Introduction to the Debate: Does Descriptive Misrepresentation by Income and Class Matter?

Abstract

This debate section features new research suggesting that class bias in the make-up of legislative bodies is a topic that deserves more attention by political scientists.1 Suppose that we drew, at random, an individual from the pool of adult citizens and another individual from the pool of elected representatives. Regardless of the country in which we carried out this experiment, the draws would likely yield two quite different individuals. Men are more likely to hold elected office than women and members of ethnic or racial minorities are less likely to hold elected office than members of “the majority.” Yet women have made significant inroads into politics in many countries over the last two or three decades and the same can be said for some minorities in some countries. Socio-economic characteristics related, at least loosely, to “class” arguably constitute a more enduring and more universal basis of descriptive misrepresentation. Everywhere, it seems, legislators and other elected officials tend to be disproportionately drawn from white-collar professions and from the business community; and the educational attainment, income and wealth of the average legislator exceeds that of the average citizen by a considerable margin. While political theorists and political scientists have devoted much attention to the under-representation of women and minorities in legislative bodies, they have been remarkably inattentive to socio-economic discrepancies between electorates and legislatures and the potential implications for the substantive representation of different citizens. As Noam Lupu notes in his contribution to this debate, the conventional wisdom of political scientists is that the social background and material circumstances of politicians are not terribly relevant to their behavior as legislators. The social background of elected representatives has arguably become a more salient question as income inequality has risen in rich democracies over the last two decades. Standard models based on self-interested income maximization imply that the median voter wants more redistribution as inequality rises or, alternatively, that more voters will support redistribution as inequality rises. Generally speaking, however, governments in more unequal countries tend to redistribute less, not more, than governments in more equal countries. And it is surprisingly difficult to identify recent instances in which elections can be said to have produced more redistributive government in response to rising inequality. While some scholars have sought to resolve this puzzle by elaborating new models of how inequality—or conditions associated with inequality, such as immigration and crime—affect demand for redistribution, others have responded by suggesting that inequality might render government less responsive to the preferences of low- and middle-income citizens.2 Briefly summarized in his contribution to this debate, Martin Gilens' recent book demonstrates that when the policy preferences of high-income earners diverge from those of low- and middle-income earners, US public policy responds to the preferences of high-income earners, but not to the preferences of middle-income earners, let alone low-income earners (Gilens 2012). The longitudinal component of Gilens' study suggests that US government was far less responsive to the preferences of high-income earners in the 1960s than it is today. Observing policy-making processes, Hacker and Pierson (2011) argue, similarly, that US politics has become increasingly pro-rich as the US has become a more unequal society. The US is surely a special case from the point of view of what Gilens calls the “outsize role of money,” but studies inspired by Gilens' work indicate that representational income bias is also a feature of West European democracies, notably Switzerland (Giger, Rosset and Bernauer 2012, Rosset 2013). In one way or another, the following essays all speak to the relevance of descriptive misrepresentation for biases in substantive representation based on education, occupation or income. Most of our contributors conceive this as an empirical question pertaining to the values, preferences and behaviors of individual legislators.3 As far as preferences are concerned, the empirics seem unambiguous and all contributors who address this issue essentially agree. In Switzerland, we are told by Anouk Lloren, Jan Rosset and Reto Wüest, income is associated with (less) support for redistribution among candidates for parliament in much the same way as it is among citizens. For the Netherlands, Armen Hakverdian reports that higher levels of education are associated with more support for multiculturalism among elected local officials as well as citizens. In the US and Latin America, according to Nick Carnes and Noam Lupu, legislators who held working-class jobs (or were union officials) before their election to public office consistently favor more progressive (redistributive and interventionist) policies than legislators who come from the business community or worked as private-sector professionals before their election. By contrast, our contributors disagree on the question of whether or not social background and material circumstances matter to the actual behavior of legislators and, perhaps, on the question of which kinds of behaviors it is most important to track. Carnes' essay refers us to a number of recent studies, including Carnes' own book (2013), demonstrating that occupation, income and wealth have statistically significant, substantively meaningful, effects on roll-call voting and other behaviors of state and federal legislators in the US (while controlling for the party affiliation of legislators and characteristics of the constituencies that elected them). Building on joint research with Carnes, Lupu presents similar evidence for Latin American countries in his essay. On the other hand, the cleverly designed study of Wüest and Lloren (2014), briefly summarized in their essay with Rosset, finds that, controlling for party affiliation, the outside income of Swiss legislators does not affect their voting on economic issues subject to popular referenda. While income predicts the votes as well as the preferences of Swiss citizens, it appears to matter only for the preferences of elected representatives, not for their voting behavior. Though Swiss parties are not famous for their discipline, one possible reason for the discrepancy between the findings reported by Lloren, Rosset and Wüest and by Carnes and Lupu is that Swiss parties are more disciplined than US and Latin American parties, leaving individual legislators with less discretion to vote based on their own values or interests. Lupu argues that occupation is more relevant than income and that the effects of class-based characteristics matter primarily for the behavior of legislators prior to final votes on legislative proposals. According to Lupu, it is in the early phases of the legislative process, when bills are drafted and discussed, that legislators are least constrained by pressures from voters, interest groups and party leaders. This important point resonates with previous scholarship on descriptive representation by gender and race. Following Thomas (1994), such scholarship has largely taken for granted that party affiliation is a powerful predictor of roll-call voting and has focused on other activities in which legislators are engaged: oversight by legislative committees and constituency service as well as sponsorship of legislative proposals (see, e.g., Minta 2011 and Broockman 2013). It seems plausible to suppose that the individual-level effects of coming from a certain class background are conditional on the share of the legislators with the same (similar) class background. Furthermore, descriptive representation may matter even in the absence of identifiable individual-level effects: having more legislators from working-class backgrounds (or more female legislators) may influence the behavior of all legislators. From this perspective, historical studies of the evolution of parliamentary recruitment by income and occupation, like the study of the Swiss Federal Assembly presented by Andrea Pilotti in his contribution to this debate, constitute an important complement to the focus on individual legislators in the work of Carnes, Lupu and others.4 While Carnes (2013) builds a compelling case for the claim that social background and material circumstances matter to the behavior of elected politicians in the US, the thorny question of the importance of descriptive representation relative to other causes of unequal responsiveness remains. As Gilens notes, Carnes finds that the occupational background of legislators matters to their positions on economic issues, but not to their positions on social (or cultural) issues. Yet Gilens' own analysis shows that income bias in government responsiveness holds across all policy domains. This suggests that descriptive misrepresentation provides, at best, a partial explanation of unequal substantive representation in the US. For Gilens, the real culprit is the dependence of candidates for public office on private donations to finance ever more expensive election campaigns. It is far from obvious, however, that campaign finance as an explanatory variable travels to other countries where inequality of substantive representation can also be observed (except perhaps to Switzerland). According to Hakverdian, the over-representation of well-educated individuals among elected Dutch officials is critical to the lack of congruence between legislation and the preferences of the average citizen on issues related to multiculturalism. Having puzzled over these essays as an ensemble, it seems to me that an important item on our collective agenda is to clarify, theoretically and empirically, how education, occupation and income relate to each other—how or when, under what conditions, they reinforce each other and add up to “class.” If legislators' behavior is at least partly determined by their own background and material circumstances, the obvious question becomes: Why do voters so consistently elect representatives who are better off than themselves? Carnes argues strenuously—and, to my mind, convincingly—that descriptive misrepresentation by class cannot be explained by a shortage of working-class individuals qualified for public office (or perceived by voters to have the necessary qualifications). More plausibly, working-class individuals are less able—also perhaps less willing—to bear the costs of running for and holding public office. As Carnes shows us, working-class representation in US state legislatures is positively associated with (state-level) union density. It seems highly plausible, as Carnes suggests, that financial support and mobilizational efforts by unions help working-class candidates overcome some of the material disadvantages that they face. It is less obvious why it is, as Carnes also shows, that working-class representation, at the level of US states, is negatively associated with legislative professionalization. This observation runs counter to Pilotti's claim that Switzerland's “militia parliament,” with very little monetary compensation for MPs, explains why so many Swiss MPs are businessmen and independent professionals.5 The costs of winning a legislative seat have increased massively over the last two or three decades in the US and legislative politics appears to have become more of a full-time job—indeed, a lifetime career—in most countries. If the costs of winning elections were what accounts for working-class under-representation, we would expect working-class under-representation to have become more pronounced over time, at least in the US. Somewhat surprisingly, however, Carnes' data shows that the working class has always been poorly represented in state legislatures as well as Congress and that very little changed in this regard over the twentieth century. Thus under-representation of the working class does not seem to be a plausible explanation of the apparent decline in government responsiveness to the preferences of working-class citizens in the US. The percentage of MPs from working-class backgrounds has declined in many West European countries since the 1950s (Best 2007: 100).6 Related to this, Alexiadou (2014) documents a sharp decline in the percentage of cabinet ministers with professional backgrounds as union officials. Arguably, the background of individuals in legislative leadership positions is particularly relevant for purposes of descriptive representation. More important for our purposes, it seems likely that the attitudes and behaviors of legislators with prior involvement in collective working-class organizations are more distinctive than the attitudes and behaviors of legislators who simply held a blue-collar job or lived in a working-class neighborhood before running for office. It is tempting, but perhaps mistaken, to reduce the question of descriptive representation by class to the question of under-representation of the working class. Shifts in the occupational background of non-working-class legislators may well be relevant for explaining changes in government responsiveness over time—on the assumption that, say, teachers are likely to more responsive to the preferences of low-income citizens than businessmen. This, then, is another item that I would like to put on our collective agenda. The final essay for this debate section is by Jane Mansbridge, a prominent contributor to the literature on representation in political theory. As indicated above, the other essays essentially ask whether working-class or low-income legislators behave differently from other legislators and whether (or to what extent) descriptive misrepresentation explains income or class bias in government responsiveness. Befitting a political theorist, Mansbridge instead asks a normative question: if it were possible to redesign our political institutions so as to improve descriptive representation of working-class or low-income citizens (for example, by changing electoral district boundaries or introducing quotas), should we do so? As Mansbridge explains, creating descriptive representation by some forms of institutional design entails significant costs. Most importantly, Mansbridge suggests, when institutional design takes the highly visible and non-fluid form of quotas and reserved seats, “the identity of citizen” may be lost, as “specific identity becomes the focus.”7 The choice available to citizens may in some cases be restricted and, when parties colonize these forms, the competence of the legislature may be diminished. In view of these potential costs, special conditions must be shown to exist to justify non-fluid measures to improve descriptive representation. Returning to the conditions that she invoked to justify descriptive representation of women and blacks in her seminal 1999 article, Mansbridge argues that none of these conditions pertain strongly in the case of social class. Yet she goes on to assert that a case for enhancing descriptive representation can be made when a group's interests are not adequately represented through other democratic mechanisms. Like other contributors to this debate, Mansbridge clearly recognizes that descriptive misrepresentation by income and class has always been a feature of representative democracies. For Mansbridge, the fundamental problem confronting us today is that “other mechanisms”—in the first instance, unions and parties—no longer serve to represent working-class and low-income citizens very well. Mansbridge's normative embrace of descriptive representation for the working class thus strikes me as an essentially defensive move. Like Gilens, I wonder whether it is realistic to think that descriptive representation of low-income, working-class citizens could be significantly improved without changing other features of the political system—most obviously, in the case of the US, laws governing campaign finance and union recognition. And if these other features were changed, measures designed specifically to enhance descriptive representation would, by Mansbridge's reasoning, become less desirable. While I am not entirely convinced that descriptive representation is the key to improving democratic responsiveness to the preferences of low-income and working-class citizens, I am convinced that this collection of essays raises a number of important questions that ought to be pursued by political scientists. To my mind, the most important analytical challenge is to articulate more clearly the conditions under which the background and material circumstances of elected politicians matter to public policy outcomes. In pursuing this agenda, parties and party systems arguably deserve more attention than they receive in the essays that follow.8 By design, our collection of essays focuses on the question, “Where do our elected politicians come from?” In the future, it might also be interesting to ask, “Where are our elected politicians going?” Eggers and Hainmüller's (2009) exemplary study of candidates for election to the British House of Commons between 1950 and 1970 shows that Conservatives who won—actually served in the House of Commons—were, at their death, significantly wealthier than unsuccessful Conservative candidates. Serving in the House of Commons does not appear to have yielded any significant wealth returns for members of the Labour Party, but it seems likely, I think, that legislative careers have increasingly become a pathway to affluence for Center-Left as well as Center-Right politicians. Setting aside the favors that might account for wealth returns to public office, could it be that elected representatives are motivated by future self-interest rather than past experience? Jonas Pontusson is Professor of Comparative Politics at University of Geneva. His current research interests include economic growth models and fiscal policy during the crisis as well as the political economy of inequality and redistribution.