Descriptive and Substantive Representation of Poor Citizens in Switzerland

Abstract

Over the past decade, a large number of studies have shown that in advanced democracies policymakers are more likely to represent the preferences of high-income citizens than the preferences of low-income citizens (e.g., Gilens 2005, 2012; Bartels 2008; Ellis 2012; Giger, Rosset and Bernauer 2012; Hayes 2012; Rigby and Wright 2013). In a quest to understand this income bias in substantive representation, several authors have recently suggested that the persistent descriptive underrepresentation of low-income groups in policymaking institutions may explain why political decisions are so frequently skewed toward the preferences of the affluent (e.g., Carnes 2012; Carnes and Lupu 2015).2 However, although this is a plausible explanation, many previous studies dealing with the consequences of descriptive representation have failed to find an effect of policymakers' incomes or social class backgrounds on their behavior in office (e.g., Matthews 1984; Norris and Lovenduski 1995). In this essay, we argue that descriptive representation may indeed have an effect on policymaking, yet for it to be consequential three necessary conditions must be satisfied. We explain in the next section what these conditions are. We then present research that we have recently conducted on representation in the Swiss parliament. Our analyses show that although Swiss legislators coming from lower-income backgrounds appear to have distinct preferences over policy, these differences in preferences do not translate into distinct behavior in the policymaking arena. Finally, drawing upon the three necessary conditions described below, the last section provides some suggestions as to why we fail to find an effect of descriptive representation on legislative behavior in the current Swiss context. The descriptive representation of different income groups may be important simply on symbolic grounds, since the legitimacy and authority of a “representative” body such as parliament can be undermined if it fails to reflect the composition of the population that it is intended to represent (e.g., Norris and Lovenduski 1995). If we assume, however, that constituents care primarily about policy outcomes (e.g., Downs 1957), rather than the composition of parliament, descriptive representation is important only to the extent that it affects legislative behavior.3 Based on this assumption, and following in part Carnes and Lupu (2015), we argue that there are three necessary conditions that must be satisfied for descriptive representation to have an effect on legislators' policymaking actions. First, legislators must share policy preferences with the group they represent descriptively, at least for a subset of the policies on the political agenda. Such preference congruence can be the result of two mechanisms. On the one hand, people with similar background characteristics may have “shared experiences,” which in turn promote similar interests and preferences over policy (Mansbridge 1999). By merely following their personal preferences, descriptive representatives then further not only their own interests but also those of the social groups they represent. On the other hand, congruence between the preferences of representatives and citizens belonging to the same socioeconomic group may be the product of responsiveness (Achen 1978). If the former have incentives to respond to the policy preferences of the latter (for instance because of reelection concerns; see, e.g., Mayhew 1974), the represented can “induce” preferences in their legislators (e.g., Snyder and Ting 2005). Second, much of the literature on legislative politics assumes that multiple principals can influence the voting behavior of legislators. Most important in this regard are party discipline and constituent pressure (e.g., Levitt 1996; Burden 2007; Carey 2009). Consequently, although legislators' behavior may be constrained by party leaders and constituents who do not belong to their descriptively represented groups, for descriptive representation to be relevant these principals must not determine legislators' actions completely. In order to be able to act according to the preferences of the groups they represent descriptively, representatives must therefore have some discretion in the policy decision-making process (see also Carnes and Lupu 2015). Third, legislators who descriptively represent a particular group must be presented with choices in the policymaking process that separate them from other legislators who do not represent the same group descriptively but are otherwise similar (e.g., in terms of party affiliation). If this is not the case, it is difficult to disentangle the effect of descriptive representation from the effects of other characteristics shared by the descriptive representatives and other (similar) representatives. Provided that these three conditions are satisfied, we expect descriptive representation to matter for legislative behavior. We turn next to a discussion of our research assessing whether and to what extent socioeconomic characteristics affect legislators' attitudes and behavior in the Swiss parliament. The Swiss parliament is a militia parliament. Hence, although the compensations legislators receive for their parliamentary activity may suffice to cover the cost of living, many retain a professional activity during their parliamentary mandates. Members of the Swiss parliament are therefore likely to have relatively diverse economic conditions, not only as a result of the different professions they exercised before entering parliament and the different levels of accumulated wealth, but also because their incomes can vary substantially depending on the professional activities conducted during their mandates. An important argument put forward by proponents of the militia system is that it allows for a more diverse composition of parliament and a closer connection between legislators and their constituents. However, in practice, the militia system does not seem to enhance the descriptive representation of different social classes, as the Swiss parliament is predominantly composed of liberal professionals, entrepreneurs, and high-rank civil servants (Pilotti et al. 2010, Pilotti this debate). These two features of the Swiss parliament – diversity in legislators' income levels, combined with a significant overrepresentation of representatives coming from high-income backgrounds – make it an interesting case for analyzing the link between the descriptive and the substantive representation of economically defined groups of citizens. In this essay, we thus report on two recent studies dealing with descriptive representation and its consequences in Switzerland. The first study focuses on policy preferences among candidates in the Swiss Federal elections of 2007 (Rosset 2013a, pp. 109-133). It is based on data from the candidate survey conducted as part of “Selects,” the Swiss electoral study. For this survey, all candidates to the two chambers of the Swiss parliament have been contacted and asked to answer a paper or online questionnaire regarding their political engagement, campaign activities, policy preferences as well as their socioeconomic background. Based on these data, the study analyzes candidates' preferences with regard to redistribution, using a question asking to what extent respondents agree with the statement that the state should do more to reduce income differences between the rich and the poor. The main independent variable of interest is the candidates' monthly household income. The analysis reveals that controlling for party affiliation, gender and education, income – even though measured only crudely by seven categories – has a significant negative effect on support for redistribution. In other words, candidates from the same party are on average less favorable to redistribution of income by the state if they themselves have higher incomes. Consequently, since citizens on average have both lower incomes and policy preferences that are much more favorable towards redistribution than their representatives (Rosset 2013b), this result suggests that the unequal descriptive representation of income groups in parliament could explain the unequal representation of their preferences on redistribution. The second study focuses on legislator responsiveness. It takes advantage of Switzerland's direct democratic system, in which parliament votes on each policy proposal that is put to a popular vote (Wüest and Lloren 2014). This institutional setting thus provides a unique opportunity to compare legislators' and citizens' policy preferences on identical scales.4 The empirical analysis consists of two steps. First, based on data from surveys carried out after popular votes (“Vox” surveys), the study examines whether citizens' policy preferences vary according to income. In line with previous research findings (Gilens 2009; Rigby and Wright 2011), this analysis shows that with regard to economic issues the poor tend to have more left-leaning preferences than more affluent citizens. After having shown that the preferences of poor and rich citizens differ, the study then assesses in a second step whether legislators with different socioeconomic backgrounds respond unequally to the policy preferences of different income groups. To do so, Wüest and Lloren (2014) regress legislators' voting decisions in parliament on three sets of independent variables: (i) indicator variables showing whether a legislator has a lower-income, middle-income or higher-income occupation (see Pilotti, Mach and Mazzoleni 2010), (ii) variables measuring the preferences of poor, middle-income and rich citizens, respectively, in a legislator's reelection constituency and (iii) a set of indicator variables to control for legislators' party affiliations.5 The analysis reveals, first, that representatives are more responsive to higher-income groups and that the preferences of the less affluent are therefore underrepresented in the legislative arena. Second, and most importantly for this essay, representatives occupying low-wage jobs are not more likely to cast a left vote than their colleagues who have middle- and higher-income occupations. Thus, conditional on party affiliation, legislators' socioeconomic backgrounds do not affect their voting decisions. Finally, the results show that legislator responsiveness to income group preferences does not depend on party affiliation. Consequently, this second study suggests that occupational income has no effect on legislators' behavior in final passage votes and that the overrepresentation of well-off representatives in parliament does not contribute to explaining the lack of congruence between legislators and lower-income citizens. At first sight, this finding contradicts a growing body of research that politicians' personal such as gender or ethnic influence their political behavior at various stages of the legislative process, thus enabling marginalized groups to voice their political concerns more accurately (Lefkofridi, Giger, and Kissau, 2012; Lloren 2015b; Preuhs 2006; Swers 2002). Some might argue that grouping legislators' according to income could be an improper measure to grasp the effect of one's position in the social structure on political decision-making (Carnes 2012; this debate; Mansbridge this debate). Accordingly, the politics of presence (Philipps 1995) postulates that personal features matter only to the extent that they generate shared social experiences, such as discrimination. Indeed, although a law student and a manual worker can be found in the same income brackets, no one would expect them to have similar social experiences or life chances, and, in turn, share a distinct set of policy preferences. However, the study carried out by Rosset (2013a) and discussed in the previous section shows that personal income determines candidates' preferences with regard to redistribution. Our findings presented above have shown that Swiss legislators' socioeconomic backgrounds are related to their attitudes towards income redistribution, yet do not seem to translate into specific patterns of voting behavior. Hence, although more research is needed to better understand the relationship between representatives' preferences and policymaking actions, we conclude this essay by offering some suggestions as to why legislators' socioeconomic characteristics do not appear to affect their behavior in parliament. As set forth in the introduction, three conditions must be satisfied for descriptive representation to have consequences for legislative behavior. Whereas the first of these conditions – that preferences of legislators are close to the preferences of the socioeconomic group they descriptively represent – is obviously satisfied, the other two are potentially violated. The second condition is violated if legislators' behavior is completely determined by party leaders and/or constituents not belonging to the groups they represent descriptively. In Switzerland, parties are decentralized organizations and legislators do not attach much importance to party loyalty as a guiding principle for their actions (Loewenberg and Mans 1988). However, this does not mean that they do not act in a cohesive manner and recent research suggests that party unity is higher on final passage votes and votes on salient issues (Traber et al. 2014). Therefore, the small influence of socioeconomic characteristics on legislators' voting behavior might be explained by the fact that their behavior is largely determined by party positions in votes on economic issues which are arguably salient. With regard to constituent pressure, we first note that given the large size of the underrepresented group of low-income citizens, legislators have a priori little electoral incentives to respond exclusively to the preferences of the more affluent. However, there is a clear class bias in political participation, which might explain why representing richer citizens could be of particular importance for reelection-seeking legislators (Rennwald 2014). Further, given the lack of party finance regulations and the low level of party subsidies in Switzerland, it may also be that politicians attach more importance to the preferences of constituents with greater financial resources they can contribute to political campaigns. Finally, the third condition for descriptive representation to matter is violated if there are no alternatives on the political agenda separating descriptive representatives from other legislators who do not represent the same group descriptively but who are otherwise similar. While coalition patterns among Swiss parties tend to differ across policy domains, recent research shows that with regard to economic issues there is a clear left-right cleavage, with center and right-wing parties favoring less welfare state than left-wing parties (Afonso and Papadopoulos 2014). Economic policies, therefore, divide legislators strongly along party lines, blurring more subtle differences in preferences that may exist among legislators with different income levels within parties (see also Mansbridge this debate). Consequently, proposals on economic issues might simply be not heterogeneous enough to allow us to disentangle the effects that socioeconomic characteristics may have on legislative behavior. This discussion also intends to serve as a cautionary note. Although in our analysis we failed to find an effect of descriptive representation on legislators' behavior, this does not imply that descriptive representation is irrelevant for policymaking. Rather, we believe that its effect is context-dependent. When the context changes (e.g., when party pressure decreases or more heterogeneous proposals are voted on in parliament), descriptive representation may become more important in structuring legislative behavior. For example, women's descriptive representation in the Swiss National Council has increased substantive representation for women (Lloren 2015a). In that case, the three conditions discussed earlier were satisfied. Women legislators are more likely to favor women friendly policy than their male colleagues and to vote together – sometimes even against their party – to support women's interests, especially on policies that yield weak left-right polarization such as moral issues. Exploring such context dependencies for other politically marginalized groups thus constitutes a fruitful avenue for future research. Finally, we should note that legislators' socioeconomic backgrounds might affect behavior other than roll call voting. For instance, some scholars have argued that legislators are less constrained by party pressure in earlier stages of the legislative process, such as when sponsoring bills or in floor debates, and therefore have more discretion to act according to the preferences of groups they represent descriptively (see, e.g., Carnes and Lupu 2015; Lupu this debate). Anouk Lloren is a postdoctoral visiting fellow at Columbia University funded by the Swiss National Science Foundation. Within the field of political behavior, her work relates to political participation, representation, gender, and socio-economic inequalities. Jan Rosset is a postdoctoral researcher at the Mannheim Centre for European Social Research (MZES), University of Mannheim, with a fellowship from the Swiss National Science Foundation. His research focuses on the relationship between economic and political inequality and on democratic political representation. Reto Wüest is a PhD candidate in the Department of Political Science and International Relations at the University of Geneva. His research interests include legislative institutions and behavior, economic inequality and political representation, and formal theory.