Abstract
We study a constitutional change in the German State of Bavaria where citizens, not politicians, granted themselves more say in politics at the local level through a state initiative election in 1995. This institutional setting allows us to observe revealed preferences for direct democracy and to identify factors which explain these preferences. Empirical evidence suggests that support for direct democracy is related to dissatisfaction with representative democracy in general rather than with an elected governing party.
Similar content being viewed by others
1 Introduction
Look at opinion surveys by leading pollsters, and you will find that citizens are relatively dissatisfied with the political system and interested in more direct democratic participation, possibly even much more direct participation. The interpretation, evaluation and discussion of such survey results on representative versus direct democracy has been largely an elite affair and political consequences are rarely drawn. This is comprehensible because stated support for direct democracy in opinion surveys need not reflect actual preferences for direct democratic participation. Hypothetical situations and erroneous self-reflections combined with moral satisfaction (Who will state that she is against democratic participation?) may introduce bias and make it difficult to draw concrete policy conclusions.
Ironically, yet coherent with the prevailing view of representative democracy, citizens have seldom been allowed to directly decide themselves on extensions of direct democratic participation rights. While political scientists and political economists have explored many facets of direct democracy, no empirical investigation so far exploits revealed preferences for direct democracy by looking at a direct democratic decision to implement more extensive direct democratic rights. Obviously, analyzing such a process of extending direct democracy would be revealing and our contribution attempts to fill this literature gap.
We analyze a constitutional change in the German State of Bavaria in 1995 where citizens, not politicians, granted themselves more say in politics at the local level through a constitutional initiative election at the state level. The constitutional change was accepted by a majority of Bavarian citizens and introduced important direct democratic instruments for municipalities, including municipal initiative petitions (Bürgerbegehren) and municipal initiative elections (Bürgerentscheid) which did not exist prior to 1995. The governing party in the state legislature campaigned against the systematic extension of direct participation rights.Footnote 1 This unique institutional setting allows us to focus on revealed preferences for direct democracy instead of analyzing opinion surveys; i.e., we focus on actual behavior of citizens in a real political decision with real consequences regarding the extension of direct democracy.
In surveys, dissatisfaction with representative democracy is regularly stated as a reason for the support of more direct democracy. The institutional setting allows us to explore factors which drive revealed preferences for more direct democratic participation. The effects of the vote share of the governing party at the state level, the difference between the vote share of this party at the state level and the municipal level, and turnout rates are revealing factors for the actual support of direct democracy. Their effects then allow tentative interpretations regarding the reasons for dissatisfaction of citizens with representative democracy in general versus suspicion and distrust with the governing party (here at the Bavarian state level). We do not measure dissatisfaction in a survey setting, but we employ an actual vote where preferences for direct democracy are revealed. Thereby, we gain in precision regarding the measure for support of direct democracy but can only argue that the observed evidence is consistent with dissatisfaction with representative democracy in general or distrust against the specific governing party. This then complements the existing literature that is currently focused on survey evidence.
Our empirical evidence for the over 2,000 Bavarian municipalities highlights three findings: (1) Support for direct democracy is strong when the support for the ruling governing party at the state level (Christian Social Union - CSU) is weakFootnote 2, (2) comparing CSU support at the state and at the local level shows that the state level is prevailing (even though the initiative has strong ramifications on the local level), and (3) higher turnout in past elections is also negatively related to our measure of support for direct democracy. While these findings constitute important evidence by itself, they also consistently highlight that dissatisfaction with representative democracy in general rather than dissatisfaction or distrust toward the state-ruling party CSU is driving the voting behavior. Apart from the main analysis, we run a number of additional tests to show the robustness of the results and the consistency of the interpretation.
Our paper is structured as follows: Section 2 presents theoretical considerations and discusses the related literature. Section 3 presents the institutional setting and the initiative at the state level for the extension of direct democracy. We discuss the data and the empirical strategy in Sect. 4 and present results in Sect. 5. Section 6 offers concluding remarks.
2 Theoretical considerations and literature
In many countries, regions and local jurisdictions citizens rely increasingly on initiatives, referendums and other forms of direct democracy to take political matters into their own hands. Numerous studies analyzed the effects that direct democratic participation and decisions have on policy outcomes (see, among others, Noam 1980; Frey 1994; Feld and Matsusaka 2003; or recently Matsusaka 2005, 2008, 2010 for overviews). While most of this literature focusses on the USA as well as Switzerland, a small emerging literature analyzes democratic elements in Germany (see Hessami 2014; Asatryan et al. 2014, 2016; Arnold and Freier 2015).Footnote 3 However, empirical evidence is comparatively scarce on the support for the introduction of direct democratic instruments. In particular, revealed preferences for direct democratic participation rights are usually unobserved. Historically, the extension of direct democratic participation rights seems to be associated with discontent with politicians in general (see Piott 2003), who then grant more participation rights in an effort to appease voters. We contribute to the literature on the demand for direct democracy by analyzing a decision of voters (instead of politicians) to grant themselves more direct democratic participation rights.
Opinion polls in numerous countries show strong, almost overwhelming demand of citizens for more direct influence on policy decisions (Bowler and Donovan 1994). According to surveys of the International Social Survey Program in 2004, over 60 % of survey respondents in the USA, Canada, Great Britain and Spain agree or strongly agree that “referendums are a good way to decide important political questions.” In Austria, Germany, and Switzerland, support is over 80 %. Other national and international pollsters offer similar numbers (Leininger 2015). Due to potential bias of surveys, the literature on the public approval of direct democracy is inconclusive (see, e.g., Dalton et al. 2001; Donovan and Karp 2006; Bowler et al. 2007). Craig et al. (2001) show that survey answers for the extension of direct democracy differ markedly by the way the question is asked: If citizens make a distinction between the political elite and themselves, i.e., “Us versus Them,” large majorities agree with direct democracy, while support is markedly weaker when asked about normal people’s capacity of participating. According to Dyck and Baldassare (2009), support for direct democracy varies on whether the survey questions focus on the abstract institution rather than concrete details, and Collingwood (2012) shows that support is lower when respondents are initially asked questions on ballot propositions. Thus, it is, unfortunately, not clear that opinion polls for more direct democracy actually correspond to citizen preferences for it and whether they over- or understate actual preferences. Instead of focusing on hypothetical support for direct democracy in opinion polls, our contribution analyzes an actual ballot proposition for more direct democracy that leads to a real constitutional change.
Extensive literature highlights the discrepancy between outcomes of opinion surveys and true preferences for politically important topics. The absence of reflected attitudes on certain issues can lead to improvised answers in surveys, and opinions may change in relatively short periods of time (Zaller and Feldman 1992; Diamond and Hausman 1994). Preceding questions, the interview manner, the context, among others, all influence the answers which may lead to different survey outcomes or seeming preference indication of survey respondents (Diamond et al. 1993; Hanemann 1994; List 2002). Proposed policies in surveys tend to be considered hypothetical, and real costs and policy consequences are only partly taken into account or even unknown. Such hypothetical bias combined with moral satisfaction can cause the survey results on politically relevant topics to be widely inaccurate (Kahneman and Knetsch 1992; Diamond and Hausman 1994; Neill et al. 1994; Cummings et al. 1997; Murphy et al. 2005; Funk 2016). Regarding prevailing norms, this may be particularly true for surveys on direct democracy.
Focusing on actual initiative elections provides a way to elicit revealed voter preferences directly. Initiative elections permit citizens to judge legislative proposals and rank them against the status quo, and they entail real policy outcomes (e.g., Schneider et al. 1981; Hersch and McDougall 1988; Frey 1994; Garrett 1999; Portmann et al. 2012; Brunner et al. 2013; Carey and Hix 2013). Voters in initiative elections put more time into thinking about a ballot decision, and the incentive to state true preferences is higher as their decisions entail real consequences. Schlaepfer et al. (2004) and Schlaepfer and Hanley (2006) compared stated preferences in professional surveys which were conducted before the awareness of an upcoming referendum with the decisions in the actual referendum and found that preferences indicated in surveys are largely incompatible with referendum decisions. By analyzing an initiative election on the introduction of direct democracy, we avoid challenges of surveys and obtain a direct measure for revealed preferences of voters regarding the support for direct democracy.
Higher-educated people are more aware of ballot propositions and have a stronger opinion on them (Bowler and Donovan 1994). Uninformed voters tend to abstain from voting in initiative elections. Thereby, initiative elections oversample the informed population (Osborne and Turner 2010; Stadelmann and Torgler 2013). However, informed answers are also overrepresented in surveys as ill-informed survey respondents often do not answer questions such that both, initiative elections and surveys, bias the outcome toward the opinion of informed voters (Althaus 1996). As open public debates precede an actual decision in initiative elections, information uncertainty is generally lower than for survey respondents (Frey 1994; Lupia 1994) and abstaining from a ballot is more closely associated with true indifference. Moreover, information is more easily accessible ahead of initiative elections than for surveys and voters generally tend to be better informed when they can participate more directly in political decisions (Feld and Kirchgaessner 2000; Benz and Stutzer 2004).
As in any political process, particular groups may try to influence referendum decisions through campaigning (Lupia 1994). Bohnet and Frey (1994) and Frey (1994) argue that referendums fulfill individual preferences and are able to break the cartel of politicians directed against voters. Nevertheless, governments and politicians may influence how citizens vote in referendums. Bowler and Donovan (1994) suggest that endorsements by political parties and politicians serve as a channel of information and that they have an influence at the ballot (see also Lupia 1994; Nalebuff and Shachar 1999; Stadelmann and Torgler 2013). Moreover, partisanship increases the probability of forming opinions by statements of the political elite (Eichenberger and Serna 1996). Results by Trechsel and Sciarini (1998) suggest an impact of political elites on voting outcomes in referendums, and Smith and Tolbert (2001) argue that political party affiliation is the most important influence on voting decisions. For initiative elections, parties tend to become involved when the issue affects the party’s ideology and parties attract voters by taking a side to withdraw voter support from another party (Smith and Tolbert 2001). Jenssen and Listhaug (2001) note that voters in initiative elections may take positions on an issue based on party cues. Our setting allows us to contribute to this literature by taking account of party positions with respect to an extension of direct democracy, and we can analyze the influence of parties on revealed preferences for direct democracy. Thereby, we extend the literature on the reasons for support of direct democracy by investigating whether suspicion of elected governing party explains actual support for direct democracy (see, e.g., Gerber 1999 or Dalton et al. 2001) or rather dissatisfaction with representative democracy in general.
3 Institutional setting
3.1 Direct democracy at the state level and political parties
Germany implemented a party-centered representative democracy at the national level after 1945. Direct democratic participation rights such as initiative petitions or initiative elections have been included in some State (Länder) constitutions. In particular, the State of Bavaria grants comparatively extended direct democratic participation rights to its citizens. In 1946, a two-stage legislation with initiative petitions and elections had been implemented for decisions at the state level. An initiative petition (Volksbegehren) constitutes an attempt of citizens to change or adapt a law or a constitutional amendment. It is addressed to the State Parliament (Landtag). Before the initiative petition gets submitted to voters for a decision in an initiative election (Volksentscheid), two signature requirements need to be fulfilled: In a first step, 25,000 signatures of eligible voters (corresponding to 0.3 %) are required to have the legal admissibility of the initiative formally tested. In a second step, 10 % of the electorate have to sign the initiative petition up to 14 days before it is submitted to the State Parliament. The State Parliament has the right to formulate a counterproposal to the initiative. If the State Parliament rejects the initiative or submits a counterproposal, a popular vote in an initiative election is necessary. The proposal with the relative majority is accepted and becomes law.Footnote 4 Constitutional amendments need to fulfill an approval quorum of 25 % of eligible voters. Until 2014, 19 referendums were held in Bavaria at the state level.
The political party landscape in Bavaria is influenced by the conservative Christian Social Union (CSU) which held an absolute majority in the State Parliament and Government from 1962 until 2008 and regained an absolute majority of seats in 2013.Footnote 5 The party only competes in the State of Bavaria. Its spectrum of supporters is broad and ranges through all strata of society (Pappi 2011). The Social Democrats (SPD) represent the party for the working class at the federal level in Germany; however, the working class votes in the same proportions for the SPD and the CSU in Bavaria. The Greens and the Liberals (FDP) enjoy electoral support at the state and municipal level in Bavaria. In state-level elections, their vote shares lie between approximately 2 to 8 %.Footnote 6 Since 1998 the Free Voters (Freie Wähler), a conservative party next to the CSU, competes in state elections and passed the electoral 5 % hurdle in 2008.
3.2 Introducing direct democracy at the local level
The municipal level in Germany is responsible for a wide array of tasks. They provide administrative services and decide on spending over child care, basic social services as well as cultural and recreational issues. They also independently set three local tax rates and raise a number of fees for different services. Nevertheless, the local level is strongly intertwined with the state level. The state level provides grants, sets tasks and decides on the rules of the game for their municipalities. Thus, there are many interactions between the state and the local level. Importantly, the rules for citizen participation at the local level are also set at the state level and have been changed through the initiative in 1995.
While the Bavarian constitution granted extended direct democratic participation rights at the state level, the situation at the local level has been entirely different until 1995. In 1951, the incorporation of local-level initiative petitions (Bürgerbegehren) and local-level initiative elections (Bürgerentscheid) was not ratified by the state legislature (see Bierl 1995; Bayerischer Landtag 1995c, p. 308 for details). In the following 40 years, multiple motions to implement direct democracy at the local level by smaller parties were all inhibited by the governing party, i.e., the CSU.Footnote 7 At the time, the CSU advanced numerous arguments against direct democratic participation at the local level against the Greens and other supporters of direct democracy, e.g., that a minority might overrule a majority and that municipalities would lose their ability to govern. CSU officials expressed concern that direct democracy at the local level would not match with the system of representative democracy (CSU Parteitag 1982) and that local direct democratic decisions would be “hijacked by demagogues and pied pipers” (Bayerischer Landtag 1995b, p. 893).Footnote 8 The CSU won an outright majority in both state elections of 1994 and 1998, with 52.8 and 52.9 %, respectively. Between the two state elections, in 1995, direct democracy at the local level was introduced through an initiative election at the state level.
In 1995, a year after the last state election, the citizen’s association for “More Democracy” (Mehr Demokratie in Bayern e.V.) formulated an initiative to introduce direct democracy at the local level through a constitutional amendment at the state level, i.e., the state constitution needed to be changed to allow direct democracy at the local level. At the state level such an initiative was possible according to the existing state constitution as outlined above. The initiative petition passed the signature requirement with 13.7 % of the electorate and was submitted to the State Parliament. The proposition was considered citizen-friendly with extensive participatory and decisive rights as well as low hurdles (see Bayerischer Landtag 1994a, b). Parliamentary criticism toward the initiative was intense, and it was particularly targeted at missing quora of approval, the non-exclusion of certain policy areas from an initiative election, and some politicians argued that “all this would serve special interests” (Bayerischer Landtag 1995c, p. 311).
As the initiative petition passed the state signature requirement, the governing party CSU formulated a counterproposal (see Bayerischer Landtag 1995a). This alternative proposal would have introduced important steps against the extension of direct democratic rights and instruments at the local level. In particular, it included a quorum requirement and subject exclusions within a legislative instead of a constitutional framework (Bayerischer Landtag 1995b, pp. 889–895) which would have made it prone to arbitrary changes ex post. A rigorous debate in parliament followed, and the counterproposal was accused as a “bluff package” (Bayerischer Landtag 1995d, p. 893). With the majority CSU voting for its own measure while the opposition voted for the initiative in the State Parliament, the initiative was not directly ratified (64 votes for the initiative, 89 against it) and the counterproposal as well as the initiative had to be put to the polls. In the intense public debate ahead of the initiative election, opposition parties, in particular the Greens, supported the initiative of More Democracy as did some other 50 small associations.Footnote 9 However, CSU-dominated large municipal head organizations “Landkreistag,” “Städtetag” and “Gemeindetag” for counties, cities and municipalities all rejected the initiative in favor of the counterproposal. The referendum took place on October 1, 1995. The CSU counterproposal was rejected by the voters with only 38.7 % supporting it. A majority of 57.8 % of voters supported the original initiative. 3.4 % rejected both amendments. The new constitutional amendment was implemented directly afterward on November 1, giving Bavarian municipalities the most extensive direct participation rights in Germany.Footnote 10 In 2010, 15 years after the implementation, Bavaria counted over 1700 local initiative petitions and 900 local initiative elections, making it the German State with the most vivid direct democratic activity at the local level.Footnote 11 Anecdotal evidence also suggests that other German states modeled extensions of direct democratic participation rights from Bavaria.
This unique institutional setting permits us to analyze the extension of direct democracy through a direct democratic process. Citizens themselves initiated a constitutional reform through an initiative petition at the state level to extend direct democratic participation rights at the local level. Past attempts to extend direct democracy through representative democracy all failed as the governing party CSU blocked them. We can, thus, analyze how support for the governing party and other factors drive citizens to vote for more or less direct democratic participation; i.e., instead of relying on opinion polls to measure potential support of direct democracy, we analyze revealed preferences for a real constitutional proposal.
4 Data and empirical strategy
4.1 Descriptive statistics
The state of Bavaria consists of 2056 municipalities, including 25 county-free cities. We obtained official results of the More Democracy initiative election (Volksentscheid) from the Bavarian State Office for Statistics (Bayerisches Landesamt für Statistik und Datenverarbeitung). On the level of the individual municipality, we observe revealed preferences for direct democracy by approval rates of the initiative election as well as approval rates for the counterproposal advanced by the State Parliament.
As our main variable of interest, we analyze the Governing Party Vote Shares (CSU) in the prior state election of 1994, \( VS ^{CSU State}\). The temporal closeness of the 1994 state election, and the 1995 state-wide initiative election contributes to ruling out changes in party loyalty over time. Moreover, the CSU obtained virtually the same state-wide support in 1998 as in 1994 and the support levels in municipalities are highly correlated for the two elections. We also gathered CSU vote shares in the last municipal elections in 1990, \(VS^{CSU Local}\). Since the CSU does not run in all municipal elections, we code a dummy indicating whether the CSU appeared on the ballot. Furthermore, we collected information on voter turnout in the state election, the preceding municipal election and the initiative election, which allows us to measure general interest in politics, political culture and citizen engagement. Turnout for the state-level initiative election measures the effect of voter mobilization on revealed preferences for direct democracy at the local level. The effect of a municipality’s support for the opposition party the Greens—which was the main political advocate for more direct citizen engagement—is captured by the Green vote share in the state election. Moreover, we have information on the number of parties running in the municipal election and whether the directly elected mayor is a member of the CSU or not.
We capture demographic, socioeconomic and other differences among municipalities by the following variables: Log(Population) size distinguishes between urban municipalities and rural ones. The age distribution is captured by the variables Share Young (under 18 years) and Share Elderly (over 65 years). The strength of the economy is proxied by the rate of Employment, and the municipal financial situation is reflected by the level of Per Capita Debt. We include the share of Catholics which stems from the 1987 census to measure conservatism and control whether a municipality is a University Town. Interest heterogeneity within the municipality is accounted for by the amount of In-migration from East Germany and neighboring states between 1987 and 1995.Footnote 12
Table 1 shows descriptive statistics for all variables used in the analysis. Average approval for the initiative by Mehr Demokratie e.V. is greater than 50 %, reflecting the fact that the initiative was successful.Footnote 13 The CSU is the dominant party on the state level, garnering on average almost 60 % of the vote in the 1994 election. At the municipal level, the CSU faces stronger competition by local parties and conservative citizen groupsFootnote 14 and it is less successful with an average vote share of 22.4 %. As the party stands for election in just 56 % of the municipalities, this implies an average vote share of about 40 % conditional on running. Osborne and Turner (2010) suggest that in common value environments referendums lead to higher welfare than a social planner because indifferent voters do not participate in the ballot. Turnout for the initiative by More Democracy was at 37.9 % on average. This corresponds to other referendums in Bavaria,Footnote 15 and similar participation rates are observed in countries with extensive direct participation rights such as Switzerland (Stadelmann et al. 2013). Turnout is highest in local elections (roughly 80 %) and about 10 % points lower in state elections.
4.2 Empirical strategy and expected effects
To estimate ceteris paribus effects of CSU strength and other factors on the revealed preference for direct democracy in the initiative election, we employ a regression control framework. Hence, we estimate the following model:
where \( PreferencesDD_i\) is the approval rate for the initiative which reflects revealed preferences for direct democracy and \(\alpha \) is a constant. \(\beta _1\) is the central coefficient of interest capturing the effect of \(VS_i^{CSU State}\), i.e., the vote share of the governing CSU in the state election. \(\beta _2\) captures the influence of the CSU strength in a municipality. The variables \(Turnout_i^{State}\), \(Turnout_i^{Local}\) and \(\# Party_i\) represent the turnout in past state- and local-level elections as well as the number of parties at the local level, respectively. \(X'_i\) is a vector of control variables, \(\theta _k\) is a fixed effect for administrative region k, and \(\epsilon _i\) is an error term. Administrative region fixed effects capture economic, demographic, social and cultural differences between the seven regions (e.g., Upper Franconia vs. Swabia). The unit of observation is the individual municipality, indexed by i (\(i=1,2,\ldots ,2031\)). We always estimate robust standard errors.
The state-governing CSU campaigned for its own counterproposal, and it was the only party campaigning against the initiative. Support for direct democracy depends on trust in politicians. If citizens generally trust the governing party’s performance (here at the state level) and its cues, and if dissatisfaction with government is low, we expect \(\beta _1<0\); i.e., CSU strength in the past state election should then have, ceteris paribus, a negative influence on approval for the initiative. If, on the other hand, the electorate is dissatisfied and suspicious of the governing party (Bowler et al. 2007), they may demand more direct control; i.e., the influence of the CSU vote share would then be positive.Footnote 16
Similar to the above discussion, we would also expect the strength of the CSU at the local level to be negatively related to the support for direct democracy, hence \(\beta _2<0\). However, traditionally, elections at the municipal level are less influenced by trust in a specific party compared to state elections where parties matter relatively more than individual politicians. Thus, we expect the (absolute) municipal CSU vote share to have a smaller influence on the initiative election for direct democracy than the (absolute) state CSU vote share, i.e., \(|\beta _1|>|\beta _2|\).
Turnout at the state-level initiative measures the effect of mobilization within municipalities for direct democracy. The effect of overall mobilization on support for direct democracy is theoretically ambiguous as it is unclear whether supporters or opponents of direct democracy are easier to mobilize with campaigning, such that support for direct democracy and turnout for the initiative are jointly determined. However, low turnout in the previous state elections can be interpreted as a sign of dissatisfaction with politics in general, such that we expect a negative relationship between turnout in the state election and support for direct democracy. The same expectation holds for turnout in the preceding municipal election.
When analyzing approval for direct democracy and interpreting the effects of party strength, we need to control for a number of other variables which may influence preferences for direct democracy and support for the governing party at the same time. The strength of the opposition Green Party which supported the initiative should have a positive influence on voting for the initiative of More Democracy. The “New Politics” theory (Inglehart 1990, 1999) suggests that support for direct democracy should be more common among the younger, more urban, better-educated and less conservative population. Therefore, we expect Share Young, Log Population, University to have a positive influence on the dependent variable and Share Elderly and Share Catholic to have a negative effect on revealed preferences for direct democracy. Economic factors such as debt levels and employment opportunities as well as in-migration may affect support for the governing party and for direct democracy, but the sign of their effect is theoretically ambiguous.
5 Results
5.1 Baseline results
Figures 1 and 2 highlight the motivation and a central result of our contribution. Panel (a) in Fig. 1 shows the revealed preferences for more direct democracy at the local level, and panel (b) contrasts it with the CSU strength across the municipalities in Bavaria. A negative relationship between CSU strength and approval for direct democracy is directly discernible. Municipalities with higher support for the CSU typically had substantially lower approval rates for direct democracy. As the CSU was and still is the governing party and campaigned actively against the extension of direct democracy, we interpret this as first evidence against the dissatisfaction hypothesis regarding governing parties; i.e., municipalities with higher support for the governing CSU trusted their representatives and voted relatively more against the extension of direct democracy. The relationship becomes even more apparent when looking at the scatterplot in Fig. 2 which visualizes the negative correlation between the CSU vote share and approval for direct democracy in the referendum. Although the overall level of support for direct democracy is high throughout, we detect a strong negative correlation (\(\rho =-0.6208\) and p value = 0.000) between the strength of the governing party CSU and the revealed preference for direct democracy.
Table 2 presents econometric evidence. Specification (1) gives the baseline model with the yes share for the Direct Democracy initiative, \(PreferencesDD_i\), as the dependent variable, and the Governing Party Vote Share in the preceding state-level election, \(VS_i^{CSU State}\), as the single explanatory variable. Ceteris paribus, the CSU strength in a municipality has a negative effect on the approval for direct democracy in the referendum. The point estimate suggests that for each percentage point increase in the CSU vote share in the state election, approval for the initiative by More Democracy goes down by 0.61 % points. This is a substantial association when taken at face value: Although the initiative was with 57.8 % clearly accepted, a difference in only approximately 1.437 standard deviations in the CSU vote strength would have sufficed to prevent the extension of direct democratic participation in Bavarian municipalities. We will show that the sign and the magnitude of the estimated relationship is highly robust, it stems from a change in the CSU strength, it can most likely be interpreted causally, and unobservables are unlikely to explain the effect.
To ensure that these results are not driven by demographic or socioeconomic differences between municipalities which affect CSU strength and preferences for direct democracy at the same time, we add control variables in specification (2). We observe that the literature’s expectations regarding the control variables broadly tend to hold: A higher share of elderly and Catholics is negatively associated with approval of more direct democracy at the local level. More urban municipalities show higher levels of support for direct democracy. A municipality’s financial situation, having a university and the share of employed are not significantly related to support levels. Importantly, qualitative and quantitative results for the CSU strength at the state election a year prior to the initiative remain virtually identical when compared to specification (1).
Specification (3) adds the CSU vote share at the previous municipal election, \(VS_i^{CSU Local}\), and an indicator variable for whether it ran as a party. CSU strength at both levels (state and local) has a negative influence on support for direct democracy. A ceteris paribus increase in the CSU vote share in the state election of 1 % point is associated with a 0.594 % points lower support for direct democracy, and for each percentage point increase in the CSU vote share in the municipal election, approval declines by 0.082 % points. As expected, \(|\beta _1|>|\beta _2|\); i.e., the influence of the CSU strength at the state level is larger than at the municipal level. The null hypothesis of equality of the two coefficients is rejected with a p value of 0.000.Footnote 17
What is also interesting about the difference between \(\beta _1\) and \(\beta _2\) is that the initiative seems to be a vote about the state level rather than the local level even though the policy consequences are mainly at the local level. Here, it is important to remember that the initiative was communicated and implemented at the state level and the political fight was entirely dominated by the independent organization (Mehr Demokratie e.V.) and some political organizations, e.g., the Greens on the one side, and the CSU on the other side. While the local considerations are not negligible (we do report significant negative effects), they are much smaller in magnitude than the state level.
In specification (4), we add further political variables—among them the vote share for the Green Party and measures of turnout in past local and state elections. The vote share of the Green Party is a positive predictor of voting for direct democracy. The Greens were the strongest advocates among state parties of more direct citizen participation at the local level in Bavaria.Footnote 18 As mayors are directly elected in Bavaria, it could be expected that they do not have any significant influence on support for direct democracy which is fully consistent with our findings. Mayors are already under direct control independent of their parties.Footnote 19 However, the number of competing parties in a municipal election proxies dissatisfaction with the political system in general and is, consequently, positively related to support for direct democracy.
As expected, turnouts in the state election and in the municipal election are both negatively correlated with approval for direct democracy.Footnote 20 We interpret this as additional evidence that dissatisfaction with the representative democracy in general is the major reason for supporting the initiative for more direct democracy and not dissatisfaction with a specific governing party. Municipalities that expressed dissatisfaction with representative democracy with lower turnout levels in past elections now also vote in favor of direct democracy. Moreover, the effect of municipal turnout remains statistically insignificant which we take as indication that the state-level initiative is indeed affected more by state-level politics even though it has local ramifications.
Importantly, the inclusion of all these political control variables does not affect the statistical significance of our main variable of interest, the magnitude of the effect remains unchanged, and CSU strength in state election has an absolutely higher effect than CSU strength in municipal election, i.e., \(|\beta _1|>|\beta _2|\) (p value = 0.000).
Not only the magnitude of the CSU vote share is important but also its explanatory power. This can best be seen by comparing the R \(^2\) of the regressions in columns (1)–(4). CSU strength at the state level generates an R \(^2\) of already 39 % in column (1). More importantly, while the R \(^2\) rises to 0.58 when including further a full set of control variables (i.e., the regional fixed effects), the strength of the governing party CSU is still the single best predictor of voting in the initiative for an extension of direct democracy. This result suggests that party cues are an important predictor even in a direct democratic decision.
As mentioned above, we interpret the sign of the CSU vote share variable, the relative magnitude of the coefficients at the state and local level as well as the evidence on turnout as indications for a dissatisfaction with representative democracy in general (rather than dissatisfaction with the governing party at the state level). It should be noted that our arguments here can only be suggestive. Compared to the evidence in surveys which can ask the citizen’s opinions about their dissatisfaction directly, our approach must be viewed as more indirect. However, as argued above, surveys are severely hampered by concerns of biased answers, something that we are not concerned with in our direct (consequential) vote. Our consistent evidence for dissatisfaction with representative democracy in general therefore complements the existing literature.Footnote 21
5.2 Robustness checks
The baseline results are robust to a variety of alternative specifications as shown in Tables 3 to 5. Robustness tests include the full set of control variables as well as administrative region fixed effects.
5.2.1 Weighting and analyses of subsamples
As municipalities have different population sizes, regressions where each observation has the same impact on the estimated coefficient can be misleading. We therefore replicate our main result from column (4) of Table 2 using population weights. This implies that a municipality with 1000 inhabitants will have twice the effect on the coefficient of interest than a municipality with only 500 inhabitants. The results can be found in column (1) of Table 3. It becomes evident that population weighting does not affect our quantitative or qualitative results.
Specification (2) looks at the subsample of 775 municipalities with fewer than 2000 inhabitants to ensure that results are not driven by small municipalities. Specification (3) includes county-free cities into the sample.Footnote 22 In both specifications, the influence of CSU strength at the state and the municipal level remains statistically significant, the absolute magnitude remains comparable, and the influence of strength at the state level is higher (\(|\beta _1|>|\beta _2|\)).Footnote 23
Analyzing municipal samples with differential strength of the CSU at the state level does not affect the statistical significance nor the magnitude of the CSU strength on support for direct democracy (specifications 4, 5 and 6). In particular, for municipalities where the CSU was above average strength (column 4) or close to the 50 % benchmark (column 5), its effect remains quantitatively similar to other specifications, which lends support to the hypothesis that dissatisfaction with the political system is not directly related to the governing party. To have a more precise proxy on whether the CSU is the local governing partyFootnote 24, we restrict the sample to only include towns with a CSU mayor (column 6). Even here, the results are almost identical, in particular not only the coefficient on the state variable, but also the effect for CSU vote share at the municipal level.
It is interesting to note that when only estimating the influence of the CSU strength at the local level (specification 7), i.e., without including the CSU vote share in the state election, the coefficient \(\beta _2\) remains negative and statistically significant, and it increases slightly in absolute size but remains comparable to specifications where \(VS_i^{CSU State}\) is also included.
5.2.2 Selection on unobservables
To further demonstrate the robustness of our results to potential endogeneity issues due to omitted variables, we conduct robustness tests in the spirit of Oster (2014) in Table 4. The basic idea of this test is that we can learn something about potential bias from unobservables by looking at coefficient movements after including observed control variables into the regression. Intuitively, the logic behind the approach is the following: If the inclusion of observed covariates increases the explanatory power of the model substantially, i.e., the R \(^2\) goes up, but leaves the coefficient of interest almost unaltered, i.e., \(\beta _1\) remains stable, then one may assume that potential unobserved variables would not affect the coefficient either, since the included controls are already the ones the researcher deemed most important in terms of reducing endogeneity issues. We apply this approach and calculate the so-called identified set under a proportional selection assumption and a given maximum potential value of R \(^2\). If this set excludes zero, the coefficient estimates reach a level of robustness “in the range of what would be seen if the treatment was randomized” (Oster 2014); i.e., the results are likely not to suffer from endogeneity due to unobservables. As required by the literature, we compare movements in the coefficient of \(VS^{CSUState}\) and the R \(^2\) from the most parsimonious specification (without controls) to the full model (with all observed covariates) and then calculate the bias-adjusted coefficient \(\beta ^{*'}_1 = \tilde{\beta _1} -{\tilde{\delta }}\frac{(\dot{\beta _1} - \tilde{\beta _1})(R_\mathrm{max} - {\tilde{R}})}{({\tilde{R}} - {\dot{R}})}\) that constitutes the bound of the identified set. The results are very reassuring: We find that the identified set always excludes zero, independent of the degree of selection on unobservables (the choice of \({\tilde{\delta }} = \{0.5,0.75,1\}\)). Hence, even under the most conservative assumptions (\({\tilde{\delta }} = 1)\) and \(R_\mathrm{max} = 1\), our results are robust to potential omitted variable bias.Footnote 25
5.2.3 Exploiting turnout and vote shares
Our results suggest a negative causal effect of the Governing Party CSU Vote Share on support for direct democracy. To ensure that the effect stems from higher individual levels of support instead of different compositions of the electorate (ecological fallacy), we control for opposition parties and turnout in the initiative election. Table 5 provides additional support for our interpretation: We estimate the relationship for four subsamples of high/low CSU vote shares and high/low turnout in municipalities. Changes in the \(VS^{CSUState}\) are likely to stem from higher individual levels of support in municipalities where the CSU vote share is high and where turnout in the initiative election is high (first quadrant), as the voting population in these municipalities is composed of individual CSU supporters. We observe that the coefficient of the variable \(VS^{CSUState}\) for this subsample is \(-0.66\) which closely resembles previous estimates. If individual voters from the opposition would drive our result or low turnout rates would drive our effect, we should expect a high absolute coefficient for municipalities with low CSU vote shares and low turnout in the initiative. This is not the case as shown in the fourth quadrant: Increasing \(VS^{CSUState}\) decreases the support for direct democracy in this subsample too, but the effect is with 0.46 % points slightly lower than in the first quadrant. This suggests that individual levels of support for the governing party matter. The second and the third quadrant provide results for municipalities where the CSU vote share is high but turnout is low and where the CSU vote share is low but turnout is high, respectively. All effects are qualitatively and quantitatively similar to earlier estimates, confirming once more the robustness of our baseline results.
5.2.4 Additional evidence in Online Appendix
To complement the analysis, we have considered additional tests which we present and discuss in the Online Appendix to this paper. In this Online Appendix, we explore how the vote shares of other local independent parties or the SPD influence our results. We also present more robustness tests in which we highlight the interaction effects of turnout and the number of parties with our main variable of interest, CSU vote share at the state level. Finally, we designed an IV approach using the amount of land in agricultural use and we use a Bayesian averaging approach, all to underpin the consistency of our main findings. Importantly, our findings remain qualitatively and quantitatively the same.
6 Conclusions
Direct democratic participation is a relevant policy issue in many countries around the world, and citizens seem to be interested in more direct democracy when looking at opinion surveys on the subject. Nowadays, almost no politician argues openly against more citizen participation prior to elections but promises for direct democracy are seldom fulfilled after elections.
Voters in Bavaria decided in a constitutional initiative election at the state level in 1995 to grant themselves more direct democratic participation rights at the local level. This unique setting allows us to directly analyze revealed preferences for direct democracy instead of relying on opinion surveys. Employing revealed preferences complements other analyses which employ opinion surveys and questions regarding the extension of direct democracy. It allows new and different insights because the decision we analyze is binding and has been implemented after the successful initiative election. In particular, the institutional setting allows us to explore factors which drive revealed preferences for more direct democratic participation at the local level and we can explore whether dissatisfaction with representative democracy or rather dissatisfaction with the governing party at the state level (and often at the local level too) is related to actual voter support for direct democracy.
Empirical results show that support for direct democracy at the local level is negatively associated with support for the elected governing party. In fact, electoral support for the state-governing party at the local level in the previous state election is the strongest single predictor for the actual extension of direct democracy. Numerous robustness analyses and testing differential hypotheses lead to the same conclusion. We also account for potential endogeneity issues and find that the effect of the strength of the governing party is most likely not affected by unobservables. Thus, our results provide tentative support to the view that dissatisfaction with politics is not linked to a specific elected governing party. Rather it seems to be the case that dissatisfaction with representative democracy in general is a driving factor for voting for direct democracy. Future analyses on effects of direct democracy may use the initial levels of support as a control variable for preferences or as a potential instrument.
Notes
Public pressure on this matter forced the governing party CSU to give up their general position against citizen participation, and they subsequently campaigned for their own proposition of a weakened version of citizen participation.
An instrumental variable approach suggests that the causal relationship most likely runs from stronger electoral support to voting against direct democracy. Results are presented in Online Appendix to this paper.
Hessami (2014) looks at the introduction of direct mayoral elections. Direct mayoral elections provide an additional form of direct voter participation which was introduced simultaneously in the beginning of the 1990s (not in Bavaria which had direct mayoral elections since 1946). Asatryan et al. (2014) and Asatryan et al. (2016) study the policy outcomes (spending and taxation) of direct democratic initiatives at the local level. Finally, Arnold and Freier (2015) investigate the prevalent relationship between the number of initiatives and formal hurdles in the state constitution.
In 2000, the Bavarian state election law (Landeswahlgesetz) was changed due to a decision of the Bavarian Constitutional Court. If a referendum and a counterproposal are submitted, voters have now more than one vote, and in case more than one option wins a majority of the votes, a tie-break vote is necessary (Bayerisches Gesetz- und Verordnungsblatt Nr. 15/2000, p. 365).
See Kauder and Potrafke (2016) for a discussion of the political landscape in Bavaria and the rent extraction of the CSU in supermajorities.
After a term in the State government the Liberals did not manage to win any seats in the 2013 election.
Famous but failed attempts to expand direct democracy include proposals in 1981, 1985, 1987 and 1991 (Bayerischer Landtag 1991, p. 400)
The protocol of the plenary session in 1995 states: “Der Bürgerentscheid ist eine Spielwiese für Volksverführer und Demagogen. Die vom Volk gewählten Vertreter in den kommunalen Parlamenten sollen entmachtet und die Mehrheiten von aktionistischen Minderheiten terrorisiert werden.”
The Social Democrats officially supported the initiative. Informal interviews and personal conversations with former social democratic mayors, however, make us believe that this position was rather taken in opposition to the governing CSU than out of support for more local direct democracy.
In 1999, the Bavarian Constitutional Court decided upon the introduction of a quorum. Since then, local initiative elections are only successful when they achieve a quorum of approval which depends on the population size of the municipality.
Today, at the national level CSU politicians often argue for more direct democratic participation.
For the IV estimation in Online Appendix, we have added additional data on the Share of Agricultural Soil Surface, which is then used as the instrument for the strength of the conservative CSU.
Note that the average weighs all municipalities equally, i.e., independent of population size. We therefore observe a small difference between average approval in our sample (54.2 %) and the official result of the initiative (57.8 %).
Some local groups affiliate with the state CSU such that there is no registered list bearing the term “CSU” in local elections.
Total turnout was 43.8 % for a change in the waste disposal law in 1991 and 39.9 % for a constitutional reform and abolishment of the Senate in 1998. Recent turnout for referendums in 2013 was higher because state elections were held the same day.
It could be argued that the electorate may vote for the initiative because voters might have an interest in setting the political agenda. However, this argument leaves open the question why a higher strength of the CSU (or other parties) should be associated with more demand for direct democracy if not because of discontent with the governing party. Importantly, discontent with the governing party does not need to result in lower support for it in elections as electoral support depends on the alternatives offered by other parties.
Note that a somewhat smaller estimate at the local level is also consistent with the idea that some of the voters indeed prefer to control their local parties. In the end, the point estimate is still negative such as to defend that the dissatisfaction effect with representative democracy is prevailing.
In the Online Appendix, we also explore the effects of other parties such as the SPD or the free voter associations and other local independent parties, for which we have no clear hypothesis for the effect.
An alternative explanation is that the effect of the CSU mayor is already captured by her/his party.
We are aware of the endogeneity of this variable. Still, we think that reporting this interesting correlation is of value to the reader. Reassuringly, results remain entirely stable when dropping turnout for the initiative from the model.
For these 25 cities, we are lacking mayoral election results as well as the number of parties running for the municipal council, which is why we excluded them in the baseline specification. Here, we include them to show that they do not bias the general results.
As we use only one cross section, we do not have a natural level to cluster the observations. Given that the municipalities are organized in counties, we tried to cluster the standard error by county (71 clusters) to allow for cross-dependencies on this level (results not reported). The results remain entirely robust.
Note that parties at the local level in Germany do not have formal coalition agreements. Decisions are made on a case-by-case basis with potentially changing majorities. Hence, it is difficult to know what party is in fact the ruling party. We turn to the mayor’s office instead and measure whether the CSU is holding the mayor’s office.
Reasoning the other way around, one can also ask which value of \(\delta \) would be needed to make the identified set include zero. We find that this maximum possible value of \(\delta \) is 6.49, implying that selection on unobservables would need to be more than six times as large as selection on observables to bias our coefficient in a way that the sign would change.
References
Althaus, S. L. (1996). Opinion polls, information effects and political equality: Exploring ideological biases in collective opinion. Political Communication, 13, 3–21.
Altonji, J. G., Elder, T. E., & Taber, C. R. (2005). Selection on observed and unobserved variables: Assessing the effectiveness of catholic schools. Journal of Political Economy, 113(1), 151–184.
Altonji, J. G., Elder, T. E., & Taber, C. R. (2008). Using selection on observed variables to assess bias from unobservables when evaluating swan-ganz catheterization. American Economic Review, 98(2), 345–50.
Arnold, F., & Freier, R. (2015). Signature requirements and citizen initiatives: Quasi-experimental evidence from Germany. Public Choice, 162, 43–56.
Asatryan, Z., Baskaran, T., & Heinemann, F. (2014). The effect of direct democracy on the level and structure of local taxes. ZEW Discussion Papers, Number 14-003.
Asatryan, Z., Baskaran, T., Grigoriadis, T., & Heinemann, F. (2016). Direct democracy and local public finances under cooperative federalism. Scandinavian Journal of Economics. doi:10.1111/sjoe.12169.
Bayerischer Landtag (1991). Plenarprotokoll 12/9. 12. Wahlperiode, 9. Sitzung, 21.02.1991, Muenchen.
Bayerischer Landtag (1994a). Gesetzentwurf Sechstes Gesetz zur Aenderung der Verfassung des Freistaates Bayern. Drucksache 13/208, 13. Wahlperiode, Gesetzentwurf SPD und BUENDNIS 90 DIE GRUENEN, Schmidt et al., 15.12.1994, Muenchen.
Bayerischer Landtag (1995a). Gesetzentwurf zur Einfuehrung von Buergerantrag, Buergerbegehren und Buergerentscheid in Gemeinden und Landkreisen. Drucksache 13/1333, 13. Wahlperiode, Gesetzentwurf CSU, Glueck et al., 26.04.1995, Muenchen.
Bayerischer Landtag (1994b). Gesetzentwurf zur Aenderung der Gemeindeordnung und der Landkreisordnung. Drucksache 13/209, 13. Wahlperiode, Gesetzentwurf SPD und BUENDNIS 90 DIE GRUENEN, Schmidt et al., 15.12.1994, Muenchen.
Bayerischer Landtag (1995b). Plenarprotokoll 13/16. 13. Wahlperiode, 16. Sitzung, 27.04.1995, Muenchen.
Bayerischer Landtag (1995c). Plenarprotokoll 13/23. 13. Wahlperiode, 23. Sitzung, 04.07.1995, Muenchen.
Bayerischer Landtag (1995d). Plenarprotokoll 13/8. 13. Wahlperiode, 8. Sitzung, 24.01.1995, Muenchen.
Benz, M., & Stutzer, A. (2004). Are voters better informed when they have a larger say in politics? Evidence for the European Union and Switzerland? Public Choice, 119, 31–59.
Bierl, P. (1995). Die Schadensgrenze fest im Auge. Sueddeutsche Zeitung—Landkreis Fuerstenfeldbruck 19.09.1995, 2.
Bohnet, I., & Frey, B. S. (1994). Direct-democratic rules: The role of discussion. Kyklos, 47, 341–54.
Bowler, S., Donovan, T., & Karp, J. A. (2007). Enraged or engaged? Preferences for direct citizen participation in affluent democracies. Political Research Quaterly, 60, 351–362.
Bowler, S., & Donovan, T. (1994). Information and opinion change on ballot propositions. Political Behavior, 16(4), 411–435.
Brunner, E., Ross, S. L., & Washington, E. (2013). Does less income mean less representation? American Economic Journal: Economic Policy, 5(2), 53–76.
Carey, J. M., & Hix, S. (2013). District magnitude and representation of the majority’s preferences: A comment and reinterpretation. Public Choice, 154(1–2), 139–148.
Collingwood, L. (2012). Levels of education and support for direct democracy. American Politics Research, 40(4), 571–602.
Craig, S. C., Kreppel, A., & Kane, J. G. (2001). Public opinion and support for direct democracy: A grassroots perspective. In M. Mendelsohn & A. Perkin (Eds.), Referendum democracy: Citizens, elites and deliberation in referendum campaigns (pp. 25–46). New York: Palgrave Macmillan.
CSU Parteitag (1982). Wir in Bayern CSU 82’ Kraftvoll in die Zukunft: Programm zur Landtagswahl ’82. Abteilung Oeffentlichkeitsarbeit der CSU-Landesleitung.
Cummings, R. G., Elliott, S., & Harrison, G. W. (1997). Are hypothetical referenda incentive compatible? Journal of Politcal Economy, 105(3), 609–621.
Dalton, R. J., Bürklin, W., & Drummond, A. (2001). Public opinion and direct democracy. Journal of Democracy, 12(4), 141–153.
Diamond, P., Hausman, J. A., Leonard, G. K., & Denning, M. A. (1993). Does contingent valuation measure preferences? Experimental evidence. In J. A. Hausman (Ed.), Contingent valuation. A critical assessment (pp. 41–89). Amsterdam: Elsevier Science Publishers B.V.
Diamond, P. A., & Hausman, J. A. (1994). Contingent valuation: Is some number better than no number? Journal of Economic Perspectives, 8(4), 45–64.
Donovan, T., & Karp, J. A. (2006). Popular support for direct democracy. Party Politics, 12(5), 671–688.
Dyck, J. J., & Baldassare, M. (2009). Process preferences and voting in direct democratic elections. Public Opinion Quarterly, 73(3), 551–565.
Eichenberger, R., & Serna, A. (1996). Random errors, dirty information, and politics. Public Choice, 86, 137–156.
Feld, L. P., & Kirchgaessner, G. (2000). Direct democracy, political culture, and the outcome of economic policy: A report on the swiss experience. European Journal of Political Economy, 16, 287–306.
Feld, L. P., & Matsusaka, J. G. (2003). Budget referendums and government spending: Evidence from Swiss cantons. Journal of Public Economics, 87(12), 2703–2724.
Frey, B. S. (1994). Direct democracy: Polito-economic lessons from Swiss experience. The American Economic Review, 84(2), 338–342.
Funk, P. (2016). How accurate are surveyed preferences for public policies? Evidence from a unique institutional setup. Review of Economics and Statistics (forthcoming).
Garrett, T. A. (1999). A test of shirking under legislative and citizen vote: The case of state lottery adoption. The Journal of Law and Economics, 42(1), 189–208.
Gerber, E. R. (1999). The populist paradox: Interest group influence and the promise of direct legislation. Princeton, NJ: Princeton University Press.
Hanemann, W. M. (1994). Valuing the environment through contingent valuation. The Journal of Political Perspectives, 8(4), 19–43.
Hersch, P. L., & McDougall, G. S. (1988). Voting for ‘Sin’ in Kansas. Public Choice, 57, 127–139.
Hessami, Z. (2014). Electoral rules for mayors and incentives to pork-barrel: Quasi-experimental evidence from german municipalities. Verein für Socialpolitik / German Economic Association, Annual Conference 2014 (Hamburg): Evidence-based Economic Policy.
Inglehart, R. (1990). Culture shift in advanced industrial society. Princeton: Princeton University Press.
Inglehart, R. (1999). Postmodernization brings declining respect for authority but rising support for democracy. In P. Norris (Ed.), Critical citizens: Global support for democratic government. Oxford: Oxford University Press.
Jenssen, A. T., & Listhaug, O. (2001). Voters’ decisions in the Nordic EU referendums of 1994: The importance of party cues. In M. Mendelsohn & A. Perkin (Eds.), Referendum democracy: Citizens, elites and deliberation in referendum campaigns (pp. 169–190). New York: Palgrave.
Kahneman, D., & Knetsch, J. L. (1992). Valuing public goods: The purchase of moral satisfaction. Journal of Environmental Economics and Management, 22(1), 57–70.
Kauder, B., & Potrafke, N. (2016). Supermajorities and political rent extraction. Kyklos, 69(1), 65–81.
Leininger, A. (2015). Popular support for direct democracy in europe. Paper presented at the ECPR Joint Sessions 2015, What citizens want from democracy: Popular Attitudes to Existing Political Processes and their Alternatives, Warsaw.
List, J. A. (2002). Preference reversals of a different kind: The “more is less” phenomenon. American Economic Review, 92(5), 1636–1643.
Lupia, A. (1994). Shortcuts versus encyclopedias: Information and voting behavior in California insurance reform elections. The American Political Science Review, 88(1), 63–67.
Matsusaka, J. G. (2005). Direct democracy works. Journal of Economic Perspectives, 19(2), 185–206.
Matsusaka, J. G. (2008). For the many or the few: The initiative, public policy, and American democracy. Chicago: University of Chicago Press.
Matsusaka, J. G. (2010). Popular control of public policy: A quantitative approach. Quarterly Journal of Political Science, 5(2), 133–167.
Murphy, J. J., Allen, P. G., Stevens, T. H., & Weatherhead, D. (2005). A meta-analysis of hypothetical bias in stated preference valuation. Environmental and Resource Economics, 30(3), 313–325.
Nalebuff, B. J., & Shachar, R. (1999). Follow the leader: Theory and evidence on political participation. American Economic Review, 89(3), 525–547.
Neill, H. R., Cummings, R. G., Ganderton, P. T., Harrison, G. W., & McGuckin, T. (1994). Hypothetical surveys and real economic commitments. Land Economics, 70(2), 145.
Noam, E. M. (1980). The efficiency of direct democracy. Journal of Political Economy, 88(4), 803–810.
Osborne, M., & Turner, M. (2010). Cost benefit analyses versus referenda. Journal of Political Economy, 118(1), 156–187.
Oster, E. (2014). Unobservable selection and coefficient stability: Theory and evidence. University of Chicago Booth School of Business Working Paper.
Pappi, F. U. (2011). CSU- und CDU-Waehlerschaften im Sozialstrukturellen Vergleich. Hanns-Seidel-Stiftung Akademie fuer Politik und Zeitgeschehen.
Piott, S. L. (2003). Giving voters a voice: The origins of the initiative and referendum in America. Columbia: University of Missouri Press.
Portmann, M. (2014). Parliamentary representation of citizens’ preferences: Explaining the differences between parliamentarians’ votes and popular referendum results (1. Aufl ed.). Neue Studien zur politischen Ökonomie. Baden-Baden: Nomos.
Portmann, M., Stadelmann, D., & Eichenberger, R. (2012). District magnitude and representation of the majority’s preferences: Evidence from popular and parliamentary votes. Public Choice, 151(3–4), 585–610.
Schlaepfer, F., Roschewitz, A., & Hanley, N. (2004). Validation of stated preferences for public goods: A comparison of contingent valuation survey response and voting behaviour. Ecological Economics, 51(1–2), 1–16.
Schlaepfer, F., & Hanley, N. (2006). Contingent valuation and collective choice. Kyklos, 59(1), 115–135.
Schneider, F., Pommerehne, W. W., & Frey, B. S. (1981). Politico-economic interdependence in a direct democracy: The case of Switzerland. In D. A. Hibbs, H. Fassbender, & R. D. Rivers (Eds.), Contemporary political economy: Studies on the interdependence of politics and economics, number 135 in contributions to economic analysis (pp. 231–248). Amsterdam: North-Holland Publishing Company.
Smith, D. A., & Tolbert, C. J. (2001). The initiative to party partisanship and ballot initiatives in California. Party Politics, 7(6), 739–757.
Stadelmann, D., Portmann, M., & Eichenberger, R. (2013). Quantifying parliamentary representation of constituents’ preferences with quasi-experimental data. Journal of Comparative Economics, 41(1), 170–180.
Stadelmann, D., & Torgler, B. (2013). Bounded rationality and voting decisions over 160 years: Voter behavior and increasing complexity in decision-making. PLoS One, 8(12), e84078.
Trechsel, A. H., & Sciarini, P. (1998). Direct democracy in Switzerland: Do elites matter? European Journal of Political Research, 33(1), 99–124.
Zaller, J., & Feldman, S. (1992). A simple theory of the survey response: Answering questions versus revealing preferences. American Journal of Political Science, 36(3), 579.
Acknowledgments
We would like to thank Florian Ade, Dirk Foremny, Benny Geys, Arndt Leininger, Serguei Kaniovski, Mark Kayser, Gebhard Kirchgässner, Katharina Hofer-Jaronicki, Manfred Holler, Peter Haan as well as anonymous referees and seminar participants at the ECPS 2015 meeting in Groningen, the IIPF at Dublin, the DIW Berlin and the FU Berlin for valuable comments and suggestions. Ronny Freier gratefully acknowledges financial support from the Fritz Thyssen foundation (Project: 10.12.2.092). The usual disclaimer applies. An earlier version of this article circulates with the title Voting for Direct Democracy: Evidence from a Unique Popular Initiative in Bavaria
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Arnold, F., Freier, R., Pallauf, M. et al. Voting for direct democratic participation: evidence from an initiative election. Int Tax Public Finance 23, 716–740 (2016). https://doi.org/10.1007/s10797-016-9398-z
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10797-016-9398-z