Does it Matter that Most Representatives are Higher Educated?

Abstract

On 1 June 2005, in the first Dutch national referendum in over 200 years, 62 percent of voters rejected the so-called ‘Treaty establishing a Constitution for Europe’. They did so despite overwhelming elite support in favor of the constitutional treaty. In fact, 85 percent of the Dutch Second Chamber, the country's lower house of parliament, supported adoption of the treaty.1 How can we make sense of this utter lack of political agreement between elected representatives and the citizens they purport to represent? I will argue that at least part of this distortion is traceable to the educational background of representatives. What makes this case relevant beyond the specific Dutch context is that influential scholarship has argued that proportional representation enhances ideological congruence between elites and masses (Lijphart 1999; Powell 2000). That is to say, the more proportional the electoral system of a country, the more its governments and voters overlap on the left-right axis. The same applies to congruence between the legislature and the electorate. Elections based on proportional representation are more likely to yield legislatures that bear closer ideological resemblance to that country's citizens (Golder and Stramski 2010). The Netherlands, as one of the most proportional democracies in the world, ought to prevent, or at the very least minimize, representational distortions of the type reported above. While most of the contributions to this symposium focus on the descriptive represenation of socio-economic variables such as social class or income, there is good reason to take the educational background of representatives into account as well. In the wake of the financial crisis issues related to economic inequality have once again emerged on the political agenda in many European countries. However, prior to this revival most of the discussion on the lack of congruence between elites and masses in Western Europe focused on socio-cultural issues such as European integration and immigration. Hanspeter Kriesi and his collaborators have famously argued that Western Europe has seen the emergence of a new societal cleavage where ‘winners and losers of globalization’ – anchored by their level of education rather than by social class or income – are pitted against one another (Kriesi et al. 2006). In order to understand unequal representation on this socio-cultural dimension, we therefore have to include education in our discussion of descriptive representation. No representative assembly in the world mirrors the electorate that it represents. Political elites are mostly drawn from the upper socio-economic strata of society and are considerably wealthier and more educated than the average citizen (Best 2007). The key question is whether or not this mismatch in descriptive representation is harmful for substantive outcomes. In her treatise on the concept of representation Hanna Pitkin (1967) presents these different aspects of representation as interconnected. A rich literature has shown that the numerical underrepresentation of certain subgroups skews policy in their disadvantage, particularly with regard to characteristics such as gender and ethnicity (see Broockman 2013 for an overview). But what about socio-economic variables? Recent research on the class background of representatives in the United States and Latin America (Carnes 2013; Carnes and Lupu 2014a) has shown that legislative behavior of working class representatives differs markedly from that of representatives from white collar professions. Class anchors attitudes towards social spending both for representatives and the represented in the sense that people from working class backgrounds are more in favor of redistributive measures. This line of inquiry would imply that working class citizens are best represented by elected officials with whom they share a class background. That is to say, agreement on political issues is likely to benefit when citizens and their representatives are more alike in socio-economic terms. European researchers have noticed the emergence of another societal cleavage, one that is anchored in educational attainment rather than social class (Kriesi et al. 2006). The issues underlying this cleavage are not redistributive, but cultural in nature, as they encapsulate current debates on multiculturalism, law and order and European integration. Whereas socio-economic issues are often anchored by social class and income, these socio-cultural issues are more closely linked to level of education (Hakhverdian et al. 2013; Hainmueller and Hiscox 2006). At the same time, legislative recruitment is increasingly based on formal schooling (Best 2007). In a series of studies Mark Bovens and Anchrit Wille (2010; 2011) have drawn attention to the fact that “the well-educated currently comprise less than a third of the population, yet they dominate every political venue in the Netherlands” (2010: 417). They even go so far as to label parliamentary democracy in the Netherlands a “platonic meritocracy” (2011: 12). At the national level about 90 percent of the Dutch lower house of parliament had completed higher vocational or university education (Bovens and Wille 2011). When we only consider legislators with an university background, the Dutch figures are broadly comparable to Germany, the UK and Italy and slightly lower than France (Gaxie and Godmer 2007). Only the Nordic countries have significantly lower shares of university trained parliamentarians. In stark contrast, the lower educated, that is, those having completed lower vocational training or less, are virtually absent from national representative bodies. Even at the municipal level, where recruitment into political office is far less competitive than at the national level, about two thirds of representatives are higher educated (Hakhverdian et al. 2014). In the general population less than 40 percent belong to this latter category. If education has similar attitudinal effects at the elite level, we would expect the preferences of higher educated representatives to be more congruent with the preferences of higher educated citizens, while the preferences of lower educated citizens ought to be more congruent with the preferences of lower educated representatives. In other words, citizens' views would be best represented by representatives from their own educational category. In order to compare attitudes between elites and masses, one would ideally have survey data where identical items were fielded to both groups. One can then quite easily calculate the degree to which political attitudes are similar across representatives and the represented. By only capturing ideological congruence as the absolute distance between the left-right position of an elite actor and the position of the median citizens (e.g. Powell 2000), one ignores the distributions underlying these attitudes which in turn can produce misleading results (Achen 1978). A more comprehensive approach would be to compare the actual distributions of these preferences (Golder and Stramski 2010; Andeweg 2011). This type of ‘many-to-many congruence’ captures congruence as the overlap between elite and mass preferences as the common area under the respective frequency distributions of these preferences. Figure 1 illustrates this measure using original survey data among Dutch local representatives, complemented with data from the Dutch Parliamentary Election Study.2 Respondents were asked to place themselves on a 7-point scale that measures attitudes towards multiculturalism.3 Figure 1 shows the frequency distributions among local representatives and citizens. We clearly see that representatives tend to be embrace multiculturalism more so than citizens. The shaded area represents our measure of ideological congruence. In this particular case, about 83% of the distribution of citizen preferences overlaps with elite preferences (and vice versa). We have shown elsewhere that congruence between local representatives and citizens is very high on redistributive variables (91%), but tends to be somewhat lower on socio-cultural issues, such as European integration (84%), multiculturalism (83%), asylum policy (82%), and law and order (77%) (see Hakhverdian et al. 2014). Having said that, congruence at the local level is markedly higher than at the national level, where these figures drop about ten percentage points using identical socio-economic and socio-cultural questions fielded to parliamentarians (Andeweg 2011; Schakel and Hakhverdian 2015).4 While by no means a formal test of descriptive representation based on education, these patterns are at least consistent with the notion that the closer educational match between citizens and representatives at the local level results in higher congruence figures than between citizens and representatives at the national level. When we look beyond collective congruence by disaggregating the electorate in subgroups based on education, a striking pattern emerges. Figure 2 shows that the preferences of Dutch local representatives on multicultural issues exhibit a much closer match with the preferences of higher educated citizens compared to lower educated citizens. We find comparable ‘congruence gaps’ of about 20 percentage points with regard to other socio-cultural attitudes. In contrast, we find no evidence for unequal representation on redistributive attitudes. At the national level, these inequalities are far more severe. Again, using the same item to tap attitudes towards multiculturalism, congruence between national representatives and the higher educated is as high as 94% (Schakel and Hakhverdian 2015). For all intents and purposes, Dutch parliamentarians and higher educated citizens hold identical views towards multiculturalism, at least in 2006 when these data were gathered. Congruence between parliamentarians and lower educated citizens was as low as 59%. Even in one of the most proportional democracies in the world, political representation remains biased in favor of the upper echelons of society. Why these patterns actually arise, remains hotly contested. In his sprawling study of political inequality in the United States, Martin Gilens (2012) dismisses the lack of descriptive representation in Congress as a potential mechanism underlying his main finding that the wealthy wield disproportionate power over policy outcomes. Still, whereas voting records of Members of Congress turn out to be uncorrelated to their outside wealth or income, the class background of representatives does matter, even when controlling for important confounders at the individual and district level (Carnes 2013). Part of the discussion seems to revolve around the particular type of policy or elite-level outcome under consideration, a matter which is also highly relevant to the present argument and to which I shall return in the concluding remarks. So do representatives from various educational backgrounds actually differ with regard to their political preferences? Unfortunately, this matter cannot be investigated at the national level, since virtually all representatives in the Dutch lower house of parliament are higher educated. The local level does enable us to test this proposition, as sufficient variation exists on representatives' educational background. We can therefore regress elite attitudes on levels of education both with and without additional controls for gender, age, religion and party affiliation. The relationship between issues attitudes and educational attainment is reproduced at the elite level for those issues that traditionally cluster on the cultural dimension (Hakhverdian et al. 2014). Higher educated local representatives exhibit more progressive attitudes on these matters compared to their lower educated counterparts. Education fails to predict attitudes towards redistribution, again echoing patterns found at the mass level. This latter finding is important, as it leaves little room for unequal representation of education groups on these socio-economic issues. It is important to note that these results hold even after controlling for party affiliation. Certain parties might recruit more actively among those who hold similar views to that party. However, even within parties, we find that education matters for elite-level socio-cultural attitudes. Finally, we turn to the main proposition that representation suffers when educational backgrounds of elites and masses diverge. In order to test this proposition empirically, we reproduce the abovementioned congruence scores, but now we disaggregate voters and representatives based on their level of education. Figure 3 shows that congruence scores for multiculturalism are higher for citizens and representatives with similar levels of education. The highest congruence score occurs among representatives and citizens with higher educational level. The worst mismatch is that between the preferences of higher educated representatives and lower educated citizens. Similar patterns exist for other socio-cultural attitudes, but not for attitudes towards redistribution. This latter finding should come as no surprise as these attitudes do not vary across education groups, neither for elites, nor for masses. As such, the mismatch between the educational background of citizens and their representatives does not translate into unequal representation on socio-economic issues. Elected representatives are not representative of the electorate that they represent on a myriad of demographic characteristics. This lack of descriptive representation spills over in the representation of the preferences that are tied to these demographics. We have seen that Dutch local and national representatives hold political views that are more in line with higher educated citizens. We have argued that part of this pattern can be explained by the mismatch in educational background between elected officials and the population at large. That is to say, ideological congruence on political issues is highest for representatives and voters with similar levels of education. Two important caveats remain. The first, and in my view most important, limitation of the reported bias in representation concerns its focus on elite attitudes rather than elite behavior or actual policy outcomes. I certainly would not equate attitudes with behavior or policy, but that is not to say that personal preferences of representatives are irrelevant. Others have shown that roll call behavior depends, at least in part, on representatives' attitudes (Levitt 1996). This is reminiscent of findings from the partisanship literature where scholars have examined the extent to which government ideology translate into different policy outcomes. For instance, left-wing governments are more likely to preserve generous welfare state entitlements than right-wing governments (Allan and Scruggs 2004), even though findings such as these might be conditional on a host of contextual factors (e.g. Rueda and Pontusson 2000). If we envisage a causal chain running from the demographic attributes of a legislature to actual policy outcomes, the analyses conducted above that focus on attitudes can be situated somewhere along such a causal sequence. Whether or not elite attitudes are found towards the demographic or policy end of this ‘chain of responsiveness’ (Powell 2004) is likely to depend on institutional features of the case at hand. Analyses that focus on roll calls are useful in some instances, but unfeasible in others. Party discipline in the Netherlands is so tightly enforced that voting records fail to contain any discriminatory information regarding the behavior of individual legislators.5 Anecdotal evidence can certainly back up the claim that attitudinal biases among representatives have policy implications – recall the referendum on the EU Constitution in the opening paragraph – but obviously systematic enquiry that includes policy outputs is to be preferred (e.g. Gilens 2013; Soroka and Wlezien 2010). The second limitation concerns the specific timing and context of the elite surveys reported above. All Dutch major national political parties participate in municipal elections in most municipalities with the exception of Geert Wilders' Freedom Party (PVV). The PVV only ran in two municipalities, The Hague and Almere, reportedly because party leader Wilders was unable to find enough suitable candidates to run on a broader scale.6 The national level survey among parliamentarians also did not contain representatives from the populist right. Given the fact that radical rightwing populist parties are more likely to espouse precisely those socio-cultural views that we have shown to be underrepresentated at the local and national level, their absence in our analyses could to some extent skew the reported underrepresentation of socio-cultural attitudes.7 Still, this potential skew is offset by at least two factors. First, at the national level, mainstream parties have responded to the sudden rise of the Dutch populist right in 2001 by adapting their policy positions to these challenger parties on socio-cultural issues (Van Spanje 2010). Second, local parties are increasingly gaining support at municipal elections. Representatives from these local parties tend to be resemble the populist right on socio-cultural issues, albeit in a less extreme manner. These representatives are included in the analyses reported above and should mitigate some of the concerns about the limited scope of PVV-participation at the local level. These limitations aside, we have shown that local representatives and citizens are more likely to agree on important societal issues if they share a common educational background. Preferences of both local and national representatives in the Netherlands align more with the preferences of higher educated citizens, but only on socio-cultural issues. We also find a clear mismatch between the political preferences of lower educated citizens and their higher educated representatives. This poses a fascinating normative dilemma. Would representative democracy function better if legislative bodies consisted of more lower educated representatives, as for instance the former President of the Dutch lower house of parliament, Gerdi Verbeet, has argued? Most people would balk at such a suggestion, as the view that higher education is inherently desirable is widespread. However, whether or not lower educated representatives actually function worse than their higher educated counterparts remains to be seen. Carnes and Lupu (2014b) fail to find any evidence to suggest that the level of education of political leaders and their performance in office are actually linked. The call for a composition of legislative bodies that is more representative of the population at large therefore does not necessarily compromise its quality. But even if formal schooling were to produce better leaders, a human capital approach to education that focuses on skill acquisition alone would still miss part of the story. Education leaves a mark on people's broad normative proclivities, both as a result of the curriculum to which one is exposed as well as through social interactions with peers (Hakhverdian and Mayne 2012). We should therefore treat education as more than a mere proxy for someone's skills and cognitive functions. Education anchors political beliefs among citizens as well as representatives and this carries important implications for our understanding of the functioning of representative democracy. Armen Hakhverdian is Assistant Professor of Political Science at the University of Amsterdam. His research interests are in the field of comparative politics with a focus on political representation, public opinion, and inequality. He has published in journals such as Journal of Politics, British Journal of Political Science, Socio-Economic Review, and European Union Politics.