Abstract and Keywords
The existing framework for explaining variation in voter turnout and vote choice across multiple levels of government is more useful for addressing aggregate patterns than for addressing micro-level behaviour. To better understand the aggregate outcomes of interest, this chapter proposes an examination of individual voter behaviour and strategies of party elites that take account of the incentives provided by multiple electoral arenas. In doing so, the chapter investigates the variation in both party and voter behaviour that is often overlooked in existing work. In the presence of multiple electoral arenas, the behaviour of both party elites and voters in one arena is affected by what happens in another. The decisions of party elites and voters are shaped by the incentive structure that is produced by the combination of multiple arenas with different electoral rules.
Scholars of electoral politics, along with the media and voters, focus overwhelmingly on national elections. Yet in contemporary democracies, political authority is dispersed across several levels of government. Voters elect representatives in subnational, national, and, in some cases, supranational elections, and policy making occurs at each of these levels. We can think of this multi-level governance as a set of ‘functional jurisdictions that enjoy some degree of autonomy within a common governance arrangement and whose actors claim to engage in an enduring interaction in pursuit of a common good’ (Zürn et al. 2010: 4). The allocation of power to different and autonomous levels of government in multi-level systems entails the existence of multiple electoral arenas for the selection of office holders. All of the questions that scholars of electoral politics ask, such as how well are citizens represented, or what determines voter behaviour, are relevant across multiple levels of political competition. The focus on national elections necessarily overshadows the link between voters, policy makers, and policy outcomes in other political arenas, leaving scholars with partial answers to their questions.
The importance of electoral arenas beyond the national level has been increasing in most countries over time. The most common expression of multi-level governance is the dispersion of political authority from the central state government to regional and local governments in order to increase the efficiency and responsiveness of government (Tiebout 1956). Decentralization of power has been a worldwide trend in recent decades, sometimes referred to as the ‘era of regionalization’ (Hooghe et al. 2010: 52). At the beginning of this century, ‘some 95 percent of democracies now have elected subnational governments, and countries everywhere—large and small, rich and poor—are devolving political, fiscal, and administrative powers to subnational tiers of government’ (World Bank 2000: 107).1 Similarly, in their analysis of the experience of regional government in a sample of forty-two democracies or quasi-democracies between 1950 and 2006, Hooghe et al. (2010: 53–5) show that
(p.2) not every country has become regionalized, but, where reform has taken place, it has generally been in the direction of greater regional authority…Of the 42 countries in our dataset, 29 saw an increase in the regional authority index over the period of evaluation, eleven saw no change, and two show a [weak] decline.
The 2008 economic and financial crisis has not substantially changed this picture.2 Governance at subnational levels, then, has been growing in importance and the recent financial crisis has not led to a recentralization process.
As earlier centralization of economic and political institutional arrangements has slowed, and in many places reversed, regional elections have risen in importance (Schakel 2013: 632). In the forty-two countries examined by Hooghe et al. (2010), the number with directly elected regional assemblies increased from eleven in the 1950s to twenty-eight in 2006. Data provided by the World Bank (2000) for 127 countries in 1999 show that some type of subnational election (regional or local) is held in at least 108 of them, while forty countries hold regional elections.3
At the same time, at the supranational level, political and economic power in European Union (EU) countries is increasingly shared between national and supranational institutions. From the initial common market in coal and steel products, the EU has evolved into an economic, social, and political union with significant authority over many areas of public policy (see Hix and Høyland 2011, chapter 1). Over time, the size of the EU has dramatically changed. The EU began in the 1950s with six states (about 175 million inhabitants) and has enlarged to twenty-eight in 2013 (more than 500 million inhabitants). As a result, the European Parliament (EP) has elected representatives from a growing number of countries: nine in the first EP election in 1979, and twenty-eight in the 2014 election. Meanwhile, in South America, the direct election of the national members of the Mercosur Parliament (PARLASUR) will take place in every member of the Mercosur trading bloc before 2020.
The existence of these multiple electoral arenas suggests that scholars should be cautious about examining single levels in isolation, because the behaviour of both party elites and voters in one arena is not independent of what happens in another. For instance, characteristics of presidential elections have been shown to affect the number of legislative parties (M. Golder 2006). Rather than considering only the incentives provided by the rules in a single election, it is important to ask if the decisions of party elites and voters are shaped by an incentive structure produced by the combination of multiple arenas with different electoral rules. For example, scholars often refer to a ‘coattail effect’, where the performance of legislative parties is largely determined by the performance of each party’s presidential candidate. With some exceptions, this coattail effect runs from the more powerful body to the less powerful one.4 In presidential systems, ‘[voters] typically pay more attention (p.3) to presidential campaigns and use the party of their presidential candidate as an information shortcut to help them decide how they should vote in legislative elections…legislative candidates have even more of an incentive to coordinate their own campaigns with their party’s presidential candidate’ (M. Golder 2006: 35). Note that neither voters nor candidates, in this account, consider the legislative and presidential election arenas independently.5
When we talk about multi-level electoral politics, then, we typically refer to the extent to which the behaviour of parties and voters in one electoral arena is affected by what goes on in another electoral arena within the country. When political actors do not only respond to the incentives provided by the election at hand, but are also influenced by coattail effects from another electoral arena, the outcomes of the election at hand differ from what would have been observed in a pure and isolated election. Multi-level governance, then, should matter for our understanding of electoral politics. Only if electoral arenas are independent can a researcher safely ignore the impact of multiple arenas.
To say that one should not ignore the impact of multiple arenas is scarcely a controversial point; many scholars agree that elections conducted on different levels should not be examined in isolation. The most common approach is to examine election results at the aggregate level, and the most influential framework for explaining multi-level electoral politics has been the second-order election (SOE) model. Originally formulated by Reif and Schmitt (1980) when analysing the first EP election in 1979, the SOE model argues that not all elections within countries are equally important in the eyes of parties and voters.6 Depending on what is at stake, elections can be divided into first-order elections and second-order elections. Elections that decide who holds national executive office, the most important electoral prize, are first-order contests, while all other elections (i.e., EP, regional, and local or upper-house elections), in which no actual executive power is at stake and/or the body to be elected has little power, are second-order races. In Reif and Schmitt’s words (1980: 8–9):
[T]he ‘first-order’ elections in parliamentary systems are the national parliamentary elections…There is a plethora of ‘second-order elections’: by-elections, municipal elections, various sets of regional elections, those to a second-chamber and the like…Many voters cast their votes in these elections not only as a result of conditions obtaining [in]…the second-order arena, but also on the basis of factors in the main political arena of the nation.
At an aggregate level, the SOE model is supported by three empirical regularities: (i) turnout is lower in second-order elections than in first-order elections,7 (ii) small parties do better in second-order elections than in first-order elections, and (iii) parties in national government do worse in second-order elections than in first-order elections.
(p.4) In the original SOE model formulated by Reif and Schmitt, there are two factors driving coattail effects. First, the influence a national election exerts on a second-order election is a function of the proximity of the two elections: the less time elapsed between the first- and second-order election, the greater the coattail effect. The most proximal second-order elections are those held at the same time as first-order elections, while the least proximal second-order elections are those held at midterm.8 Second, the influence of a first-order election on a second-order election depends on the gap in the powers of the bodies to be elected: the greater the gap, the greater the coattail effect. Accordingly, coattail effects should be greater when bicameralism is asymmetrical than when it is symmetrical (Cox 1997: 20–7), and in EP elections compared to regional elections, as no executive office is chosen in the former. Given the increasing powers of the EP and the trend towards decentralizing power within states, though, coattail effects should decrease somewhat in EP and regional elections as time goes by (see Blais et al. 2011).
As illustrated in the literature on presidential systems (Samuels 2000, M. Golder 2006), the distinction between first- and second-order elections within a country entails the existence of coattail effects running from the more important body to the less important one/s.9 In a parliamentary system, we would expect national legislative elections to be the most important while in a semi-presidential system (like France) the presidential elections are expected to influence electoral outcomes at other levels. For instance, Clark and Rohrschneider (2009: 646) refer to coattail effects in the European arena as the transfer hypothesis: ‘national considerations dominate voters’ decisions even in EU elections. Voters presumably rely on the popularity of national government to decide whom to support in EU elections; they rely on the national economy to allocate their support accordingly, and so on.’
In regional elections, people are concerned with choosing a government (or at least influencing the formation of one), and whether regional elections should be considered to be second-order elections in the same way that EP elections are is a matter of dispute (see Dandoy and Schakel 2013, Jeffery and Hough 2006, or Schakel 2015). For instance, Schakel and Jeffery (2013) found strong second-order effects for some regional elections, but clearly not for the majority of them. Indeed, for some groups of voters, subnational elections may actually be more salient than national elections (Cutler 2008a). However, Heath et al. (1999) argue that the SOE model applies not only for EP elections, but also, albeit with less force, to the British local elections. In a similar vein, based on data from regional elections in six European countries, Jeffery and Hough (2006: 252) conclude that ‘the general finding, then, is that most substate elections do indeed appear to be second-order, subordinate to voters’ considerations of state-level politics’. In sum, the largest coattail effect running from the first- to the second-order election should emerge under the following conditions: elections are simultaneous, and there is a significant difference in the powers of the electoral bodies.
The SOE model is a useful framework for thinking about multi-level electoral politics. Once we know which office is the most important prize in an electoral democracy, the direction of the coattail effect can be easily determined. Differences in the coattail effect over time seem to be explained by the proximity of the first and second-order elections, while the gap in the powers of the bodies to be elected helps us to understand why coattail effects vary across countries.
However, the SOE model neglects an important factor affecting aggregated coattail effects in second-order elections: the difference in electoral systems across the different levels of elections. We know that electoral rules provide incentives for voters and parties to engage in strategic behaviour (such as strategic voting and strategic entry). It has been shown that the coattail effect running from the first-order election to the second-order election will be moderated if the permissiveness of the two electoral systems differs. As explained by Cox (1997: 21) when analysing upper-house and lower-house elections,
if a party can run and elect candidates under the more permissive system, it may decide to run candidates in the other systems as well—not to win seats, perhaps, but to keep its electoral organization in good trim, to establish its blackmail potential, or other reasons. In this case, the party system in each chamber should be influenced by that of the other, in such a way as to lessen observed differences.10
In line with this argument, M. Golder (2006) shows that the effect of presidential elections on reducing legislative fragmentation declines as the number of presidential candidates increases. It is generally the case that electoral rules are less permissive in national elections than in regional (Lago and Montero 2009: 183, Dandoy and Schakel 2013) or EP ones (Hix and Høyland 2011: 150–1).11 This line of research implies that the greater the difference in the electoral systems between the first- and second-order elections, the greater the coattail effect.
The SOE model stops short of providing a satisfactory story of multi-level electoral politics because the micro mechanisms that generate the coattail effect are not made explicit. That is, although the SOE model claims that the most important electoral arena drives coattail effects, it does not present an explicit account, based on individual behaviour, of why this occurs. Without these generative mechanisms, based on the actions of individual actors, parties, and voters, the SOE model is a ‘black-box’ explanation. For instance, if the largest difference in outcomes between first and second-order elections is observed when the latter are held around the midterm of the national election (p.6) cycle, it is not possible to determine whether the explanation is a coattail effect, as the SOE suggests, or whether this is simply a reflection of the fact that as more time has elapsed between the first- and second-order election the economic and political situation is more likely to have changed. Similarly, the different composition of electorates might explain the correlation between the strength of the coattail effect and the gap in the powers of the bodies to be elected.
In fact, the outcome of an election is merely one observed implication of the SOE model and, following King et al.’s advice (1994: 31) it should not be privileged over other observable implications. This is precisely Hobolt and Wittrock’s point (2011: 29) when they argue that the widespread use of aggregated-level data to test the SOE leads to the problem of observational equivalence because the empirical evidence is compatible with different explanations. Even if we are primarily interested in the aggregate level of analysis (i.e. election results), leverage about the SOE model’s veracity can be gained by looking at individual-level data if one has a theoretical argument about individual voting behaviour that is consistent with SOE outcomes.
However, the lack of micro mechanisms in the SOE model means that it is not helpful in explaining individual-level electoral behaviour. When all individuals are exposed to the same situation, it is not possible to explain why only some of them are affected by the coattail effect. The SOE only covers the macro-to-micro transition (in Coleman’s (1986) terminology, that is, a situational mechanism), and then at most it can explain aggregate differences in the coattail effect; how an individual assimilates (or not) the impact of the coattail effect is clearly out of reach. As Marsh and Mikhaylov (2010: 10) note,
[t]he evidence that the results of EP elections typically reflect the expectations initially developed by Reif and Schmitt is overwhelming…What is more problematic is our understanding of how this pattern comes to light. Reif and Schmitt do not offer an explicit theory of a European voter. Their work is essentially an aggregate-level one. Aggregate patterns, however, require an explanation in terms of individual choice.
Not surprisingly, the lack of robust micro foundations in the SOE model has recently fostered a flurry of mechanism hunting when explaining multi-level electoral politics or, more specifically, when explaining how vote choices in EP and regional elections are shaped by being second-order elections. A non-exhaustive list of proposed micro mechanisms includes instrumental and expressive motivations (Oppenhuis et al. 1996, Heath et al. 1999), second-order elections as barometer (Anderson and Ward 1996) or balancing elections (Kedar 2006, Kern and Hainmüller 2006), the mobilization deficit in second-order arenas (Weber 2007), the transfer hypothesis (Clark and Rohrschneider 2009), midterm effects (Weber 2011), a two-dimensional spatial model (left-right and European integration) (Hobolt and Wittrock 2011), (p.7) the role of personal motivation (Söderlund et al. 2011), the interaction between the state of the economy and party type (Schakel 2015), and the timing of elections (Schakel 2013, Schakel and Dandoy 2014).
Although proposing micro foundations of voting behaviour that are empirically consistent with the SOE model is useful, this list of micro mechanisms does not mean that we have a good understanding of the phenomenon. The leverage these mechanisms provide over multi-level electoral politics is low, because these possible explanations require a host of causal variables: much is required to explain a little. The empirical evidence typically used in such studies does not come from different electoral arenas, but from single levels of elections. Despite the worldwide prominence of multi-level electoral politics, there are no comparative studies covering supranational, national, and regional arenas. In the specific case of regional elections, comparative contributions are sparse and restricted both in their theoretical scope and in the number of cases examined (Schakel 2013: 632). Clearly, scholars still have ample opportunities to make theoretical and empirical contributions to the study of multi-level politics.
1.3. Re-Examining Multi-Level Electoral Politics
The objective of this book is to explain how party and voter behaviour in a given election is affected by the existence of multiple electoral arenas for the selection of political office holders within the country. We study both party strategy and vote choice in elections at three levels of government—European, national, and subnational—in three well-established democracies from 2011 to 2015: France, Germany, and Spain. These countries have been chosen in order to maximize variation in electoral arrangements within and across them. For both national legislative and subnational elections, France has a majority (two-round) system, Germany has a mixed system, and Spain utilizes proportional representation. However, for EP elections, the three countries all use proportional representation.
Our research is designed to maximize leverage over multi-level politics in two ways. The first is that the focus is on the choices that citizens make in multiple electoral arenas within countries. However, as voting behaviour is intertwined with the decisions that parties make and the coordination problems generated by electoral systems, our theory about how voters’ choices are affected by coattail effects addresses three elements—entry decisions by party elites, campaign strategies, and voter choices—that generate a variety of observable implications.
(p.8) First, citizens’ choices cannot be understood without first addressing the entry decisions made by party elites. Without understanding where this menu of choices comes from, it is not possible to fully make sense of voter behaviour. Well before the election occurs, political parties have to decide, in every district, whether they should enter the race independently, engage in some form of pre-electoral coordination with another party, or stay out. Since electoral systems vary across electoral arenas within countries, the supply of parties may differ, which suggests that voting behaviour may also differ. Second, once parties enter the race, they have to define their campaign strategies with the goal of influencing voters’ decisions. That is, they have to select policy positions, define the salience of issues, allocate their resources across districts, and select their candidates. Again, there are no reasons to expect that parties follow the same campaign strategy in each kind of election. If parties’ campaign strategies vary across electoral arenas within countries, voting behaviour should change in response. Finally, once the menu of choices is set, citizens first have to decide whether to vote or not, and second, those who vote have to choose a specific party. As Dalton and Wattenberg note (1993: 193–4), ‘any discussion of voting behaviour is ultimately grounded on basic assumptions about the electorate’s political abilities—the public’s level of knowledge, understanding, and interest in political matters’. Given that voters’ political abilities can vary across electoral arenas, voting behaviour could change accordingly. Although we do not deny the existence of stable patterns of electoral behaviour, at the end of the day ‘every election is different. Candidates and issues change…the dominant issues were not identical in any two elections…Sometimes even the parties are different’ (Beck 1986: 263).
Another way in which we maximize leverage is to examine multiple arenas within and across countries. To the best of our knowledge, this book provides the first systematic analysis of multi-level electoral politics at three different levels—subnational, national (lower house), and EP elections—across countries, using original qualitative and quantitative data. This allows us to examine many implications of our theoretical argument. In order to keep the composition of the electorate manageable, while allowing party systems to differ substantially, we focus on two specific regions in each of our three countries: Île-de-France and Provence in France, Lower-Saxony and Bavaria in Germany, and Catalonia and Madrid in Spain. For every election in each region, two analyses have been conducted: an analysis of party strategies through a systematic content analysis of campaign materials, as well as semi-directed interviews with campaign managers, and an analysis of voter behaviour using online pre- and post-election panel surveys with the same technical characteristics and similar questionnaires. The data collection was designed with the goal of being able to compare across countries and electoral arenas.
(p.9) We are able to use these data to examine the ways in which voter and party behaviour is connected across subnational, national, and European elections. In doing so, we take account of institutional features such as electoral rules as well as the historical and political context that shapes the nature of the party system(s) across levels. We also consider variation in party characteristics and voter characteristics, because we do not expect that all parties, or all voters, will behave in the same way, either within or across electoral arenas.
In Chapter 2, we present a theoretical framework for examining multi-level elections. Our goal is to explain how electoral processes in different electoral arenas can influence one another by focusing on the incentives facing political actors in these arenas. Party elites, for example, have mobilization strategies that take account of the features of each electoral arena, and make decisions about how to allocate their resources across these arenas accordingly. Larger or more nationalized parties may make very different choices compared with smaller, under-resourced, or regionally based parties. Voters, in turn, may face different incentives in one electoral arena versus another, both in terms of whether to turn out to vote, and in terms of which party to support once in the polling booth. However, not all voters feel connected to each electoral arena in the same way. For some, their identity and the issues they most care about are linked to politics at the national level. For others, the regional level may offer the political community and political issues that most resonate with them. In some EP elections, a sense of European identity may affect a voter’s behaviour.
To get a sense of the institutional variation across our three countries and six regions, in Chapter 3 we discuss electoral rules, regime type, and party systems in our cases. For example, our data come from elections held under majoritarian, proportional, and mixed electoral rules, and in both parliamentary and semi-presidential systems. The party systems vary in the extent to which they are nationalized, with some countries having important differences at the subnational level, such as regional parties that compete in a single region, and other countries having party systems that look similar across multiple arenas. To put the elections we study in context, we also discuss the political and economic situation facing citizens of the different regions during the period under analysis. Notably, our analyses cover a period of financial stress in these countries. Here, too, though, we see variation in how much different countries, and regions within countries, suffered from the global financial crisis that began shortly before our study, and we expect parties and voters to vary accordingly in their reaction to the crisis.
We address the decision-making calculus of the party elites in Chapter 4, turning our attention to individual voter decisions in Chapters 5 and 6, where we examine turnout and strategic voting, respectively. Of course, the decisions of party elites when deciding how to allocate their resources to mobilize their electorate are based on their expectations of how voters will behave. And the (p.10) mobilization tactics employed by the parties will affect turnout, so in this sense Chapters 4 and 5 tell us about two different sides of the same coin. In both sets of analyses, we consider how differences in party and voter characteristics affect these decisions; we do not expect that all parties make the same choices, or that voters form a monolithic bloc with the same preferences and attitudes toward voting across the three electoral arenas. Chapter 6 focuses on voters’ decisions when they go to the polls, specifically, whether they are likely to cast a sincere or a strategic vote. We caution against a naïve approach that assumes that voters are voting strategically if they vote for different parties in different levels of elections, and explore the nuances of this decision in more depth.
Finally we look at the link between voters, elected politicians, and policy outcomes more broadly. In Chapter 7 we ask how voters evaluate the performance of their national, subnational, and European governments, and to what extent this evaluation affects their voting behaviour. On the one hand, asking voters to vote, and hence, evaluate governments in different electoral arenas gives voters a tremendous amount of input into which parties get to enter government at each of those levels. On the other hand, requiring voters to assess which level of government is responsible for which policy outcome can be difficult. As Gschwend (2008: 230) notes,
[Multi-level systems of governance] provide several challenges and important opportunities for electoral accountability and for our understanding of representative democracy. Nevertheless multi-level systems of governance also raise serious concerns about their democratic deficit. Not only citizens but also elected MPs find it hard to attribute responsibility to certain actors correctly when actual policy-making processes are obfuscated by the number of state and non-state actors, lobbyists, specialists and the like who participate in it. This, of course, has important consequences for the legitimacy of the policymaking process.
We focus on two policy outcomes in particular. One is the state of the economy, as viewed by the voters. We expect that the extent to which the voters attribute blame or credit for economic outcomes to the government at any one level will depend on whether the voter believes that the government at that level plays an important role in shaping the economy. We also examine voter opinions about corruption in government, again across all three levels. This is an issue that should affect voter satisfaction with, and trust in, their democratic institutions.
We conclude with a discussion summarizing the results of our investigation into multi-level elections in France, Germany, and Spain. We emphasize the importance of taking seriously the heterogeneity both of voters and of parties in our three countries in analysing how these political actors compete and vote (p.11) in subnational, national, and European elections. One clear conclusion is that scholars should be encouraged to develop a habit of asking about multiple electoral arenas when surveying voters, rather than focusing on a single level (typically the national one). The evidence we present in the following chapters suggests that there is a great deal of interesting variation across party and voter behaviour, and that there is ample room to extend and build on this current project.
(1.) In Chapter 3, we discuss different kinds of decentralization, and the extent to which each of our countries can be considered to be decentralized along de facto or de jure grounds. De facto decentralization can be thought of as fiscal decentralization, while de jure decentralization can be thought of as federalism (versus a unitary state). See Bednar (2009) for a detailed discussion of these distinctions.
(2.) According to the OECD (2013: 14–15), subcentral tax revenues and intergovernmental transfers have remained almost stable since 2008, while the subcentral fiscal consolidation has been achieved mainly through spending cuts.
(3.) By ‘regional’, we mean elections at the level of the state, province, region, department, or other elected entity between the local and the national government.
(4.) A coattail effect occurs when ‘the fortunes of electoral parties are tied to the fate of their party’s presidential candidate’ (M. Golder 2006: 35). In some circumstances, scholars also find evidence of ‘reverse-coattail’ effects (see Ames 1994, Cutler 2008a, Lago and Montero 2009, Samuels 2000, Gélineau and Remmer 2006, Bechtel 2012, or Garmendia Madariaga and Ozen 2015).
(5.) Although the coattail literature typically focuses on presidential regimes, for example in Latin America or the United States, its central logic can be applied to any multi-level electoral setting. Note also that although a ‘coattail effect’ often has a positive connotation, the effect could be positive or negative depending on how the national incumbent has been performing.
(7.) This difference in turnout presupposes that the elections are not held at the same time. If multiple levels of elections are held concurrently, distinguishing between turnout levels in that election makes little sense.
(11.) Since the 1999 election, all EU member states have been required to use some form of proportional representation for EP elections.