that individuals attitudes are shaped by their perceived not just actual economic status. This paper, therefore, examines how individuals perceived confidence in their household income can affect their tolerance towards immigration. Are Europeans who have less confidence in their household income are more likely to reject immigrants? My findings show strong evidence that this is true a respondent s comfort in his/her household income influences his/her attitudes towards immigrants. The less comfortable respondents are, the more likely they are to reject immigration; conversely, people who are more comfortable with their income are less likely to reject immigration. These results are consistent across all models and are highly significant at conventional levels (p<0.1 and below). Additional investigations using a Probit model and 2SLS, yield results identical to my baseline regressions. Additional robustness tests also show that respondents who are less comfortable with their income are more likely to reject Gypsies, Jews, Muslims, and homosexuals. Therefore, a 19

26 person s comfort with his/her income can play a key role in determining his/her attitudes towards others, on the levels of ethnicity, nationality, religious denomination, and sexual orientation. Consequently, individual level economic concerns should be considered when crafting policy to promote integration. The European Union s refugee and migrant crisis is highly unlikely to subside in the near future, so policymakers should consider long term response strategies. Large numbers of migrants and asylum seekers have already taken residence in European countries. While host countries may suffer initial economic losses from the influx of immigrants, citizens and policymakers should consider the benefits of long term integration when evaluating policy. For example, research has shown that when allowed to participate in the workforce, refugees and migrants can bring net economic gains for their host countries (Maystadt and Verwimp 2009, Carpio and Wagner 2015). This phenomenon may prove helpful to European nations with aging populations and declining workforces. It is important to note that immigration policy that is designed to be integrative in the long term has the best chance of yielding positive results (IMF 2015). In addition, education has consistently proven be a significant variable in defining attitudes towards immigration. Consequently, education initiatives may prove fruitful policy options. That said, interventions in the educational system and awareness campaigns for receiving countries should not be seen as just a quick fix, but rather opportunities to deliver positive results in the long run. Critics of such approaches may point to some countries rapid rise in youth unemployment, sluggish economic recoveries, and scarcity of low skilled jobs. Against this 20

27 background, integrating refugees into recipient countries labor markets may be a challenging task for some countries. Nevertheless, uncoordinated, ad-hoc policies such as Germany s can produce more social tensions, stronger hostility towards immigrants, and further disagreement among an already strained European Union. In addition, as noted above, refugees and migrants are already present in Europe, and are likely to continue arriving in the future. The refugee and migration crisis opens a window of opportunity to foster European integration and promote innovation in the design of new integration policies emphasizing social inclusion and integration. This may be possible particularly if policymakers remain attuned to individual perceptions of personal economic wellbeing. Understanding the relationship between individuals economic comfort and their tolerance can help policymakers formulate policy that promotes long term economic growth, rather than creating further social tensions. Careful policymaking will, at worst, mitigate the refugee and migrant crisis and, at best, transform this crisis into an opportunity. 21

28 WORKS CITED Amann, Melanie, Matthias Bartsch, Jan Friedmann, Nils Minkmar, Michael Sauga, and Steffen Winter. "The Hate Preachers: Inside Germany's Dangerous New Populist Party - SPIEGEL ONLINE." SPIEGEL ONLINE. Web. 5 Mar "Migrant Crisis: Migration to Europe Explained in Seven Charts." BBC News. 04 Mar Web. 21 Mar Bridges, Sarah, and Simona Mateut. "Should They Stay or Should They Go? Attitudes Towards Immigration in Europe." Scott J Polit Econ Scottish Journal of Political Economy 61.4 (2014): Web. Brug, Wouter Van Der, Meindert Fennema, and Jean Tillie. "Anti-immigrant Parties in Europe: Ideological or Protest Vote?" European Journal of Political Research 37.1 (2000): Web. Card, David, Christian Dustmann, and Ian Preston. "Immigration, Wages, And Compositional Amenities." Journal of the European Economic Association 10.1 (2012): Web. Carpio, Ximena V. Del, and Mathis Wagner. "The Impact of Syrians Refugees on the Turkish Labor Market." Policy Research Working Papers (2015). Web. "Unemployment Rates, Seasonally Adjusted, February 2016." EuroStat. Web. 15 Mar "Weighting European Social Survey Data." European Social Survey. Web. 20 Oct "Asylum Statistics." Eurostat. Web. 10 Apr Fischer, G. Unpublished Manuscript MS. Hainmueller, Jens, and Michael J. Hiscox. "Educated Preferences: Explaining Attitudes Toward Immigration in Europe." International Organization Int. Org (2007). Web. 22

29 IMF. "INTERNATIONAL MIGRATION: RECENT TRENDS, ECONOMIC IMPACTS, AND POLICY IMPLICATIONS." International Monetary Fund. International Monetary Fund, 12 Nov Web. 13 Apr Judge, Timothy A., and Daniel M. Cable. "The Effect of Physical Height on Workplace Success and Income: Preliminary Test of a Theoretical Model." Journal of Applied Psychology 89.3 (2004): Web. Lublin, David. "Anti-Immigrant Parties on the Rise Internationally." Seventh State. 30 Nov Web. 12 Feb Mayda, Anna Maria. "Who Is Against Immigration? A Cross-Country Investigation of Individual Attitudes toward Immigrants." Review of Economics and Statistics 88.3 (2006): Web. Maystadt, Jean-François, and Philip Verwimp. "Winners and Losers among a Refugee-Hosting Population." Economic Development and Cultural Change 62.4 (2014): Web. Meuleman, Bart, Eldad Davidov, and Jaak Billiet. "Changing Attitudes toward Immigration in Europe, : A Dynamic Group Conflict Theory Approach." Social Science Research 38.2 (2009): Web. Paas, Tiiu, and Vivika Halapuu. "Determinants of People s Attitudes Towards Immigrants in Europe." SSRN Electronic Journal SSRN Journal (2012). Web. Rojas, Mariano. "Happiness, Income, and Beyond." Applied Research Quality Life Applied Research in Quality of Life 6.3 (2011): Web. Scheve, Kenneth F., and Matthew J. Slaughter. "Labor Market Competition and Individual 23

30 Preferences Over Immigration Policy." Review of Economics and Statistics 83.1 (2001): Web. Stockemer, Daniel. "Structural Data on Immigration or Immigration Perceptions? What Accounts for the Electoral Success of the Radical Right in Europe?" JCMS: Journal of Common Market Studies (2015). Web. Massimiliano Calì, and Samia Sekkarie. "Much Ado about Nothing? The Economic Impact of Refugee 'invasions'" The Brookings Institution. 16 Sept Web. 12 Mar "UNHCR Refugees/Migrants Emergency Response - Mediterranean." UNHCR Refugees/Migrants Emergency Response - Mediterranean. Web. 10 Apr "Europe Blank Location Map." Wikimedia Commons. Web. 9 Apr

31 APPENDIX Table 7: Summary Statistics Variable Obs Mean Std. Dev. Min Max Income Feeling 26, Education 26, Age 26, Gender 26, Born in Country 26, Left Right Scale 24, Interview 26, Month Country 26,

32 Table 8: Variable Definitions Variable Reject Factor Income Feeling Definition Dependent variable created using factor analysis as an index of three immigration survey questions: 1- Allow many/few immigrants from poorer countries outside Europe 2- Allow many/few immigrants of same race/ethnic group as majority 3- Allow many/few immigrants of different race/ethnic group from majority. Higher values indicate a higher probability of a respondent to reject immigrants. Independent variable showing a respondent's selfreported comfort in their household's income. Higher values indicate a respondent s belief that they are less able to cope at their present income. Income Decile Education Age Gender Respondent's self-reported income level based on a 10 decile ranking. Ranking of the respondent's education level based on 10 deciles, ranging from no high school to a Master's degree. Respondent's age, 15 years and above. Respondent's gender, male or female. Born in Country Whether or not a respondent is native born. Left Right Scale Respondent's self-reported position on a left-right political spectrum. 26

33 Variable Definition Interview Month What month the interview took place on, January- December. Country LGBT Rights What European country the interview took place in. Whether or not the respondent believes that gays and lesbians should be free to live life as they wish. Allow Muslim Respondent's self-reported willingness to admit Muslims into the country. Allow Jew Respondent's self-reported willingness to admit Jews into the country. Allow Gypsy Respondent's self-reported willingness to admit Gypsies into the country. Health Respondent's self-reported health. Happy Respondent's self-reported happiness level. Height Respondent s self-reported height. 27

34 Figure 1: Attitudes towards immigration by country 28

35 Table 9: Ordered Probit and Instrumental Variable (1) (2) VARIABLES Ordered Probit IV Height Coping on income 0.270*** (0.0228) Tough on income 0.532*** (0.0374) Very tough on income 0.640*** (0.0790) Income Feeling (IV) 1.176*** (0.251) Controls Y Y Constant *** (0.009) (0.483) Observations 24,504 24,373 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 29