Chance or threat? Effects of non-citizens voting rights on natives attitudes towards immigrants

Transcription

1 Very Preliminary: Please do not quote, cite or distribute without permission of the authors Chance or threat? Effects of non-citizens voting rights on natives attitudes towards immigrants Anna Maria Koukal 1 and Marco Portmann 2 University of Fribourg, Switzerland Abstract: In a world shaped by globalization and international mobility most democratic participation rights are still tied to traditional forms of citizenship. Thus, noncitizens form the largest disenfranchised group in the contemporary society. The scare but evolving literature attributes positive effects of noncitizens voting rights to immigrants integration. Interestingly the evaluation of natives with respect to noncitizens enfranchisement is rarely discussed. We combine new data about the Swiss noncitizens enfranchisement process with individual data. This allows us to analyze differential effects of enfranchising noncitizens on natives attitudes towards foreigners and democracy. Applying a difference in difference approach we find that natives seem to be more reluctant towards additional migration but show a tendency to reduce negative attitudes towards present noncitizens. Moreover, our results suggest a stronger satisfaction with democracy and a positive effect on the attitude towards noncitizens of those natives who live in municipalities which initially opposed the suffrage extensions. Keywords: noncitizens voting rights, attitudes towards foreigners, democracy satisfaction JEL classification: D02, D72, J15 1 Université de Fribourg, Boulevard de Pérolles 90, 1700 Fribourg, Switzerland. annamaria.koukal@unifr.ch 2 Université de Fribourg, Boulevard de Pérolles 90, 1700 Fribourg and CREMA, Switzerland. marco.portmann@unifr.ch 1

2 1. Introduction Suffrage is a key instrument for the legitimacy of democratic systems. Claims such as women s voice matter and no taxation without representation reflect the notion that only those who possesses a vote in the political process are genuinely able to express their preferences at the polls. While globalization and digitalization have led to an ever-increasing mobility of people, in contrast, most democratic participation rights are still tied to conventional models of citizenship. Consequently, migrants are today s largest disenfranchised group. According to Aleinikoff (2013) starting in the late 1970s, the granting of local voting rights for noncitizens has been slightly emerging. Since 1993 European citizens have local voting rights in any EU member state, with Northern European states in the lead. For instance, Sweden grants non-discriminating local voting rights to migrants after three years of residence (for a detailed description of noncitizen suffrage across Europe see Earnest, 2003, Groenendijk, 2008 or Aleinikoff, 2013). Increasing international migration, and, thus, increasing numbers of politically excluded residents bring about a growing urge to gain empirical insights on the effects of decoupling political participation rights from citizenship. Policy makers are interested in the consequences of enfranchising noncitizens, since those are said to influence key areas of integration such as civic virtue, satisfaction with democracy, trust or attitudes towards foreigners. However, to the best of our knowledge, so far the empirical literature has focused on the effects on enfranchised noncitizens themselves (e.g. Tillie, 1998, Vernby, 2013, Slotwinski et al., 2017). We contribute to the literature on noncitizens enfranchisement by assessing the effects on natives regarding their attitudes towards foreigners and their evaluation of democracy. Specifically, we contribute by exploring noncitizens voting rights in Switzerland. In contrast to most other countries in which voting rights are installed via parliaments, in direct democratic Switzerland, foreigner s suffrage is granted by the native electorate through cantonal and municipal initiatives. Thus, the design of voting rights for noncitizens differs across cantons and suffrage is installed at different points in time. We take advantage of three different datasets, allowing us to combine individual data from the Swiss Household Panel (SHP) with municipal information about the enfranchisement process and sociodemographic characteristics. Using a difference-in-difference approach our empirical analysis evaluates how the enfranchisement of noncitizens at the local level affects 2

3 natives attitudes in two important dimensions. Namely, their attitudes towards foreigners and their evaluation of democracy. In the empirical analysis we consider two different perspectives of natives reactions to noncitizens voting rights. In a first step, we evaluate the voting behavior of municipalities in national referenda about topics connected to foreigners, whereas in a second approach we analyze individual data. Our preliminary results indicate that natives tend to be less open towards additional immigration as their skepticism against the EU and additional immigration increases when they are exposed to enfranchisement of noncitizens at local level. We find that suffrage extensions bring about limited changes in natives attitudes towards current immigrants on average. However, focusing subpopulation, we find that natives with low education who tend to oppose a suffrage extension reduce their bias against foreigners, while highly educated foreigners opposing an extension rather increase their bias against foreigners. In addition, natives experience an increased satisfaction with democracy and political influence when compared to individuals who are not exposed to noncitizens voting rights. This paper is organized as follows. The next section briefly reviews the related literature. In section 3 we provide a short overview over the Swiss institutional setting. Section 4 describes the structure of our dataset. Section 5 introduces our estimation strategy and hypotheses. Section 6 presents and discusses our first result. In section 7 we summarize our main results, draw some conclusions and provide an outlook. 2. Literature Review This paper contributes to the astonishing scarce literature on the effects of enfranchising noncitizens. One potential channel of noncitizens voting rights impacting integration arises from immigrants themselves. The few empirical papers on the effects on noncitizens voting rights address this channel. Similar to the literature about the relationship between enfranchising women and the size and structure of government spending 3, Vernby (2013) analyzes the effect of suffrage extension to noncitizens on public policy. Drawing on a sample of 183 Swedish municipalities, Vernby (2013) finds a positive effect of enfranchising 3 See for example Abrams and Settle, (1999), Lott and Kenny, (1999), Aidt et al., (2006). 3

4 noncitizens on public policy spending regarding education, family and social services. Slotwinski et al. (2017) exploit an institutional regulation in Denmark granting suffrage to immigrants automatically after three years. Using registry data, they evaluate the effect of granting suffrage on individual compliance with the law. They find that the first possibility to vote enhances male immigrant s compliance with the law. Tillie (1998) adds to the literature by analyzes the party choices of foreigners in Dutch municipalities. Working with survey data he finds that foreigners stick to the political discourse of a left-right dimension and that in principal all parties are eligible for foreigners. In addition to the scarce empirical contributions on the consequences of foreigner s suffrage, some papers analyze the voting behavior of former noncitizens after their naturalization (e.g. Wüst, 2003; Savodelli, 2006; Kroh and Tucci, 2009). These contributions based on surveys conclude that the former nationality is a key predictor of party choice and that former foreigners tend to prefer larger parties to smaller ones. Compared to the literature on the effects of enfranchising noncitizens on attitudes and democracy satisfaction, the literature connected to labor market outcomes is much richer. The literature provides mixed results about the so-called competition hypotheses. This literature explores the question to what extent worries and uncertainties about the effects of immigration on employment and wages shape anti-immigrant attitudes. While some authors find evidence in favor of the competition hypothesis having a bolstering effect on antiimmigrant attitudes (e.g. Scheve and Slaughter, 2001 or Mayda, 2006) other contributions question this result (e.g. Sides and Citrin, 2007 or Hainmueller et al., 2015). Despite the sizeable literature regarding competition in the labor market, so far little is known about competition in the political market and potential effects on political outcomes and attitudes. Without explicitly linking to the question of immigrants voting rights, several contributions highlight that participatory democratic institutions turn out to foster cooperation (Acemoglu and Robinson, 2012), trust (e.g. Rainer and Siedler, 2009) and civic virtue (Frey, 1997). According to Frey (1997) the boost effect of participation on civic virtue occurs, when extensive participation rights are seen as an act of trust. Moreover Frey (1997) also emphasizes the possibility of a crowding in of civic duty, if a person is able to take part in the decision making process. Likewise democracy and especially democratic participation seem to foster individuals sense of self-determination, which results in a positive procedural utility (see Frey et al., 2004). Regarding the effect of extending suffrage to foreigners, the role of 4

5 political discourse and its effect as a discovering mechanism (Hayek, 1968) for the political market (Frey and Bohnet, 1993) may also play an important role when thinking about the effects of noncitizens voting rights on natives. From this literature review, we conclude that despite the extensive literature on native s attitudes towards foreigners connected to the labor market, to the best of our knowledge, little is known about the effects of noncitizens voting rights on natives. Most strikingly, little is known about the role of competition concerning the political market and the satisfaction with democracy. This paper aims to contribute to these gaps. 3. Swiss Institutional Setting In most countries which introduced foreigner s suffrage, national parliaments enacted the enfranchisement of noncitizens. In highly decentralized and direct-democratic Switzerland, the decision has to be made by the citizens themselves at the ballot boxes. So far this resulted in approximately 40 votes about enfranchising noncitizens in 16 out of 26 cantons (sub national states) 4. However, voting rights for foreigners neither exist nor were subject to public debate at the federal level as of yet. The power to extend the franchise for cantonal and municipal votes and elections lies with the cantonal voters. Hence, voters in each canton decide separately, whether they extend the franchise to noncitizens. This results in a considerable number of votes on foreigner s enfranchisement as can be seen from Table 1. The table lists all cantons in which suffrage has been extended to noncitizens. The cantons autonomy on the matter also resulted in a variety of configurations with respect to which rights exactly they grant to noncitizens. For instance, some cantons decided to grant active, passive or both voting rights (denoted as full in the table), or they may grant them contingent on the duration of stay. In some cantons, voters have decided to enfranchise noncitizens either for cantonal or municipal matters only. Hence, in some cases, the majority of cantonal voters has decided to impose an enfranchisement of foreigners at the municipal level, even though majorities of voters in some municipalities within the canton have opposed the suffrage extension. To avoid such conflicts, three cantons 4 The respective cantons are: Aargau, Appenzell Ausserrhoden, Bern, Basel-City, Fribourg, Geneva, Grison, Glarus, Jura, Lucerne, Neuchâtel, Schaffhausen, Solothurn, Uri, Vaud, Zurich 5

6 adopted opt-in rules for municipalities. In those cantons the majority of cantonal voters decided to concede the right to extend the franchise to noncitizens at municipalities within the cantons. Eventually in those three cantons, voter majorities in 26 municipalities decided to extend the franchise for municipal matters to noncitizens. Most voting rights concern the municipal level, though the canton of Jura and Neuchâtel also offer voting rights at the cantonal level. Table 1: Overview: Suffrage Introductions at the Cantonal Level Vote date Acceptance Effective date Canton Suffrage type Enforcement Municipalities w/ enforcement % Jura active local; active cantonal enforced all Cantonal per Appenzell full local conceded to 4 assembly municipality A.R. municipalities % Neuchâtel active cantonal enforced all % Vaud full local enforced all % per municipality Grison full local conceded to municipalities % Fribourg full local enforced all % per municipality Basel-City full local conceded to municipalities % Geneva active cantonal enforced all % Neuchâtel passive local enforced all % Jura passive local enforced all Sources: Cantonal chancelleries, cantonal constitutions, Alder et al. (2015). 4. Data To answer our research questions, we need rich information about voting rights in municipalities and cantons and individuals attitudes. Therefore, we rely on three different datasets: (1) a full dataset covering the results of votes on enfranchising noncitizens at the municipality level, (2) a dataset covering results of national votes on issues connected to 6

7 foreigners and (3) individual data on characteristics and attitudes from the Swiss Household Panel (SHP) Referenda on suffrage extensions to noncitizens We use a novel, self-collected dataset about the enfranchisement decisions. Enfranchising noncitizens lies in the power of cantonal and municipal voters and authorities and so does the conduction and subsequent archiving of the respective referenda. Therefore, all data on referenda were collected and digitalized from cantonal archives, because most parts of the data were only available as paper printouts. A full list of votes on the introduction of noncitizens voting rights is given in Appendix Table A. Concerning cantons and language regions the voting procedure is different across Switzerland. The French speaking cantons are pioneers in terms of noncitizens voting rights with two cantons even granting suffrage at the cantonal level. However, vote activities on enfranchising noncitizens are spread across Switzerland. Up to now, votes were conducted in 16 cantons. In the German-speaking part of Switzerland three cantons adopted opt-in rule. So far 26 municipalities in two cantons installed local voting rights on the municipality level. The actual implementation of noncitizens voting rights takes various forms: full local rights, active rights only, cantonal active only. The majority of cantons only grants local voting rights to noncitizens. However as a survey conducted by Adler et al. (2015) indicates noncitizens are visible in the political process. An example is the canton of Vaud in which noncitizens can also be elected at the local level (passive voting rights). Out of 165 municipalities taking part in the survey 115 stated to currently have active noncitizen parliamentarians in their municipality Adler et al. (2015, p. 24). As Appendix Table A illustrates, the vote types are different. A part of the votes on enfranchising noncitizens is integrated in constitutional revisions. For the future, we plan to gain a deeper understanding of the other elements of those constitutional revisions to make sure that the effects we find derive from the franchise extension and not other aspects of the constitutional revisions National Votes Connected to Immigrants As we focus on the effect of noncitizens voting rights on natives preferences and attitudes toward them, a challenge of the paper is to find suitable measures of attitudes. Due to Switzerland s strong direct democracy, citizens frequently reveal their preferences on a variety of topics in referenda and numerous votes concerning laws on immigration. Therefore, our first approach is to proxy openness towards immigrants using votes. 7

8 While referenda take place at the local, cantonal and federal level, we specifically use federal votes as proxies of attitudes towards noncitizens. Since referendum results at the federal level are available disaggregated for municipalities, this source offers comparable, nation-wide information about attitudes, that can be matched with our dataset on foreigner voting rights at the municipal level. Moreover, foreigners are as of yet not entitled to vote at the federal level, and hence, the preferences we measure are those of citizens only and are not permeated with noncitizens who have gained the right to vote. Usually at three to four ballot meetings per year the Swiss electorate is invited to decide about initiatives, referenda or counterproposals on the federal level. We construct a dataset covering those votes connected to immigrants in the 1990 to 2016 period. Appendix Table B provides an overview of the full list of 23 5 votes. The dataset consists of the respective results on the municipality level. To control for various municipal characteristics, we can either take advantage of the panel structure employing municipality fixed effects or we draw on federal census data. In addition to the census data we can also rely on measures provided by Ladner who offers information about the institutional settings of municipalities starting in For a further analysis this might provide additional insights with respect to the influence of direct democracy on attitudes towards immigrants. In order to proxy the effect of noncitizens voting rights on natives attitudes towards foreigners we can either take the full list of votes connected to foreigners or analyze votes that are repeatedly voted upon. A main concern of our approach might be that votes are different and difficult to compare. There are at least four domains of votes that allow us to observe similar votes before and after the introduction of noncitizens voting rights to control for this concern. The following categories are: (1) votes about the EU ( ) with 5 votes in the 2000 to 2009 period, (2) votes concerning regulation of additional immigration ( ) with 3 votes in the 1996 to 2016 period, (3) votes connected to refugees and (4) with 6 votes in the 1999 to 2016 period, votes about naturalization 6 ( ) with 3 votes in the 1994 to 2008 period. The allocation to domains is listed in Appendix Table B. In order to make the referenda data comparable the municipal referenda result is transformed in such a way, that a higher yes share corresponds to a higher degree of openness towards the immigrant related 5 In the current version of the paper the results of two votes are not integrated due to a lack of data. 6 In this preliminary version of the paper two votes concerning naturalization are missing in our data set. In the future we will integrate those in the analysis. The two missing votes are took place in 2004 and 2008 and are listed in the appendix. 8

9 question. This allows us to observe voting behavior of municipalities over the course of time. Consider as an example the initiative against the construction of minarets which took place in November In this case a higher yes share indicates a preference against the construction of minarets. Therefore, we change the sign, such that a share of 100 corresponds to full openness towards foreigners while a share of 0 corresponds to complete isolation. Other interesting outcome variables at the municipality level are the number of naturalizations and turnout of the native population, which we plan to consider in a later version of this paper Individual Perspective: Swiss Household Panel Our second approach to measure citizens attitudes towards foreigners is to draw on individual survey data. The Swiss Household Panel (SHP) provides detailed information about individuals, information on their attitudes towards foreigners, their evaluation of democracy and a rich set of individual background characteristics. As the SHP is constructed as an annual panel questionnaire, it allows us to exploit the variation over the observed period on the individual level. The SHP data collection started in 1999 with 13,000 interviews in approximately 5,000 households. When using the sample weights, as we do throughout all regressions that draw on SHP data, the panel is representative of the whole Swiss population. Since we analyze the effects of voting rights on citizens and the legal minimum voting age for voting is 18 in Switzerland, we restrict the sample to all Swiss citizens 18 or older. For the moment, we restrict the analysis to SHP questions that were posed in each wave. Due to missing observations, we end up with around 50,000 observations. The variables of interest for our analysis can be divided into two groups: attitudes towards foreigners and evaluation of democracy. To capture the first, we employ the variables and. measures on a scale from 1 to 3 whether natives should have the same chances as foreigners in life ( ) or whether natives should be treated preferentially ( ). measures on a scale from 1 to 3 whether Switzerland should become a member of the European Union ( ) or should not become a member of the European Union ( ). For the second dimension of attitudes, we employ and. The first question asks participants how satisfied they are with the way democracy works in Switzerland, while the second question asks about participants perceived influence in the political process. Both variables range from zero (unsatisfied/no influence) to 10 (completely satisfied/very strong influence). We know the municipality of residence for all households in the survey and are therefore able to merge 9

10 data on attitudes with municipality characteristics and in particular information about foreigner voting rights in the municipality of residence. Summary statistics for all variables from the SHP are given in the Appendix Table C. 5. Empirical Strategy and hypotheses 5.1. Basic approach To measure the effects of noncitizens voting rights on natives we pursuit a difference-indifference estimation approach. We compare how attitudes of individuals and municipalities change once municipalities extend the franchise to noncitizens. As a control group we use those municipalities and individuals who are not exposed to foreigner suffrage at municipal level. As outlined in section 4, we have two approaches, either to use municipalities or individuals as units of observation. These two approaches yield the regression equations that take the following forms: In equation (1), the units of observation are the voting results of municipality in federal referendum. For each municipality and referendum either takes the value 0 if the respective municipality does not have suffrage for noncitizens at the time of voting and 1 otherwise, i.e., if noncitizens have some voting rights. In most specifications we include municipality fixed effects and time fixed effects. In some cantons and municipalities additional votes on the extension of the franchise to foreigners take place. For instance, they may reject to grant foreigners the right to vote at the cantonal level or there is an accepted vote to concede the right to introduce voting rights for noncitizens at the municipal level. Thus, we use a series of control variables for those votes, denoted by 7 in some specifications. The coefficient is the coefficient of interest and corresponds to differential effect of the enfranchisement of noncitizens on citizens living in 7 We use separate dummy variables accepted and rejected suffrage extensions at the cantonal level and cantonal concessions for municipalities to decide about the introduction of voting rights for foreigners. Detailed results for these variables are available upon request. 10

11 municipalities which have extended the franchise. Note that our setting differs from the classical difference-in-difference 8 setting where additionally treatment group and after treatment variables are required. In our case this is not necessary as we employ municipality and time fixed effects. Moreover, our setting gains strength, as municipalities enfranchise foreigners staggered, and thus, issues of spurious correlations are reduced. Moreover, our setting gains strength, as municipalities enfranchise foreigners staggered, and thus, issues of spurious correlations are reduced. In equation (2), the units of observation are the responses of SHP participant on an attitude question in survey year. Analogously to equation (1) and the analysis based on municipal data, denotes whether individual lives in a municipality with a municipal noncitizen suffrage in force. Again, we take advantage of the panel structure of our dataset and employ individual and time fixed effects and control for other votes that affect suffrage for foreigners. Compared to our first approach based on municipal data, we can thereby analyze within individual changes and exclude ecological fallacy issue, i.e., the problem that municipal voting results in federal referenda may change due to compositional changes of those voters who decide to turn out Hypotheses As we estimate the effect of noncitizens suffrage on different attitudes measures, the meaning of changes with the dependent variable employed. With regard to the municipal dataset may indicate that enfranchising foreigners reduces the in-group toward the enfranchised group. Regarding the literature on the contact hypothesis (see for example Amir, 1998), the political discourse and contact with the newly enfranchised group may reduce information asymmetries, increase trust, reduce prejudices and therefore shape natives attitudes towards greater openness towards foreigners. 8 In the future we plan to analyze the votes conducted about easing the naturalization process. Moreover the data structure allows estimating a difference in difference approach in a traditional manner, as we are able to identify if a municipality agreed to introduce noncitizens suffrage or not. Therefore we have 5 groups of municipalities in our dataset: (1) municipalities that never voted about enfranchising noncitizens, (2) municipalities that accepted suffrage and got it, (3) municipalities accepting without getting it, (4) municipalities rejecting suffrage but getting it and (5) municipalities rejecting suffrage and not receiving suffrage. Going forward with the project we will further elaborate on these possibilities. 11

12 Conceptional distinction between different groups of foreigners Especially with regard to the categories or additional immigration, could be interpreted as a signal of feared enhanced political competition. In particular, with respect to additional migration the increased competition in the political market could lead to a conceptional distinction between noncitizens already present and future migration to Switzerland. Noncitizens belonging to the resident population could be regarded as being closer to the own group (reduction of in group bias), whereas the price of additional migration will increase as they now have the potential to use the option of voice in the political process. Drawing from these considerations we would expect if attitudes towards additional immigration are concerned and in the case of attitudes towards noncitizens already present 9 in the country. Analogous to the consideration made with respect to the municipal dataset we expect enfranchisement to have a negative effect on the individual measure as this question proxies attitudes towards additional immigration. Procedural and instrumental utility of democratic involvement In the spirit of the literature showing participatory institutions fostering cooperation, trust and compliance with the law (Acemoglu and Robinson, 2012 and Slotwinski et al., 2017), we would expect for. All the more since noncitizens have a say in politics as soon as they are enfranchisement. Consequently, incentives for natives to treat them as equal partners should increase. Either because they fear their vote or because prejudices are reduced through additional contact opportunities in the political sphere (Amir, 1998). With respect to, the theory connected to procedural utility of democratic participation (Frey et al., 2004) we would expect. The enfranchisement of a so far excluded group could enhance the political debate and therefore enrich the political market with additional ideas. Moreover, we would expect as the enfranchisement brings about an increase of the institutional congruence. The inclusion of the so far excluded group should bring about additional information in the democratic process, and thus leads to better results. 9 This will be elaborated regarding the votes on naturalization of noncitizens. 12

13 Since many suffrage extensions for foreigners are decide in cantonal votes, we can also identify those municipalities, in which a majority of voters rejected the extension. Therefore, we also estimate models according to equation (4), which includes. This variable takes the value zero, if individual I lives in a municipality that has been outvoted and zero otherwise. Thus, captures the differential effect for outvoted municipalities. There are at least two reasons why we expect coefficient. First, external interventions have, at least in the short term, the potential to crowd out intrinsic motivation (see Frey, 1997). A second possibility is that municipalities opposing a suffrage extension may differ in their sociodemographic characteristics and expect voting rights for foreigners to have negative consequences. For the moment we cannot disentangle these two explanations but plan to do so in the future. Competition on the political market Taking into account the literature connected to we would expect as natives are sharing their political influence with additional voters (see Koukal and Eichenberger, 2017). If sharing voting rights is perceived as costly in the sense of, we would expect the effect to be strongest for individuals living in municipalities with a high share of noncitizens. As we are measuring the perceived political influence of individuals could also indicate that own individual influence is now regarded as relatively high as the native population has access to all levels of democratic involvement, whereas noncitizens are restricted to specific areas. could therefore reflect some sort of benchmarking process. Applying the hypothesis of Meltzer and Richard (1981) to suffrage extensions, enfranchising new groups could shift the pivotal voter to another income bracket. In other words, the perception of political influence could be related to the sociodemographic characteristics of the enfranchised noncitizens. At the individual level Sides and Citrin (2007) mention various important predictors of attitudes towards foreigners such as economic interests and information about immigration. Therefore, we consider heterogeneous effects across sociodemographic, municipal or institutional dimensions. 13

14 6. Results 6.1. Results for the municipal level Table 2 maps a first descriptive approach to the data and suggests that local suffrage seems to foster openness. The mean of openness is larger in those samples in which noncitizens are already enfranchised. This is true for different samples ranging from the full sample of municipalities and votes to those municipalities enfranchising noncitizens over the observed period. However, it is important to note that this picture is still shaped by municipality characteristics, vote effects or time effects. Table 2 also provides a glance of the rich variation that our dataset offers. For example, we can distinguish voting behavior of municipalities that agreed to the rule from those which were outvoted in the respective enfranchisement vote. In future version of the paper, we will further explore these differential effects. For the moment, we show first results on outvoted municipalities for regressions based on individual level data only. Table 2: Mean Comparison for the Dependent Variable Openness 10 Sample Local enfranchisement N Mean Max Min Sd yes Full sample no Recent efranchisements Treated Outvoted treated Accepted treated EU Immigration yes no yes no yes no yes no yes no yes no Regarding our estimations we first look at the full sample of Swiss municipalities. The respective results are shown in panel A of Table 3. In panel B we reduce the sample to recent 10 In the full sample all cantons are considered, recent enfranchisements is defined as cantons being observed before and after the enfranchisement, treated refers to municipalities enfranchising noncitizens in the period of observation, outvoted refers to municipalities rejecting the enfranchisement but being overruled by the canton. 14

15 enfranchisement allowing us to observe municipalities before and after the introduction. Therefore we exclude Jura and Neuchâtel, which introduced noncitizens voting rights already before our analyzed period. In both panels we estimate three specifications in which either the subsamples on votes about the EU, immigration or the full sample of votes on topics related to migrations are used. All results are estimated using municipality fixed effects, vote fixed effects and year fixed effects. Additionally, we estimate a model with and without control variables. To consider urbanization we add population in logs, the share of agriculture and the share of foreigners, whereas the share of employees takes into account the evolution of the economic environment. Finally, we control for the share of married persons in a municipality. Quite contrary to the descriptive picture presented in Table 2, Table 3 suggests that enfranchising noncitizens on the local level seems to negatively impact openness of natives towards foreigners regarding openness when considering EU related topics, additional immigration and general openness. Enfranchising noncitizens on the local level seems to reduce openness with respect to EU-related questions by 3 percentage points when compared to municipalities without noncitizens enfranchisement. This effect seems substantial especially when considering tight vote outcomes. 15

16 Table 3: Effects of enfranchising noncitizens on citizens evidence from municipal data A: Full sample Dependent variable: openness Subsample EU Immigration All votes Local Enfranchisement *** *** *** *** *** *** (0.309) (0.323) (0.343) (0.379) (0.214) (0.238) Control Variables yes yes yes Municipality fixed effects yes yes yes yes yes yes Vote fixed effects yes yes yes yes yes yes Year fixed effects yes yes yes yes yes yes n. Obs. 10,945 10,945 6,510 6,510 46,160 46,160 R B. Cantons with recent enfranchisements and control group Dependent variable:openness subsample EU Immigration All votes Local Enfranchisement *** *** *** *** *** *** (0.309) (0.323) (0.345) (0.379) (0.211) (0.236) Control Variables yes yes yes Municipality fixed effects yes yes yes yes yes yes Vote fixed effects yes yes yes yes yes yes Year fixed effects yes yes yes yes yes yes n. Obs. 10,530 10,530 6,261 6,261 44,363 44,363 R Notes: Robust standard errors in parentheses clustered at the municipality level:*** p<0.01, ** p<0.05, * p<0.1.control variables: population in logs, foreigners (share), married (share), employed (share), agricultural (share). 16

17 Table 4: Effects of enfranchising noncitizens on all citizens evidence from individual data A. Full sample Dependent variable NativePriority AntiEUAttitude DemocracySatisfaction PoliticalInfluence LocalEnfranchisement (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (0.0213) (0.0214) (0.0231) (0.0505) *** (0.0211) *** (0.0212) ** (0.0232) (0.0505) * (0.0467) ** (0.0470) ** (0.0514) (0.1106) ** (0.0631) *** (0.0633) LocalEnfrOutvoted Controls for other votes no yes yes no yes yes no yes yes no yes yes Wave fixed effects yes yes yes yes yes yes yes yes yes yes yes yes Person fixed effects yes yes yes yes yes yes yes yes yes yes yes yes Month fixed effects yes yes yes yes yes yes yes yes yes yes yes yes n. Obs *** (0.0701) R B. Cantons with recent enfranchisements and control group Dependent variable NativePriority AntiEUAttitude DemocracySatisfaction PoliticalInfluence LocalEnfranchisement ** (0.1308) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (0.0222) (0.0222) (0.0237) (0.0524) *** (0.0220) *** (0.0221) ** (0.0239) (0.0518) ** (0.0482) ** (0.0485) ** (0.0528) * (0.1130) ** (0.0647) ** (0.0650) LocalEnfrOutvoted Controls for other votes no yes yes no yes yes no yes yes no yes yes Wave fixed effects yes yes yes yes yes yes yes yes yes yes yes yes Person fixed effects yes yes yes yes yes yes yes yes yes yes yes yes Month fixed effects yes yes yes yes yes yes yes yes yes yes yes yes n. Obs *** (0.0715) R Notes: Robust standard errors in parentheses clustered at the individual level:*** p<0.01, ** p<0.05, * p< * (0.1304) 17

18 With around 2 percentage points, the effect is slightly smaller with regard to votes connected to additional migration. Both effects are highly significant at the one percent level and stay robust in the estimated specifications. This result hints in the direction of a conceptional distinction between noncitizens already present in the country and additional migration. Moreover, this result is a first indication that granting noncitizens voting rights is not automatically associated with an increased openness towards immigrants. Additional insights concerning this hypothesis will be gained when analyzing questions regarding the regulation of naturalization affecting a group of noncitizens already present in the country Results for the individual level First results based on municipal data showed that noncitizens voting rights seem to reduce openness concerning additional migration, EU matters and general openness. We now advance to the analysis based on individual survey data. Individual survey data do not only allow to better investigate heterogeneity of potential effects. Compared to data aggregated on municipal data, we also rule out the potential problem that results may be driven by compositional effects because either citizens move between municipalities or because different voters turn out for the analyzed referenda. All voters Table 4, panel A presents results for all individuals and the period of 1999 to 2016, which constitutes the complete sample of the Swiss Household Panel. We investigate the effect of the municipal enfranchisement of noncitizens on citizens with respect to two relevant measures of attitudes towards foreigners and two measures on attitudes towards the political system. In all specifications we run OLS regressions. We use the corresponding sample weights provided by the SHP and calculate robust standard errors clustered for individuals, as we observe each individual multiple times 11. While all specifications include fixed effects for individuals, survey waves and months we present three specifications differing in the number of controls related to votes on enfranchisements. In specifications (1), (3), (7), and (10) we do not include further voting variables apart from. All other specifications include dummy variables signifying that accepted or rejected referenda on the 11 We performed robustness checks with different clustering approaches, in which standard errors essentially remained in the same ballpark is in the specifications shown in the paper. 18

19 introduction of suffrage extensions to foreigners for cantonal matters took place or if cantons voted about conceding the right to decide about enfranchisements to municipalities. Finally, in specifications (3), (5), (9), and (12) we include which captures the differential effect for those individuals living in municipalities in which a majority of voters opposed the introduction of al local suffrage for foreigners 12. Turning first to, we observe that individuals tend to be less in favor of supporting the view that natives should be treated preferentially after noncitizens have been locally enfranchised. However, these effects are not significant at any conventional levels. Contrastingly, for the second variable on attitudes ( ) we find that suffrage extension is associated with a more opposing view on Switzerland becoming a part of the European Union. Given that the variable is measured on a scale from one two three, these changes of 0.5 to 0.6 are substantial. Advancing to those variables measuring attitudes concerning the working of the political system, we see that satisfaction with the working of democracy in the country ( ), as well as individuals perceived political influence ( ) increase. Both effects are statistically significant in all specifications. Since a majority of voters has voluntarily decided to extend the franchise, these findings are not completely surprising for the overall population of citizens. While the coefficient that captures the differential effect on for municipalities in which a majority of voters opposed the suffrage extension is negative, the effect is not robust. However, individuals from outvoted municipalities exhibit a significantly negative effect on their perceived political influence after locally enfranchising foreigners when compared to those from municipalities which supported the law. Since we include time fixed effects and month fixed effects and the suffrage extensions take place staggered, our findings should not pick up time trends. However, since a large fraction of observations for individuals living with noncitizens municipal suffrage originate from the Cantons of Jura and Neuchâtel and since those cantons introduced local voting rights for foreigners before the period of observation, we cannot exploit within individual variation for 12 Note, that we cannot measure the individual voting behavior. 19

20 these cantons. Therefore, in panel B we exclude those individuals from the sample. Essentially, we find that panel B confirms the results from panel A. While there are mixed results with respect to suffrage extensions effects on attitudes towards foreigners (no or rather negative effects on ) and positive effects on, panel B indicates again positive effects on and. In contrast to the whole sample however, we find statistically significant results for both variables that individuals living in outvoted municipalities express a lower satisfaction with democracy, as well as feeling less influential in politics. In sum, the results connected to confirm the picture arising from 6.1. As discussed in section 5 this could indicate that the out-group definition changes in such a way that additional immigrants are perceived as additional competition. Although not significant, the opposite direction seems to work with respect to noncitizens already present in the country and affected by the new regulation. Regarding the results for the results are first evidence for the potential of a crowding in effect, especially if the regulation was self chosen. Note again, that for changes in attitudes towards the working of the political system, we cannot rule out that other aspects of the referendum or general effects of referenda are driving the result. However, those concerns might be less of a problem when taking a comparative view and looking at subsamples of voters. 20

21 Table 5: Effects of enfranchising noncitizens by educational attainment A. Low education Dependent variable NativePriority AntiEUAttitude DemocracySatisfaction PoliticalInfluence LocalEnfranchisement LocalEnfrOutvoted (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (0.0674) -3.4e-05 (0.0673) (0.0738) ** (0.1370) * (0.0520) * (0.0524) * (0.0571) (0.1308) (0.1405) (0.1414) (0.1491) (0.3501) (0.1718) (0.1730) Controls for other votes no yes yes no yes yes no yes yes no yes yes Wave fixed effects yes yes yes yes yes yes yes yes yes yes yes yes Person fixed effects yes yes yes yes yes yes yes yes yes yes yes yes Month fixed effects yes yes yes yes yes yes yes yes yes yes yes yes n. Obs R B. High education Dependent variable NativePriority AntiEUAttitude DemocracySatisfaction PoliticalInfluence LocalEnfranchisement (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (0.0388) (0.0390) (0.0379) (0.0391) (0.0393) (0.0426) ** (0.0824) ** (0.0832) *** (0.0882) * (0.1011) * (0.1015) (0.1845) (0.3698) (0.1075) * ** LocalEnfrOutvoted (0.1004) (0.0917) (0.2065) (0.2536) Controls for other votes no yes yes no yes yes no yes yes no yes yes Wave fixed effects yes yes yes yes yes yes yes yes yes yes yes yes Person fixed effects yes yes yes yes yes yes yes yes yes yes yes yes Month fixed effects yes yes yes yes yes yes yes yes yes yes yes yes n. Obs R Notes: Robust standard errors in parentheses clustered at the individual level:*** p<0.01, ** p<0.05, * p<

22 Educational attainment Swiss institutions require direct-democratic legitimacy of suffrage extensions. Hence, average effects for the whole Swiss population as shown in Table 4 may not show negative changes in natives attitudes as long as the consequences of their decisions turn out to meet their prior expectations. However, as discussed in Section 2, effects may differ substantially for subpopulations. As Scheve and Slaughter (2001) and Sides and Citrin (2007) point out, education turns out to be an important factor in shaping attitudes towards immigrants in a labor market context. At the same time Matsusaka (1995) and Hessami (2016) highlight education as an important factor shaping both democratic participation and processing information about the democratic process. In Table 5 we study the effects of enfranchising noncitizens on citizens by educational attainment. Panel A. shows results for natives with low education (defined as ISCED 1,2, 3c), whereas Panel B presents results for highly educated natives (defined as ISCED 5, 6). Analogously to Table 4, in panel B we restrict the sample to recent suffrage extensions, i.e., drop individuals who are exposed to municipal suffrage for noncitizens but cannot be observed before the introduction of suffrage for foreigners. With respect to preferential treatment for natives ( ) we observe that specifications (1) and (2) do not hint at changes in attitudes of low educated citizens after enfranchising foreigners. However, those from outvoted municipalities exhibit a statistically significant reduction, implying a change towards more favorable treatment of foreigners. At a first glance this result seems counterintuitive. However, this result supports the idea of contact via political institutions being a channel of reducing prejudices. However, since education can be seen as a proxy for economic status, this results could also point in the direction of Hainmueller and Hopkins (2014) who point out, that negative effect of economic factors on attitudes towards foreigners are overestimated. At the same time, low educated citizens became more hostile to the idea of Switzerland joining the EU after suffrage extensions, confirming the results of most other EU specifications. While all coefficients for satisfaction with democracy and political influence show a negative sign, none of them is statistically significant. Comparing panel A to B reveals differential effects for high educated citizens. Even though high educated do not systematically change their attitudes on prioritizing natives over 22

23 foreigners, high educated from outvoted municipalities react with an increasing bias against foreigners, as can be seen from specification (3). Conversely, natives with high education do not increase their rejection of a Swiss accession to the EU. While highly educated persons generally increase their satisfaction with democracy and fell more influential in the political process after municipal enfranchisements of foreigners, those from outvoted municipalities also show an increasing dissatisfaction with democracy as specification (9) shows. These results suggest that citizens with high education react stronger to political issues when they belong to the winner of a vote, as well as when they lose a vote. This result may be driven by higher turnout among high educated and their higher knowledge about politics. It is unclear to what extent this is a general feature of referenda or specific to referenda on suffrage extensions. Share of foreigners in the municipality Swiss municipalities greatly differ in their shares of foreigners. The local share of foreigners may mediate the effect of enfranchising noncitizens on citizens attitudes in multiple ways. While contact with foreigners, knowledge about them and prejudices may vary, the impact on local policy outcomes may also vary with the size of the newly enfranchised group. Therefore, we investigate in Table 6 differential effects for municipalities with either a low or high share of foreign population. We focus on the two subsamples with a share of less than 12.3 % and more than 27.8 % of foreigner in the municipalities, which corresponds to the bottom and top quartile of observations within the analyzed dataset. As before, we restrict the analysis on recently decided franchise extensions, enabling us to study within individual changes over time. For natives living in municipality with a small foreign population (panel A), we observe no statistically significant but throughout positive effects on, and, and mixed effects for. However, regressions (4) to (6) exhibit a significant increase in the anti-eu-membership attitude. Moreover, when controlling for municipalities which have been outvoted, we find a statistically significant increase in satisfaction with democracy and individuals perceived political influence. 23