Information, perceived education level and attitudes towards refugees: Evidence from a randomized survey experiment *

Transcription

1 Information, perceived education level and attitudes towards refugees: Evidence from a randomized survey experiment * Philipp Lergetporer, Marc Piopiunik, and Lisa Simon WORK IN PROGRESS. PLEASE DO NOT CITE WITHOUT PERMISSION. Abstract In the past few years, Europe has witnessed an unprecedented influx of refugees. While natives attitudes towards refugees are decisive for the political feasibility of asylum policies, little is known about how these attitudes are shaped by refugees characteristics. We conducted a survey experiment with almost 5,000 students in Germany in which we randomly shifted the perception of refugees education levels through information provision. We find that the perceived education level significantly affects respondents concerns regarding labor market competition and the fiscal burden associated with refugee migration, but these concerns do not translate into general skepticism towards refugees. We underpin these findings with survey evidence on respondents opinion formation process towards refugees and extensive sub-group analyses. Keywords: refugees, asylum seekers, human capital, education, survey experiment, labor market JEL classification: H12, H53, I38, D83, D72, P16 * For helpful comments, we would like to thank Katharina Werner, Ludger Woessmann, and seminar participants at the Ifo Institute in Munich. Lergetporer: Ifo Institute at the University of Munich; CESifo; lergetporer@ifo.de. Piopiunik: Ifo Institute at the University of Munich; CESifo; piopiunik@ifo.de. Simon: Ifo Institute at the University of Munich; simon@ifo.de. 1

2 1. Introduction Since 2014, Europe has been facing an unprecedented influx of refugees. 1 In 2015 alone, more than 1.5 million individuals applied for asylum in Europe, with Germany registering the highest total number among European countries of some 440,000 applications (Eurostat, 2016). 2 The current refugee movements are exceptional not only in terms of magnitude, but also in terms of the refugees origin countries: Since Syria, Afghanistan, and Iraq constitute the main source countries, current refugees are perceived as culturally more distinct than those who sought asylum during previous refugee crisis such as the 1990 crisis after the Balkan wars (see Dustmann et al., 2016). Against this background, European politicians now face the difficult challenge to implement and enforce policies which, on the one hand, honor international commitments such as the 1951 Geneva Convention for Refugees or the Dublin Convention. 3 On the other hand, it is vital that policies towards refugees are supported by domestic voters in order to foster the political feasibility of these reforms and to preserve national solidarity and unity. The fact that public support for anti-immigration parties increased markedly in several European countries since the beginning of the refugee crisis indicates that voters skepticism towards refugees and asylum policies have not been fully appreciated by policy makers. 4 Despite the apparent importance of public attitudes towards refugees, little is known about the determinants of these attitudes and how responsive they are to specific characteristics of refugees. In this paper, we study the causal effect of refugees perceived education level on natives attitudes towards them. The education that refugees bring with them is decisive for policy makers since it determines the refugees prospects of labor market- and social integration and thus defines the policies required to foster integration. At the same time, the assessment of an refugees true education level has proven very difficult and is subject to a high degree of uncertainty (see Section 2). To answer the research question at hand, one requires (i) a measure for attitudes towards refugees among natives (preferably at the individual level) and (ii) an exogenous shifter of the perceived education level of refugees. To this end, we implemented a randomized online survey experiment with almost 5,000 students from four German universities. To exogenously shift the perception of the education level of 1 While the term asylum seeker describes a person seeking asylum and the term refugee a person whose asylum status has been approved, we use the more common term refugee throughout the paper. This is also the term that we used throughout the entire survey. 2 The Federal Ministry of Internal Affairs registered a total or more than 1.1 million asylum seekers entering Germany in 2015 (see Bundesministerium des Inneren, 2016). 3 The Geneva Convention broadly defines the rights of refugees and the obligations of hosting countries. The Dublin Convention came into force in and established the principle that the EU member state through which an asylum seeker first enters the EU is responsible for processing the asylum claim (see Dustmann et al., 2016). 4 Examples for electoral outcomes which have largely been attributed to voters rising anti-immigration sentiments in the course of the current refugee crisis include the Brexit referendum in the United Kingdom to exit the European Union (see Bansak et al., 2016) and the success of the Alternative für Deutschland (AfD), a right-wing populist party in Germany. The AfD won significant vote shares in several state elections, including the election in Mecklenburg-West Pomerania in which it outperformed chancellor Merkel s Christlich Demokratische Union (CDU) in Merkel s own home state (21 percent versus 19 percent). Furthermore, a historic share of 47 percent of the electorate voted for Norbert Hofer, the candidate from the far-right Freiheitliche Partei Oesterreich (FPOE), in the Austrian presidential elections in

3 refugees, we randomly assigned respondents to one of three experimental groups. The control group completed the online survey on attitudes towards refugees without any further information. Before answering the same questions, respondents in the HIGH SKILLED treatment were informed about a study which finds that refugees are rather well-educated (see UNHCR, 2015). In the LOW SKILLED treatment, we induced the opposite perception by informing participants about a study which shows that refugees tend to be poorly educated (see Woessmann, 2016). We find that these information treatments strongly shift respondents perceptions of the education level of refugees in the expected way. The treatments also affect respondents economic concerns. For example, compared to the control group, respondents in the HIGH SKILLED treatment are more likely to state that refugees will increase labor market competition, and respondents in the LOW SKILLED treatment are less likely to think that refugees will bring more revenues than costs for the government. These pronounced effects on respondents economic concerns, however, do not translate into shifts in general attitudes towards refugees: The information treatments have no significant effects on (i) preferences for how many refugees should be admitted to Germany in the future, (ii) perceptions that Germany admitted too many refugees in the previous year, (iii) preferences for allowing refugees to stay in Germany permanently, and (iv) satisfaction with the government s asylum- and refugee policy. The finding that economic concerns do not translate into general attitudes towards refugees is consistent with the previous literature which suggests that non-economic concerns are more important for attitude formation towards immigrants than economic concerns (see, for instance, Hainmueller and Hiscox (2010)). This notion is corroborated by a post-experimental survey question in which we asked our respondents how important several aspects are for the formation of their attitudes towards refugees. On average, humanitarian aspects is the most important one, whereas economic aspects are least important. Our results contribute to several strands of research. Within the extensive literature on attitudes towards immigration (see, for instance, Dustmann et al., 2016; Steinmayr, 2016), a growing number of studies employ survey experiments to investigate the causal determinants of these attitudes. In largescale surveys, Grigorieff et al. (2016) show that randomly provided information about immigration, such as the share of immigrants in the population and immigrants unemployment or incarceration rates, yield more favorable attitudes towards immigrants, but does not affect policy preferences. In a similar vein, Hainmueller and Hiscox (2010) study experimentally how concerns about labor market competition and about the fiscal burden on public services shape attitudes towards high- and lowskilled migration. In contrast to the theoretical predictions of a labor market competition model and a fiscal burden model, they find that both high-skilled and low-skilled natives prefer high-skilled immigrants and both groups equally dislike low-skilled immigrants. Since these studies do not explicitly investigate attitudes towards refugees 5, it is not clear to what extent their findings are transferrable to the current asylum situation in Europe. We are only aware of one experimental study 5 Furthermore, these surveys were fielded before the current refugee crisis. 3

4 which refers to the current refugee crisis: Bansak et al. (2016) present a survey experiment among 18,000 eligible voters in 15 European countries in which respondents evaluate different profiles of refugees. The specific characteristics of refugees have been varied experimentally across nine broad domains. They find that refugees with higher employability, more consistent asylum testimonies and vulnerability, and those who are Christian rather than Muslim are more likely to be accepted. We complement these important findings by investigating the impact of refugees perceived education level, a characteristic not studied by Bansak et al. (2016). From a methodological viewpoint, our paper is related to the growing literature which studies the effects of informing survey respondents on their attitudes and preferences in different areas. For instance, Cruces et al. (2013) investigate the effects of alleviating biased perceptions regarding one s relative position in the income distribution through information provision on preferences for redistribution. Focusing on a different policy area, Elias et al. (2015) study how preferences for markets for human organs are shaped by information about the current state of organ shortage and about academic studies on different strategies to alleviate the shortage. Further papers which investigate the causal effects of information provision in surveys include Kuziemko et al. (2015) on preferences for redistribution, Lergetporer et al. (2016) on preferences for public spending, and Bursztyn (2016) on the effects of information on local government spending on the public s ratings of their local government. The rest of this paper is structured as follows. In the next section, we discuss the importance of refugees educational level for labor market integration and labor market policies. We also discuss the difficulties of assessing the education level of refugees obtained in their home countries. In Section 3, we describe the opinion survey and the experimental design. Section 4 presents and discusses our results and Section 5 concludes. 2. The educational level of refugees and natives attitudes The successful integration of refugees into society critically depends on their integration into the labor market. Besides preserving national solidarity and social cohesion, successful labor market integration is also desirable from an economic viewpoint: Refugees who are integrated into the labor market are self-sufficient and thus not dependent on government aid. Furthermore, the large class of small to medium sized enterprises in Germany hopes that their deficit of skilled labor will be mitigated by the influx of refugees. The ability of policy makers to implement the policies necessary for successful integration depends on at least two circumstances: First, the natives attitudes towards refugees and second, the availability of accurate information regarding the skill level of refugees. Economic theories on natives attitudes towards immigrants The increasing popularity of anti-immigration parties in several European countries, including the AfD in Germany, after the onset of the European refugee crisis reflects the fact that immigrationcritical sentiments are on the rise among the European public. Thus, voters attitudes towards 4

5 immigration might be a key political obstacle to the implementation of integration policies. Economic models on natives attitudes towards immigration underline the importance of natives (perceived) education levels of immigrants. In particular, Hainmüller and Hiscox (2010) discuss two competing theories on how the skill level of immigrants affects attitudes towards them: According to the labor market competition model, natives are most opposed to immigrants with similar skill levels as their own because they expect that immigrants will increase competition on the labor market. For our sample of university students (who represent the upper tail of the skill distribution in Germany), this theory implies that they should be more skeptical about refugees if they consider them well educated. In contrast, the fiscal burden model predicts that high-income natives are more opposed to low-skilled immigration because low-skilled immigrants impose net burdens on public finance whereas highincome natives are net contributors. The present study provides a direct test of these competing theories in the context of the European refugee crisis. Note that these economic channels might not be the only ones through which refugees educational attainment affects natives attitudes. For instance, it might well be that natives are more favorable towards highly educated refugees because they expect that highly educated individuals can be integrated into society more easily6 or that they exhibit less criminal behavior (see, for instance, Lochner and Moretti, 2004). In this paper, we experimentally test the relevance of these alternative channels. Evidence on the education level of refugees Whether or not the society can reap the gains from successful labor market integration of refugees critically depends on (accurate information on) the skill level they bring with them.7.a major challenge in this context is that information on refugees characteristics, including their educational levels, is surrounded by a large degree of uncertainty. The sheer amount of incoming refugees throughout 2015 and 2016 created an enormous administrative challenge in solely registering the individuals, let alone documenting their educational attainment. Problems arises due to missing verifiable credentials such as graduation certificates, but also because of a lack of comparability between degrees from European educational institutions and from the refugees source countries. As a consequence, different studies which aim at approximating the education level or skills of refugees yielded inconsistent and seemingly contradicting conclusions. One of the first assessments of refugees education levels which received considerable media attention is UNHCR s interview study on Syrian refugees (UNHCR, 2015). The study draws a positive picture of refugees education level 6 For instance, Hanushek et al. (2017) show that returns to skills are larger in faster growing, which is consistent with the hypothesis that skills are particularly important for adapting to economic change. Under the assumption that skills are also important for adapting to changing life circumstances, it seems natural to assume that the refugees ability to integrate is positively correlated with their skills.. 7 From a legal perspective, granting prosecuted individuals temporary refugee status is a humanitarian act which is independent from economic considerations or the person s education level (see Dustmann et al., 2016). At the same time, it is crucial that policy makers take all (education-) measures necessary for integrating these individuals into the labor market (and hence, into society), because the majority of individuals who entered Europe during the refuge crisis are unlikely to return to their countries of origin in the foreseeable future (see Woessmann, 2016). 5

6 since it finds that 43% of them state to have some university education and another 43% state to have completed secondary education (UNHCR, 2015). The data for the study was collected by UNHCR border protection teams who conducted interviews with a non-random sample of Syrian refugees in various locations in Greece.8 Since the majority of the interviewees (50 percent) intended to request asylum in Germany, the study has been interpreted as a rough proxy for the education level of refugees in Germany.9 In contrast to that, Woessmann (2016) reaches more pessimistic conclusions regarding the education level of refuges: Comparing multiple data sources of the education level of refugees in Germany, Woessmann (2016) contends that only around 10 percent of refugees have a university degree, while two thirds do not have any professional qualification. Moreover, using data from Syrian 8th-graders who participated in the Trends in International Mathematics and Science Study (TIMSS) in 2011 (i.e., from the last test before the Syrian civil war started), the author finds that 65% fail to achieve the most basic proficiency level in mathematics and science as defined by the OCED. Comparing these outcomes with German school children of the same age,, the difference in skill levels amounts to 4 to 5 years of schooling.10 We use these contradicting findings from both studies, which were widely reported in the media, for designing two information treatments which shift our respondents perception of refugees skill levels. In particular, the fact that both studies reach contradicting conclusions allows us to design symmetric information treatments, one which shifts perceived education levels of refugees upwards and one which shifts it downwards (see Section 3.2).11 8 The asylum seekers arrived in the country between April and September Note that the authors of the study cautiously noted that the interviews were voluntary and interviewees were not required to verify their statements with credentials. 9 See Buber-Ennser et al. (2016) for a similar interview study with refugees in Austria. 10 Note that results from the TIMSS should be viewed as an approximation of the actual skill level of refugees who entered Germany: Syria is the most relevant, but not the only source country of the refugee crisis. Furthermore, little is known regarding the extent to which Syrian refugees who arrive in Germany represent a selected subgroup with respect to their skill levels. 11 A study conducted by the German Socio-Economic Panel in cooperation Institute for Employment Research (IAB) of the German Federal Employment Agency (BA), Research Centre on Migration, Integration, and Asylum of the Federal Office for Migration and Refugees (BAMF-FZ) provided newer evidence on refugees education levels. Their survey results show that 32 percent of refugees aged 18 years report to have a high-school degree and 13 percent of them hold a University degree. 19 percent report to have no formal or only primary schooling. Note that these results became available in late 2016, well after our survey was conducted. 6

7 3. Data and empirical strategy 3.1 The opinion survey General framework To implement the survey experiment, we ran an online survey with 4,846 students from four large German universities. 12 We were granted access to the universities mailing lists and invited students to participate in a short survey on refugees via . The informed students that the completion of the survey would take about 5 minutes, that participants would have the chance to win Amazon gift vouchers and that the survey would be anonymous. 13 The survey was computerized using Qualtrics software (Qualtrics, Provo, UT) and the field time was from June to August The timing of the survey is key: While most previous studies investigate attitudes towards immigration before the current refugee crisis, our survey was fielded just after the unprecedented refugee influx from 2015 slacked off. At this time, the public debate s focus shifted from the sheer number of refugees to questions on how to integrate them. Of course, the education level of refugees is central to the latter debate. As is typical for experiments in economics, our study relies on a self-selected sample of university students. In the context of this paper, university students are a highly relevant focus group: They represent the future high-skilled work force and high-income earners in Germany for which economic theories on attitudes towards immigration (i.e., the labor market model and the fiscal burden model) have clear, yet contradicting, predictions (see Section 2). Our experiment provides a clean test to discriminate between these two competing theories. Survey questions We designed three survey screens to investigate the effect of refugees perceived education level on attitudes towards them. In the following, we describe the content of each screen in detail. Screen 1: On the first screen, we used the following four questions to measure general attitudes towards refugees: 1. Compared to the current situation, should Germany admit more refugees, less refugees or the same number in the future? Answer categories: much more, somewhat more, the same amount, somewhat less, much less 2. What do you think about the number of refugees which Germany admitted last year? 12 The Technical University of Dresden, the University of Munich, the University of Konstanz, and the Technical University of Chemnitz participated in our study. Throughout the paper, we use university fixed effects. 13 We were able to guarantee anonymity and the chance to win an Amazon gift voucher (which was delivered through ) because survey answers were saved in a different file than addresses. This fact was known to the respondents. Multiple participations with the same computer were not possible. 7

8 Answer categories: far too many, somewhat too many, about the right magnitude, somewhat too few, far too few 3. Do you favor or oppose that refugees are allowed to stay in Germany permanently? Answer categories: strongly favor, somewhat favor, neither favor nor oppose, somewhat oppose, strongly oppose 4. How satisfied are you with the government s asylum- and refugee policy? Answer categories: very satisfied, somewhat satisfied, neither satisfied nor unsatisfied, somewhat unsatisfied, very unsatisfied These questions give us measures for general attitudes towards refugees. Following the wording of the general public discourse in Germany, we refer to refugees as refugees, even though only a subset of those individuals entering Germany in the wake of the refugee crisis qualify for refugee status according to the Geneva Convention (see Dustmann et al., 2016). We eschewed this fine distinction for the benefit of tangibility and comprehensibility of our survey questions. Screen 2: The second screen of the survey comprised specific statements on refugees which presumably (i) are responsive to refugees perceived education level and (ii) affect general attitudes towards refugees. Respondents were asked to articulate their agreement with the different statements on a five-point scale (completely agree, somewhat agree, neither agree nor disagree, somewhat disagree, completely disagree). In order to verify that our information treatments shifted our respondents perception of refugees education level in the intended directions, we included the following statement: On average, the refugees are well educated. The labor market competition model and the fiscal burden model propose different channels through which refugees perceived educational level affects attitudes of high skilled natives towards them. To assess the relevance of these competing channels, we asked our respondents to what extent they agree to the following statements: The refugees will increase competition on the labor market for me personally, In general, the refugees will increase competition on the labor market, The refugees will bring more revenues than costs for the government, Due to the government spending for refugees, I will have to forgo government benefits in the future and Due to the government spending for refugees, I will have to pay more taxes in the future. While these statements were designed to capture the channels of the labor market competition model and the fiscal burden model, there are certainly other potential channels through which refugees perceived education level might affect attitudes. Therefore, we included a number of statements which capture the multiple dimensions of the current discussion on integrating refugees and might uncover these alternative channels. The statements relate to cultural concerns ( The refugees are a cultural enrichment for Germany ), general economic concerns ( The refugees are good for the German economy ), concerns about the ability to integrate refugees ( Germany will succeed in integrating the refugees into the labor market, Germany will succeed in integrating the refugees into 8

9 society and Lack of language skills of the refugees are an obstacle for their labor market integration ) general assessment of the effects of the refugee crisis ( Generally speaking, the refugees are beneficial for Germany ) and concerns about the criminal behavior of refugees ( The crime rate will rise due to refugees criminal behavior ). Screen 3: On the third screen, we directly asked how important the following aspects are for the respondents opinion formation process towards refugees: Humanitarian aspects, Economic aspects, The refugees willingness to integrate, Religion/culture of the refugees, Refugees criminal behavior and Personal experience with refugees. Respondents were asked to rate each of these aspects on a five-point scale: very important, somewhat important, neither important nor unimportant, somewhat unimportant, very unimportant. Answers to these questions will generate important insights into the correspondence between specific statements on refugees (Screen 2) and general attitudes towards them (Screen 1). Within each screen, the order of questions was randomized in order to avoid primacy effects. At the end of the survey, we elicited a set of demographic characteristics. 3.2 The Survey Experiment To establish a causal link between natives perceived education level of refugees and the former attitudes towards the latter, we randomly assigned respondents to one of three experimental groups (control group, treatment HIGH SKILLED, and treatment LOW SKILLED) which differed in terms of the information provided at the top of Screen 1 and Screen 2. Members of the control group received the following information when answering both screens: With this survey, we would like to learn about your opinion on refugees. Please think of the current refugee situation in Germany when answering the survey. While this information was intended to fix ideas within survey respondents, note that it does not contain any reference to refugees educational level. Individuals in treatment HIGH SKILLED were presented the following information: With this survey, we would like to learn about your opinion on refugees. Please think of the current refugee situation in Germany when answering the survey. In this context, a study has found that the education level of refugees is rather high since 43 percent of the refugees from Syria have attended a University. The figure provided is based on a study by the UN Refugee Agency (UNHCR) which conducted 1,245 interviews with Syrian refugees in Greece between April and September Most refugees stated that their country of intended destination was Germany (see UNHCR, 2015). Respondents assigned to treatment LOW SKILLED received the following information: 9

10 With this survey, we would like to learn about your opinion on refugees. Please think of the current refugee situation in Germany when answering the survey. In this context, a study has found that the education level of refugees is rather low because 65 percent of the school students in Syria do not reach the basic level of academic competencies. This figure stems from Woessmann (2016) and was calculated from the 2011 wave of TIMSS (Trends in International Mathematics and Science Study), in which Syrian 8 th graders participated. Both studies have been published in quick succession (UNHCR (2015) was published in September 2015; Woessmann (2016) was published in January 2016). Despite the fact that both studies do not provide education measures for refugees entering Germany, they entered the German public discourse prominently. We take advantage of the high degree of uncertainty which surrounded the education level of refugees and use the information provided in these two studies as exogenous shifters of respondents perceived education level of refugees. The fact that we refer to seemingly contradicting studies on refugees education level allows us to implement symmetric treatments (i.e. a positive and a negative exogenous shifter of perceived education level) without deceiving our respondents. Our method to randomly inform survey respondents about research findings is similar to Elias et al. s (2015) survey experiment on public preferences for creating a market for organs. One potential shortcoming of our information treatment is that it only refers to Syrian refugees and remains silent about other source countries. This restriction was necessary due to a lack of data for other source countries. Note, however, that Syria is the major source country in the current refugee crisis. In the next section, we show that our information treatments indeed shifted the perceived education level of refugees. 3.3 Econometric Model In order to evaluate the impact of our information treatments, we estimate different versions of the following regression model: (1) where y i is the outcome of interest for individual i, HIGH SKILLED i and LOW SKILLED i are indicators of whether individual i was assigned to the respective treatment, X i is a vector of control variables, and ε i is an error term. In this specification, the average treatment effects of information provision on the outcome variables are given by the parameters α 1 and α 2. In our subgroup analysis, we investigate heterogeneous treatment effects across subgroups (defined over sociodemographic characteristics, for instance). To this end, we extend the basic regression model as follows: 10

11 SKILLED SKILLED SKILLED SKILLED (2) where Subgroup i equals one if respondent i is member of the respective subgroup, 0 else. In this specification, the treatment effects for those who are not part of the subgroup are given by α 1 and α 2 while α 4 and α 5 measure the additional effects of the information treatments on subgroup members. 3.4 Balancing test Table 1 compares respondents observable characteristics between the control group and the two treatment groups in order to test whether the randomization in our information experiment successfully balanced respondents characteristics across experimental groups. We find small, but statistically significant differences only in 5 out of 25 pairwise comparisons. This makes us confident that the randomization worked as intended. Since the number of observations in treatment HIGH SKILLED is somewhat lower than in the other two experimental groups, non-random attrition is a potential concern. In this context, nonrandom attrition would occur if respondents in treatment HIGH SKILLED were less likely to complete the survey as compared to respondents in the other experimental groups. Such non-random attrition would be a threat to the internal validity of our estimates because, if present, differences in answering behavior could be attributed to selection rather than the information provided. We tested for nonrandom attrition with a regression of the relative share of respondents who did not complete the survey on treatment indicators. The fact that treatment status cannot predict attrition makes us confident that our estimates are internally valid and that the lower number of observations in treatment HIGH SKILLED is due to pure chance. 4. Results 4.1 Descriptive evidence In this section, we present descriptive evidence on the association between respondents sociodemographic characteristics and their general attitudes towards refugees. Table 2 shows OLS regressions of answers to the questions on Screen 1 on respondent s background characteristics. 14 We restrict this analysis to the control group because their responses are uncontaminated by the information treatments. 14 We employ OLS models throughout the paper. Note, however, that (Ordered) Probit specifications yield qualitatively identical results. 11

12 4.2 The effects of information provision on specific and general attitudes towards refugees Table 3 reports OLS estimates based on Eq. (1). The dependent variables are the respondents approvals of the specific statements about refugees from Screen 2 of the survey. Column (2) of Table 3 shows that the information treatments shifted respondents perceived education level of refugees in the intended directions: Compared to the control group, the share of respondents who belief that refugees are well educated is 14 percentage points higher (5 percentage points lower) in treatment HIGH SKILLED (LOW SKILLED). Consistent with the labor market competition model, exogenously increasing refugees perceived education level increases concerns that the refugees increase competition on the labor market (for the respondent personally (see column 3) and in general (see column 13). In accordance with the fiscal burden model, treatment LOW SKILLED has a negative impact on respondents beliefs that refugees will bring more revenues than costs for the government. Apart from these economic concerns, the information treatments did not have any effects on the remaining specific attitudes towards refugees. The next question is whether these exogenous shifts in concerns regarding refugees impacts on the labor market and on the government s budget translate into general attitudes towards refugees. Table 4 presents OLS regressions based on Eq. (1) in which the dependent variables are answers to the questions regarding general attitudes towards refugees (Screen 1). Interestingly, our information treatment has only little effects on these general questions: The only significant effect of the information treatments is that treatment HIGH SKILLED decreases the share of respondents who think that Germany should permit more refugees in the future. Apart from that, all other treatment effects are insignificant. In line with previous studies on attitudes for immigration (see Hainmueller and Hiscox, 2010), it seems that economic considerations (which have been exogenously shifted by our information treatments) play only a minor role in the process which governs the formation of attitudes towards refugees. The lacking correspondence between economic concerns and general attitudes towards refugees is corroborated by answers to our question on the importance of different aspects for our respondents attitude formation towards refugees (Screen 3). Figure 1 depicts the importance attribute to each aspect. While the refugees willingness to integrate and humanitarian aspects are most important (88 percent respectively 86 percent consider them very important or somewhat important), religion/culture of refugees (important for 46 percent) and economic aspects (important for 42 percent) appear to be least important. Personal experience with refugees and their criminal behavior take intermediate positions with 70 percent respectively 53 percent considering them to be important. These findings show that concerns regarding the labor market and fiscal imbalance do not have a pronounced impact on general attitudes towards refugees because the importance of economic concerns for shaping general attitudes towards refugees is dwarfed by other aspects. 12

13 4.3 Effect heterogeneities of information treatment to be written 5. Conclusion During the past few years, Germany and other European countries have witnessed a massive influx of refugees, mostly from the Middle East. Therefore, European politicians face the important challenge to implement policies which simultaneously honor international commitments, foster the integration of those who have been granted refugee status, and are supported by domestic voters. The latter is particularly important in the light of rising voters skepticism towards refugees and asylum policies. The design of adequate integration policies is furthermore complicated by the fact that host countries know relatively little about the characteristics of the new arrivals, for instance, in terms of their education level. In this paper, we present evidence from a randomized survey experiment among almost 5,000 German university students to study the effect of refugees perceived education level on natives attitudes towards them. Respondents were randomly assigned to one of three experimental groups (one control group and two treatment groups). While the control group answered a short survey about refugees without any further information, individuals in treatment HIGH SKILLED (LOW SKILLED) were informed about a recent study which finds that refugees are rather well (poorly) educated. These information treatments significantly affected our respondents economic concerns regarding refugee immigration: Individuals in treatment HIGH SKILLED are more concerned that refugees might increase competition at the labor market while individuals in treatment LOW SKILLED are more inclined to think that the refugees bring more costs than revenues for the government. Consistent with previous studies which find that noneconomic aspects are more important than economic aspects for shaping attitudes towards immigrants, our results indicate that economic concerns do not translate into general attitudes towards refugees. This finding is corroborated by answers to our post-experimental questionnaire, in which our respondents stated that economic aspects are relatively unimportant for shaping their attitudes towards refugees (in contrast to refugees willingness to integrate or humanitarian aspects, for instance). While survey experiments are certainly subject to some artificiality, we consider this method well-suited for addressing the research questions at hand for at least two reasons. First, in order to study the causal effects of the perceived education level of refugees on natives attitudes towards them with naturally occurring data, one requires detailed measures for attitudes towards refugees as well as a truly exogenous shifter of perceived education levels. We are not aware of any data source which fulfills these two requirements. Second, Blinder and Krueger (2004) argue that public opinion surveys 13

14 are important for the political process because politicians devote tremendous resources to assess public opinions with surveys. In particularly, in the light of the current refugee crisis, much of the public and political discourse has focused on natives stated attitudes towards refugees and asylum policies. 14

15 6. References Bansak, Kirk, Jens Hainmueller, Dominik Hangartner (2016). How economic, humanitarian, and religious concerns shape European attitudes toward refugees. Science 354 (6309): Bénabou, Roland, Jean Tirole (2016). Mindful Economics: The Production, Consumption, and Value of Beliefs. Journal of Economic Perspectives 30 (3): Blinder, Alan S., Alan B. Krueger (2004). What Does the Public Know About Economic Policy, and How Does It Know It? Brookings Papers on Economic Activity (1): Buber-Ennser, Isabella, Judith Kohlenberger, Bernhard Rengs, Zakarya Al Zalak, Anne Goujon, Erich Striessnig, Michaela Potančoková, Richard Gisser, Maria Rita Testa, Wolfgang Lutz (2016). Human Capital, Values, and Attitudes of Persons Seeking Refuge in Austria in 2015.PLoS ONE 11(9): e Bundesministerium des Innern (2016). 2015: Mehr Asylanträge in Deutschland als jemals zuvor. Press release of 6 January Available at asylantraege-dezember html, accessed 5 December Bursztyn, Leonardo (2016). Poverty and the Political Economy of Public Education Spending: Evidence from Brazil. Journal of the European Economic Association 14 (5): Cruces, Guillermo, Ricardo Perez-Truglia, Martin Tetaz (2013). Biased Perceptions of Income Distribution and Preferences for Redistribution: Evidence from a Survey Experiment. Journal of Public Economics 98: Dustmann, Christian, Francesco Fasani, Tommaso Frattini, Luigi Minale, Uta Schoenberg (2016). On the Economics and Politics of Refugee Migration. CReAM DP No. 16/16. Dustmann, Christian, Kristine Vasiljeva, Anna Piil Damm (2016). Refugee Migration and Electoral Outcomes. CReAM DP No. 19/16. Elias, Julio J., Nicola Lacetera, Mario Macis (2015). Sacred Values? The Effect of Information on Attitudes toward Payments for Human Organs. American Economic Review P+P 105 (5): Eurostat (2016). Asylum in the EU Member states: Record number of over 1.2 million first time refugees registered in News release of 4 March Available at accessed 5 December Grigorieff, Alexis, Christopher Roth, Diego Ubfal (2016). Does Information Change Attitudes Towards Immigrants? Evidence from Survey Experiments. Working Paper. Hainmueller, Jens, Michael J. Hiscox (2010). Attitudes towards Highly Skilled and Low-skilled Immigration: Evidence from a Survey Experiment. American Political Science Review 104 (1): Kuziemko, Ilyana, Michael I. Norton, Emmanuel Saez, Stefanie Stantcheva (2015). How Elastic Are Preferences for Redistribution? Evidence from Randomized Survey Experiments. American Economic Review 105 (4): Lergetporer, Philipp, Guido Schwerdt, Katharina Werner, Ludger Woessmann (2016). Information and Preferences for Public Spending: Evidence from Representative Survey Experiments. CESifo Working Paper No Steinmayr, Andreas (2016). Exposure to Refugees and Voting for the Far-Right: (Unexpected) Results from Austria. IZA Discussion Paper No UNHCR (2015). Syrian Refugee Arrivals in Greece: Preliminary Questionnaire Findings. UNHCR, Genf. Woessmann, Ludger (2016). Bildung als Schlüssel zur Integration: Nur eine realistische Flüchtlingspolitik wird Erfolg haben. Ifo Schnelldienst 69(1):

16 Figures and Tables

17 Table 1: Comparison of means of socio-demographic characteristics across treatments High skilled Low skilled Control group Difference to control group Mean Survey completed Dresden (0.014) (0.014) (0.010) Munich (0.010) (0.009) (0.007) Konstanz (0.010) (0.010) (0.007) Chemnitz (0.005) (0.005) (0.003) Male (0.018) (0.017) (0.012) Age (0.128) (0.129) (0.089) Bachelor (0.016) (0.016) (0.011) Master (0.014) (0.014) (0.010) Diploma (0.015) (0.016) (0.011) PhD (0.010) (0.010) (0.007) Other study level (0.012) (0.012) (0.008) Semester (0.111) (0.111) (0.077) Born abroad (0.010) (0.009) (0.006) Both parents born in Germany (0.013) (0.012) (0.009) One parent born abroad (0.008) (0.009) (0.006) Both parents born abroad (0.010) (0.010) (0.007) No parent has college degree (0.017) (0.017) (0.012) Government aid (0.017) (0.017) (0.012) Fraction w/ lower income (0.573) (0.571) (0.395) Not encountered refugees (0.012) (0.012) (0.009) Respondents 1,604 1,629 1,668 Notes: The first two columns display the difference in means between the control group and the respective treatment group. Significance levels of Difference stem from linear regressions of the background variables on the respective treatment dummies. The final column reports unweighted means of the control group. Standard errors reported in parentheses. Significance levels: p<0.10, p<0.05, p<0.01.

18 Table 2: Correlations between perceived education level of refugees and attitudes towards refugees Germany should admit more refugees in future Number of refugees Germany admitted last year Refugees should be allowed to stay in Germany permanently Satisfied with government s asylum and refugee policy Refugees are a cultural enrichment for Germany Refugees will increase labor market competition for me personally Refugees are good for the German economy Refugees will bring more revenues than costs for government Refugees will succeed in integrating into the labor market Refugees will succeed in integrating into society Lack of language skills is obstacle for refugees labor market integration Due to government spending for refugees, I will have to pay more taxes in future Due to government spending for refugees, I will have to forgo government benefit Generally speaking, refugees are beneficial for Germany Crime rate in Germany will rise due to refugees criminal behavior Refugees will increase labor market competition in general Opinion formation: Humanitarian aspects Opinion formation: Economic aspects Opinion formation: Refugees criminal behavior Opinion formation: Religion/culture of refugees Opinion formation: Refugees willingness to integrate Opinion formation: Personal experience with refugees Notes: Correlations between perceived education level and responses to other survey questions reported. Correlations only based on control group. For description of variables and answer categories, see Section 3.1. Perceived education level has five categories, ranging from 1 (completely disagree) to 5 (completely agree). Significance levels: p<0.10, p<0.05, p<0.01.

19 Table 3: Effect of information treatment on refugees perceived education level Five-point scale Agree Disagree (1) (2) (3) (4) (5) (6) High skilled information (0.035) (0.034) (0.015) (0.015) (0.017) (0.016) Low skilled information (0.033) (0.032) (0.013) (0.012) (0.017) (0.017) Covariates No Yes No Yes No Yes Control mean Observations 4,831 4,831 4,831 4,831 4,831 4,831 Adj. R Notes: Dependent variable: Refugees are well educated on average: Columns (1)-(2): integer values from 1 to 5 (1= completely disagree, 2= somewhat disagree, 3= neither agree nor disagree, 4= somewhat agree ; 5= completely agree ); Columns (3)-(4): dummy variables (1= completely agree or somewhat agree, 0 else); Columns (3)-(4): dummy variables (1= completely disagree or somewhat disagree, 0 else). Covariates include all socio-demographic characteristics (see Table 1) and 10 binary faculty indicators. Robust standard errors reported in parentheses. Significance levels: p<0.10, p<0.05, p<0.01.

20 Table 4: Effect of information treatment on attitudes towards refugees (1) (2) (3) (4) (5) (6) (7) (8) Agree Disagree Agree Disagree Agree Disagree Agree Disagree Panel A: Labor market Competition for me Competition in general Refugees integrate successfully Language skills obstacle High skilled information (0.008) (0.013) (0.016) (0.018) (0.017) (0.016) (0.011) (0.008) Low skilled information (0.007) (0.012) (0.015) (0.018) (0.017) (0.016) (0.010) (0.008) Control mean Observations 4,843 4,843 4,840 4,840 4,844 4,844 4,850 4,850 Panel B: Economic aspects More revenues than costs Pay more taxes Less government benefits Good for economy High skilled information (0.016) (0.017) (0.015) (0.018) (0.011) (0.015) (0.017) (0.014) Low skilled information (0.016) (0.017) (0.015) (0.017) (0.011) (0.015) (0.017) (0.014) Control mean Observations 4,834 4,834 4,843 4,843 4,844 4,844 4,841 4,841 Panel C: Cultural aspects Cultural enrichment Integrate into society Beneficial for Germany Increase crime High skilled information (0.016) (0.014) (0.017) (0.016) (0.017) (0.014) (0.016) (0.017) Low skilled information (0.016) (0.014) (0.017) (0.016) (0.017) (0.014) (0.016) (0.017) Control mean Observations 4,846 4,846 4,843 4,843 4,841 4,841 4,846 4,846 Notes: Dependent variable: Columns (1), (3), (5), and (7): dummy variables (1= completely agree or somewhat agree, 0 else); Columns (2), (4), (6), and (8): dummy variables (1= completely disagree or somewhat disagree, 0 else). All regressions include the same covariates as in Table 3. For a definition of the 12 outcomes, see Section 3.1. Robust standard errors reported in parentheses. Significance levels: p<0.10, p<0.05, p<0.01.

21 Table 5: Effect of information treatment on opinion about number of refugees and asylum policy Dependent variable: Admit more refugees in future #Refugees admitted last year Allowed to stay permanently Satisfied w/ asylum policy More Less Too few Too many Favor Oppose Safisfied Unsatisfied (1) (2) (3) (4) (5) (6) (7) (8) High skilled information (0.016) (0.017) (0.014) (0.017) (0.016) (0.014) (0.014) (0.017) Low skilled information (0.016) (0.017) (0.014) (0.017) (0.016) (0.014) (0.014) (0.017) Covariates Yes Yes Yes Yes Yes Yes Yes Yes Control mean Observations 4,823 4,823 4,828 4,828 4,851 4,851 4,850 4,850 Adj. R Notes: Dependent variable: Columns (1), (3), (5), and (7): dummy variables that equal 1 for the two most positive answer categories, 0 otherwise; Columns (2), (4), (6), and (8): dummy variables that equal 1 for the two most negative answer categories, 0 otherwise; all outcomes are measured on a five-point scale. Survey questions are (in the order of the dependent variables): Cols. 1+2: "Compared to the current situation, should Germany admit more refugees, less refugees or the same number in the future?" Cols. 3+4: "What do you think about the number of refugees which Germany admitted last year?" Cols. 5+6: "Do you favor or oppose that refugees are allowed to stay in Germany permanently?" Cols. 7+8: "How satisfied are you with the government s asylum- and refugee policy?" All regressions include the same covariates as in Table 3. Robust standard errors reported in parentheses. Significance levels: p<0.10, p<0.05, p<0.01.