Abstract
Focusing on the effects of singular technical elements, electoral system research has neglected the question of whether specific systems are supposed to achieve overarching normative goals. Due to their importance for any polity, such principles of representation are widely assumed to be determined in the constitution. Therefore, a world-wide survey of constitutions presents a promising opportunity to study electoral systems’ general goals in a comparative manner. In providing such a survey and investigating the causes, contents and consequences of constitutional provisions, this analysis shows that constitutional embeddedness of the electoral system is contingent upon factors such as region (with constitutional principles especially typical in Europe and Latin America) and decision-context. Importantly, the ‘proportionality principle’ is much more prone to enter a constitution and receive increased protection than its antipode, the ‘majority principle’. Furthermore, mixed principles calling for a balance between aforementioned extremes exist, suggesting that mixed electoral systems are not always merely technical compromises. Finally, constitutional embeddedness seems to have a context-dependent effect on specific technical elements but generally leads to greater stability of the overall electoral system. The key implication for future research is that the normative principle-dimension of electoral systems has to be considered alongside their technical design.
Introduction
When investigating the role of electoral systems, we are used to detailed analyses of the effects of specific technical elements on outcomes such as the disproportionality between vote and seat shares, minority representation, Member of Parliament (MP) behaviour and government formation. However, much less frequently do we have information about whether a particular electoral system, notwithstanding its exact technical design, is supposed to follow a specific principle and achieve a particular normative goal. Automatically associating specific technical designs (especially pure proportional representation (PR) and first-past-the-post) with specific normative goals and deriving design implications based on outcomes might be highly problematic if designers of technically similar systems have followed different normative agendas. So far, mostly case studies (e.g. in Gallagher and Mitchell, 2005) have assessed whether electoral systems are supposed to follow a principle or whether they rather were the result of diffusion processes or self-interested bargaining lacking a genuine principle. The consideration of this principle-dimension of electoral systems could add tremendously to our ability of thoroughly evaluating electoral system performance and, eventually, supplying practical advice (see Gallagher, 2005b: 568; Farrell, 2011: 188 in this context).
While there is an abundance of detailed and fairly unambiguous information about the technical design of electoral systems, the situation is radically different when it comes to the principle-dimension. However, one line of argument suggests a (nearly) universal and rather parsimonious way of learning about an electoral system’s overall goal. This argument posits that the choice of an electoral system is a decision of constitutional importance when setting up the institutional framework and that – as a result of this choice – the general principle of an electoral system will be included in a country’s constitution (see Lijphart, 1991, 1999; Nohlen, 1984, 2009). Specifically, designers ought to be deciding between the accurate (proportional) reflection of citizens’ opinions and the production of stable majorities as competing general principles of representation for the electoral system. This principled decision then would imply a clear guideline for the respective electoral system – of whichever technical design – and its evaluation (see especially Nohlen, 1984, 2009) and at the same time provide for a high level of (constitutional) protection to this key institution. A survey of constitutions could thus prove tremendously helpful in closing the gap of the neglected principle-dimension in electoral system research.
Therefore, the following pages will explore constitutional documents as a source of principles of representation and provide a first assessment of the potential causes and consequences of constitutional embeddedness. In surveying constitutions, two central tasks are at hand: firstly, the analysis will assess whether constitutions indeed typically convey information about the electoral system and especially its general guiding principles or whether this function is contingent upon contextual factors. Secondly, the focus will be on what types of principles are stated, what they require from election outcomes and what they imply for the technical design of the electoral system. Addressing these questions will then also speak to the general validity of constitutional provisions for measuring overall goals of electoral systems.
The paper is structured as follows. Firstly, we will revisit the argument of the electoral system as being of constitutional importance for polities, derive expectations as to the content of constitutional provisions and consider potential causes and consequences of the constitutional embeddedness of the electoral system. Secondly, the methodological approach of soliciting information about the principle of representation from constitutions is discussed. Thirdly, an empirical overview based on the analysis of the currently valid constitutions of 86 democracies is presented and – in combination with information about the technical design of electoral systems – used to carry out the tasks outlined above. Finally, the conclusion suggests fruitful avenues for future research.
The constitutional embeddedness of the principle of representation
When defining the principle of representation, Dieter Nohlen (1984: 86) poses ‘the question as to the political goals of political representation concerning the aggregate nation-wide outcome of elections’ and clarifies that this principle is related to the technical design of an electoral system as an end is to its means – where the latter may differ given the circumstances (see also Katz, 1997: 309; Raabe and Linhart, 2012). When considering the overall normative goal of an electoral system – ‘what you want your legislature and executive government to look like’ (Reynolds et al., 2005: 9) – we are concerned exactly with this overall principle without inferring it from the technical design of an electoral system.
For Lijphart (1991, 1999), Nohlen (1984, 2009) and Powell (2000), the decision for such an overall principle carries great implications for the overall style of democracy practised in a given country. Generally, a choice has to be made between a majoritarian and a consensus vision of democracy. In this framework, the associated decision for what the electoral system is supposed to produce as an aggregate outcome and how it should shape a country’s political landscape is seen as a landmark decision of political system design (Birch, 2003: 8; Norris, 2004: 66–77; Powell, 2000). Following this logic of competing styles of democracy, an electoral system may either be called upon to foster the formation of accountable (single-party) governments and limit the fragmentation of the party system or to facilitate the accurate (proportional) representation of the different political interests of the citizenry – importantly including various minority groups – within the parliament (see, e.g. Bernauer et al. 2013; Dunleavy and Margetts, 1995; Farrell, 2011: 10–11; Norris, 1997; Pinto-Duschinsky, 1999; Przeworski et al., 1999). These two principles are usually understood as incompatible and thus present system designers with a crucial trade-off decision (Duverger, 1984; Lijphart, 1994: 144; Rae, 1967; Reynolds et al., 2005: 9–15; Taagepera and Shugart, 1989). 1
Due to the wide-ranging implications associated with the choice of a principle of representation, it is generally held that this choice is a ‘vital element in democratic constitutional design’ (Lijphart, 1991: 72) leading to the principle of representation’s inclusion into a country’s constitution (Lijphart, 1991; Nohlen, 1984, 2009: 145; Steiner, 1971: 64). The choice of the specific technical design is then left to ordinary law (e.g. Reynolds et al., 2005: 20). This expectation also seems reasonable when considering that constitutions are supposed to state and protect the key rules and general guiding principles forming the building blocks of a democratic political system (see Brennan and Hamlin, 2008; Elkins, 2010). While constitutions lend special attention and, even more importantly, protection to these principles (Benoit, 2007: 382; Flinders, 2010: 55), the more specific institutional design may then typically be undertaken via ordinary law (Brennan and Casas Pardo, 1991: 54). Following the above, it may thus be assumed that the principle of representation is defined within any constitution.
However, research on electoral systems as well as on constitutions suggests that the existence of constitutional principles might vary between polities. Firstly, and contrary to what is suggested above, an electoral system may lack a clear-cut principle of representation. This seems especially likely if elite bargaining leads to a compromise without a genuine principle (see Birch, 2003: 3–27). Secondly, the need for flexibility in reforming the electoral system and/or for broad support of the constitution might affect the propensity of a particular principle entering the constitution (Brennan and Hamlin, 2008: 340; Melton et al., 2013: 403; Taagepera, 1999: 14). Thirdly – as is the case with constitutions generally (Elkins, 2010) and also the technical design of electoral systems (Blais and Massicotte, 1997) – patterns of constitutional embeddedness might arise in regional clusters resulting from shared socio-political developments as well as diffusion processes. Finally, principles of representation could themselves lead to different patterns of embeddedness. For example, the demand for accurate, inclusive representation is often tied closely to the process of democratization, including the inherently democratic (Elkins, 2010: 974) writing of constitutions. The election of representatives from single-member districts, on the other hand, is an older concept not always clearly tied to a general normative principle of constitutional rank. The first part of the empirical analysis will therefore assess whether the constitutional embeddedness of principles of representation is indeed universal or whether it is contingent upon different contextual factors.
When it comes to the content of constitutional principles, it is typically claimed that the decision for such principle is dichotomous – one may either go for concentrating the party system or for accurate representation while meeting halfway is no reasonable option (Colomer, 2004: 58; Nohlen, 1984: 85–87, 2009: 138; Sartori, 1997: 53). Yet, in light of the surge of mixed electoral systems and their latent promise of reaching a superior middle ground in electoral system design (see Carey and Hix, 2011; Shugart and Wattenberg, 2001), this dichotomy-claim seems unconvincing. The emphasis put on a compromise between different normative goals among political scientists and system designers (see, e.g. Birch, 2003: 28–51) points towards a continuous understanding of the principle-dimension. Accordingly, the general principle of an electoral system may well be a mixed one – whether or not it is paired with a multi-tier electoral system. Therefore, constitutions will be assessed as to what types of – pure and mixed – principles are stated and whether a subset of constitutions demands a balance between antithetic pure principles.
Turning to the implications of constitutional principles of representation, it is important to consider whether and how constitutional embeddedness affects design and stability of an electoral system, that is, whether and how constitutional embeddedness matters. The above-stated means–end relationship of design and principle might call into question a deterministic association between these dimensions. Given variations in socio-political circumstances, different countries may rely on different technical designs to fulfil similar goals or vice versa. For instance, the normative goal of proportional representation may be combined with a mixed-member electoral system – suggesting the principle of proportional representation does not necessitate a pure PR system nor does the existence of a mixed system necessarily signal the desire for a balance between different principles. Hence, it is expected that the same general principles are paired with varying technical designs and vice versa. Evidence for this would cast doubt on the often implicit assumption that a specific technical design clearly signals a specific normative desire. Whereas the above suggests a certain level of indeterminacy of constitutional principles regarding the actual technical design of an electoral system, it is plausible that such principles affect specific technical elements in the same manner even under varying contextual situations. District magnitude, prompting an ‘effective threshold’ (Lijphart, 1994: 27), and the legal threshold stand out as two factors that are most closely linked to the trade-off between achieving accurate representation and the fostering of a legislative majority (Lijphart, 1994). A pure constitutional principle (e.g. proportionality), functioning as a focal point for design and lending uplifted legal protection – rendering reform more complicated and making possible an appeal to the constitutional court – should lead to a more pure technical design. Via rendering the adoption of technical rules running counter to the respective principle less likely, constitutional embeddedness of pure principles can be expected to lead to, in the case of the proportionality principle, lower (legal and effective) thresholds and to higher thresholds in the case of the majority principle.
Finally, notwithstanding the exact technical design, constitutional embeddedness and the ensuing superior legal protection for the principle of representation should lead to greater stability of an electoral system. This is because constitutional amendments typically require legislative supermajorities – two-thirds majorities are the norm and in the (e.g. Danish) case where an absolute majority is sufficient, a successful popular referendum has to confirm the amendment. Furthermore, constitutional provisions regarding the electoral system provide constitutional courts with considerable leverage to repeal reform bills. Therefore, constitutional embeddedness can be expected to lead to greater resilience of an electoral system against electoral reform.
Surveying constitutions
The cases under investigation include the currently valid constitutions in 86 democracies according to the 2011 Polity IV index (see Marshall et al., 2012). New Zealand, San Marino and the United Kingdom were excluded from the sample as they neither possess a written constitution nor a clearly identifiable collection of constitutional documents. Constitutional texts were gathered from official sources (governmental or parliamentary institutions) as well as from the Oxford Constitutions of the World archive. 2 The following empirical overview is concerned with electoral systems responsible for the seat distribution in national parliaments. 3 Thus, the data collection has focused exclusively on passages concerning elections to the unicameral legislature or the lower house of a bicameral legislature. Accordingly, the relevant passages referring to the electoral system are usually contained in the sections dealing with the legislative branch of government. Less frequently, the sections dealing with general principles for the political system overall and specific sections concerned with elections or the electoral system included (additional) relevant passages.
As concerns principles of representation, after an initial identification of the relevant passages these were investigated as to whether they conveyed general principle(s) or goal(s) as to what the electoral system is supposed to achieve at large. As regards the substantive content of these passages, it was expected that statements of general principles would circle around the trade-off between concentrating the party system and fostering an accurate reflection of public opinion. However, constitutional passages were read in an open-minded way in order to also consider potential other goals which are not expected to be defined within the constitution (such as personal representation; see Colomer, 2011). In order to provide the reader with both the original as well as with more comprehensive, processed information, an online appendix (accessible at http://ips.sagepub.com) includes the relevant constitutional passages stating a general principle as well as a grouping-variable, while more condensed information is presented in Tables 1 and 2. According to the expectations derived above, principles were allocated to the following groups: ‘proportionality principle’ and ‘majority principle’ (following Steiner, 1971) as well as what Nohlen, Sartori and others would probably consider residual categories, ‘mixed principle’ and ‘other principle’ (including those principles that cannot be associated clearly with the central proportionality-majority trade-off). For countries whose constitutions do mention the electoral system and where the decision about whether the text refers to a general principle was not straightforward and actually went in the direction of ‘no principle stated’, the online appendix includes explanations as to the coding decision. In order to investigate the link between principle and technical design, Golder and Bormann’s (2013) dataset on the technical design of electoral systems will be used in addition to the constitution data.
Constitutional provisions.
Notes: Countries highlighted bold state both a principle and a technical procedure in their constitution; regions are based on Golder and Borman (2013).
Overview over different groups of countries.
MMM: mixed-member majoritarian; MMP: mixed-member proportional; PR: proportional representation; STV: single transferable vote.
Based on Golder and Bormann (2013).
Denmark is counted twice both as MMP and as PR since the constitution allows for both.
Since, despite all efforts to avoid this, the coding of constitutional texts by a few individuals may not be entirely free of coder-bias, two strategies have been adopted in order to alleviate this problem. Firstly, similar to the strategy employed by the researchers within the Comparative Constitutions Project (see Melton et al., 2013), the identification of relevant text passages and the substantial codings were discussed by two coders who then agreed on a final coding. Secondly, the online appendix will provide readers with the original text so that readers may scrutinize the coding decisions.
Assessing constitutional principles of representation
Constitutional embeddedness
Out of 86 countries surveyed, 42 countries’ constitutions neither include information about principles of representation nor about the technical procedure used to convert votes into seats (these countries are listed in the online appendix). Thus, while a slight majority of 44 countries specify at least either a guiding principle or a technical procedure, the argument that constitutions typically include a statement about the general normative goal of the electoral system is clearly called into question by the empirical evidence. Clearly, the electoral system is not always included in the constitution of a country and thus constitutional entrenchment has to be considered a contingent phenomenon. Table 1 presents the results – listing the countries whose constitutions include provisions about the electoral system by region – and provides a starting point for the assessment of factors that might affect constitutional embeddedness.
Table 1 includes a majority of the countries surveyed and suggests that constitutional entrenchment of the electoral system occurs frequently (but far from universally). Yet, Table 1 also differentiates between statements of genuine principles and of general technical procedures via which votes are to be converted into seats. Including this information ensures that, even if the constitution is not explicit in defining, for example, majority facilitation as an overall goal, the possibility that defining a technical procedure is also meant to imply a principle of representation is not ignored. 4 Yet, if we fully differentiate between these types of statements, it becomes obvious that actually only a minority of constitutions (26) define genuine principles for the electoral system. We will later revisit this distinction.
Turning to the assessment of the contingency of constitutional embeddedness and regional differences, Table 1 clearly shows that the constitutional provisions concerning the electoral system are especially common in Western Europe and, to a lesser extent, Eastern Europe and Latin America. In Western Europe, 14 countries deal with key aspects of the electoral system in their constitution, with nine constitutions specifying a genuine principle. In Latin America, six out of 16 countries explicitly state a principle (with four additional constitutions specifying only a technical procedure). In Eastern Europe, those countries leaving electoral system matters to non-constitutional law are in the majority. Here, only seven of 19 constitutions include a principle statement, while two additional constitutions state a technical procedure. The regions of Europe and Latin America then account for 22 out of 26 countries stating a principle of representation and 33 out of 44 countries whose constitutions include key provisions about the electoral system. In sum, regional clustering seems to occur (likely based on common historical developments and processes of diffusion) – with Western European countries typically dealing with the electoral system in their constitution – but this hardly accounts fully for patterns of constitutional embeddedness. As will become apparent below, the consideration of further parameters even suggests that the correlation between region and constitutional embeddedness is spurious.
Going beyond regional patterns, elite bargaining resulting in compromise electoral systems and the necessity to gather broad support for a constitution seem to affect the constitutional embeddedness of the electoral system. This connection is suggested by the literature on electoral system development in Eastern Europe (e.g. Birch, 2003). This literature highlights the co-occurrence of the desire – in part caused by the high level of uncertainty about future election outcomes – to serve collective interests and integrate society (see also Benoit, 2007: 381) and the relatively frequent constitutional entrenchment of electoral systems in the region, strikingly often in the form of a principle-statement (see Table 1). At the same time, countries such as Bulgaria, Croatia or Hungary – lacking a constitutionally embedded electoral system – experienced intense bargaining between new and old elites, with the result of compromise systems without a general principle (Benoit, 2005; Shugart and Wattenberg, 2001: 579). In Western Europe, Germany (Scarrow, 2001) and France (Elgie, 2005: 119–121) experienced complex bargaining over the electoral system, resulting in a lack of constitutional provisions. In France, different parties finally agreed upon the electoral system but were divided on whether or not to make it constitutional (Wahl, 1959: 366–367). In Italy, the desire for continued flexibility in electoral system design seems to have been the key reason for the lack of constitutional entrenchment (Baldini, 2011: 645–647). Considering countries whose constitutions include principles of representation, Ireland (Gallagher, 2005a: 512), Israel (Rahat and Hazan, 2005: 334–336) and Switzerland (Lutz, 2004) are examples of how there existed general agreement over the importance of proportional representation and its potential to foster broad (popular and elite) support for the constitution. Thus, pending in-depth assessment of constitutional choices, the above suggests the following: under circumstances in which electoral system choice is not based heavily on strategic partisan interests, it is more likely that provisions about the electoral system will be included in the constitution.
Finally, differences between countries and regions might also be caused by electoral systems and principles of representation themselves. The co-evolution of the PR system and democratization in Europe (see Blais et al., 2005) and the inclusiveness of proportional representation, as well as its potential to generate widespread support, seem to render the proportionality principle prone to be included in constitutions. On the contrary, as mixed electoral systems often result from bargaining revolving around the antithetic principles of majority and proportionality (Benoit, 2007; Birch, 2003), they seem less likely to be perceived as following a normative principle deserving of constitutional protection. Furthermore, the case study literature highlights that in Canada, the United Kingdom and the United States system designers ‘were hardly aware that they had “chosen” an electoral system’ (Gallagher, 2005b: 538–539). Consequently, the respective constitutions would not have to include a principle statement because there was no awareness of a principled choice to be made. In order to assess the potential consequences of such differences among electoral systems and their principles, the analysis now turns to the content of the constitutional provisions.
What types of principles are proclaimed?
When it comes to the content of constitutional provisions, the proportionality principle is mentioned overwhelmingly often: 22 out of 26 countries stating a principle proclaim the proportionality principle (see Table 2; the online appendix for detailed information). At the same time, it is striking that none of the constitutional principles is that of majority. 5 Therefore, on the constitutional level, the proportionality principle is well established, while its antipode as well as mixed and other principles are rarely, in the case of majority never, featured in the constitution. It thus appears that different principles indeed vary in their propensity to be embedded constitutionally. Comparing these findings with those regarding the constitutional statement of a general technical procedure suggests that the key differences seem to rest with the normative principle-dimension. While only two constitutions (Comoros and Liberia) clearly state that a plurality system will be used in single-member districts, 6 the over-representation of proportionality as a stated principle is not mirrored by a similar over-representation among constitutions stating a technical procedure. In the latter category, PR systems account for 60 per cent of the cases. Looking at all electoral systems in the dataset shows how this percentage closely reflects the percentage of PR systems among all electoral systems (see Table 2). The most intriguing difference between statements of principles and of technical procedures concerns mixed principles/electoral systems. While genuine mixed principles are rarely stated, mixed technical procedures are stated relatively often. Does this imply that, while mixed systems are also fairly often embedded constitutionally, they at the same time lack a guiding normative idea? This would be quite contrary to the idea that these systems often aim at providing a superior middle ground and could be part of the reason why institutional features of these systems are strategically exploited by party elites (see, e.g. Bochsler, 2012). A closer look at individual countries will disclose whether there also exist mixed principles aiming for a balance between majority and proportionality goals.
Starting with a closer look at the proportionality principle, constitutional provisions typically adhere closely to the associated normative concept, most frequently stating that elections shall be ‘in accordance with the principles of proportional representation’ (e.g. Austria, Czech Republic; see online appendix). In an alternative, multiple constitutions make explicit that the parliamentary seat distribution has to equal the vote distribution among parties (e.g. Bolivia, Norway and Sweden). Furthermore, a few constitutions refer to the concept of inclusive representation lying behind the proportionality principle. For example, the Costa Rican constitution states ‘representation of minorities’ and ‘political pluralism’ as goals. The Danish constitution requires the electoral system to ‘secure equal representation of the various opinions of the electorate’.
From a mixed principle of representation one would expect that the constitutional statement made concessions to the majority principle alongside the proportionality goal, calling for a balance between pure principles. In this regard, four constitutions stand out as they proclaim principles that differ from what a dichotomous choice between proportionality and majority would suggest. The Mexican constitution refers to the ‘majority principle’ with respect to the single-member district tier of its mixed electoral system and to the ‘proportional representation principle’ with regard to the PR tier. Mexico’s constitution explicitly demands a combination of both principles and may thus be read to state a mixture of the dimensions of proportionality and majority. Likewise, Turkey’s constitution has been coded as conveying a mixed principle. Yet, Turkey’s constitution even explicitly refers to the trade-off inherent in electoral system design, clearly aiming at balance: ‘[t]he electoral laws shall be drawn up in such a way as to reconcile the principles of fair representation and consistency in administration’. While one could debate the Mexican case, the Turkish case clearly highlights how the general normative guideline for an electoral system may well be placed in-between the extremes of proportional representation and majority-facilitating concentration of the party system. While these cases of mixed principles are exceptional, they still highlight that a general goal for an electoral system need not derive exclusively from a dichotomous choice between two extremes. These findings underline how mixed electoral systems do not necessarily lack a guiding principle, even though overall there are relatively few mixed principles. 7
When it comes to the principles of representation in Kenya’s and Serbia’s constitutions, it is less clear whether they can be placed on the same dimension as the principles discussed above and thus count as mixed principles. These two constitutions were therefore coded conservatively to state an ‘other principle’. Kenya’s constitution calls for ‘fair representation’ and ‘fair representation of persons with disabilities’, and requires that at least one-third of the parliament consist of female MPs. The Serbian constitution states similar demands in a slightly less specific fashion, in that it calls for the provision of ‘equality and representation of different genders and members of national minorities’. As these constitutional provisions demand fair inclusion of specific socio-demographic groups, they may also be understood to constitute mixed principles if we assume that fostering unambiguous government formation is the goal with which these demands have to be reconciled.
Thus, overall, the central trade-off between providing accurate representation of public opinion and concentrating the party system indeed seems to be the key choice to be made within the constitution – bearing in mind that only 26 out of 86 constitutions clearly state a principle for the electoral system. However, while proportionality is called for by many constitutions and while other principles present a balance of the antithetic general goals, the majority principle is basically absent from constitutions throughout the world. Based on the evidence, it appears that while the trade-off dimension is identified correctly by electoral system researchers, the idea of the majority principle as a (literally) constitutional goal may be called into question. Potential reasons are that the majority principle is unlikely to generate support by all or most veto players during the constitution making process and, as one of the reviewers points out, that majority electoral systems are seen as the traditional way of choosing representatives not in need of constitutional entrenchment. However, caution is in order concerning these conclusions, for the sample of constitutions including a principle of representation consists largely of European and Latin American countries. Although the East Timorese, Kenyan and Turkish constitutions also fit the general pattern, the question of whether the absence of majority as a constitutional principle is partly based on this principle is applied mainly in regions in which it is uncommon to state a constitutional goal for the electoral system has to be tackled by future research.
Consequences of constitutional embeddedness
Is it warranted to assume that the same general goal leads to uniform technical design? Assessing the proportionality principle would at first sight suggest that principle choice and technical design are indeed closely associated. Golder and Bormann (2013) classify 21 countries out of 22 that proclaim the proportionality principle as also having a proportional electoral system (Bolivia being the exception in using a mixed-member proportional (MMP) electoral system). Yet, a closer look reveals how Malta and Ireland combine the proportionality principle with the single transferable vote (STV) system with fairly small district magnitudes. Israel votes in a single nation-wide district while, for example, Portugal uses multi-member districts. Furthermore, seven out of these 22 countries use national top-up seats to compensate for disproportionalities arising due to the use of multi-member districts of varying sizes (e.g. Austria and Denmark). Finally, legal thresholds of varying height are applied. Thus, the same general normative goal is combined with very different technical designs. Reasons for this may be highly idiosyncratic, as in the Irish case where system designers desired proportional results but were unaware of suitable technical options other than STV (Gallagher, 2005a: 513). Where these nuanced differences might be of minor importance, an investigation of principles other than proportionality highlights that principle clearly does not dictate design and vice versa. While the mixed principle in Mexico’s constitution is paired with a mixed (mixed-member majoritarian (MMM)) electoral system using two independent tiers to elect MPs, the Turkish case shows how mixed principles need not indicate the implementation of mixed-member electoral systems. Here, a PR electoral system with multi-member districts is accompanied by a very high legal threshold of 10 per cent. That different means may be called upon to achieve the same ends can also be inferred from comparing the electoral systems of Kenya and Serbia: while both call for representation of women and minorities, Kenya uses a rather special mixed system (with a PR-tier ensuring the election of at least one-third female MPs and a plurality tier for all other seats 8 ), whereas Serbia uses one nation-wide PR district and exempts minority parties from the five per cent threshold. Costa Rica also calls for minority representation yet employs PR in multimember districts without any legal threshold. In sum, the expectation that the same general principle is paired with varying technical designs and vice versa is borne out, especially when looking at technical details. Obviously, it is unlikely that any country would consider a first-past-the-post system to be suitable to fulfil the proportionality principle. However, system designers in different countries seem to make use of the many potential ways of reaching the same normative goals with a myriad of technical designs.
After highlighting that technical design is not dictated by constitutional principles, the remainder of the analysis seeks to explore the practical consequences of constitutional provisions. Regarding the expectation that the existence of a constitutional principle will lead to more pure technical design, the evidence is inconclusive. Neither for average district magnitude nor for the height of the legal threshold is there any evidence that constitutionally embedded systems aiming for proportionality are significantly different from their non-embedded counterparts. 9 Furthermore, there are no meaningful differences between regions, suggesting that the ‘purity’ of proportional electoral systems is generally not enhanced through the existence of constitutional principles. Are constitutional provisions thus more or less meaningless and unable to protect certain normative goals? When it comes to specific technical elements, the answer may very well be ‘it depends’ as, for example, legal thresholds do not seem to be regarded as violating the principle of proportionality by many electoral system designers (for example in Eastern Europe, see Birch, 2003: 11). Kostadinova (2002: 31) also concludes that ‘the electoral threshold emerges as a powerful mechanism for reducing fragmentation in the assembly […] without changing the fundamentals of the system itself’. Comparing Austria and Estonia, whereas each set out the proportionality principle in their constitution and employ a legal threshold (four per cent in Austria, five in Estonia), this threshold has been contested with reference to the constitutional text heavily in Estonia (Taagepera, 1999: 11–12), while no such reform pressure exists in Austria (Müller, 2005: 408). This contrast even carries to the constitutional level: while the Icelandic constitution includes a legal threshold alongside the proportionality principle (Art. 31), a legal threshold is prohibited by the Portuguese constitution (Art. 152). Whereas the legal threshold often seems to be viewed as belonging to a standard proportional electoral system, legal thresholds are explicitly used to strike a balance between the majority and proportionality principles in Turkey, and, to a lesser extent, in Spain (see Hopkin, 2005). Thus variation in the way that legal thresholds are used not only highlights again that principle choice does not determine technical design, but also that the impact of constitutional embeddedness on the ‘purity’ of the electoral system hinges upon what different societies deem as violating, for example, the proportionality principle. A similar potential reason for the lack of impact of constitutional embeddedness on district magnitude could be that oftentimes regional boundaries are used to group parliamentary seats into different districts. This distribution of seats to regions itself follows a logic of proportional representation (for example in Costa Rica), which may then lead to relatively low average district magnitude (e.g. in Finland or Switzerland). In this case then, low magnitude is not seen as violating the proportionality principle – Ireland is also a case in point (see Gallagher, 2005a: 512–517). Looking at technical details, the above discussion suggests that especially legal thresholds may come under attack based on constitutional provisions but that constitutional embeddedness has no uniform impact on such details.
Again revisiting country narratives, it seems that the more clear-cut impact of constitutional principles is on the longevity and stability of the overall electoral system. Countries lacking a constitutional principle experience constant quarrels over and significant changes of the electoral system – for example France and Italy (Baldini, 2011; Elgie, 2005) – and wide discrepancies between competing reform options (e.g. in Romania 2007; Nikolenyi, 2011: 618–620). On the other hand, constitutional embeddedness has led to the stability of the electoral system, protecting it against self-interested proposals by particular parties in many cases (see, e.g. Benoit, 2007: 382; Gallagher, 2005a: 514; Nikolenyi, 2011). Birch (2003: 32) cites constitutional embeddedness as a key explanation of the variation in electoral system stability in Eastern Europe. The case of the Czech Republic highlights how constitutional embeddedness may protect an electoral system in two important ways: firstly, in 1999 a majoritarian reform of the constitutionally embedded PR system failed as the governing coalition planning the reform lost the three-fifths majority in the Senate necessary for constitutional amendment. Later, a similar reform alternative was largely trumped by the constitutional court with reference to the constitutional principle of proportionality (see Nikolenyi, 2011: 617–618). In Slovenia in 1996 even the outcome of a popular referendum was overruled by the constitutional court in favour of the PR electoral system (Birch, 2003: 32). Thus, next to the increased legal protection (typically, even a two-thirds majority is needed to amend a constitution), a constitutional principle of representation may function as a point of reference for judges as they evaluate the constitutionality of a reform proposal. In sum, the case study literature appears to echo the expectation that constitutional embeddedness leads to greater resistance of the electoral system to electoral reform.
Finally, in light of these mixed findings, it is important to address the validity of inferring general normative goals from constitutional texts. While the empirical evidence suggests that constitutional principles typically emerge where the goal of the electoral system is generally agreed upon and where uncertainty about future election outcomes is high, this is not to say that constitution drafters might not succeed in strategically placing insincere principles into constitutions. Here, future research will need to investigate individual constitutional choices more closely in search for such insincere normative provisions. Yet, based on the case study literature cited above, the validity of constitutional texts might be questioned more strongly from another perspective: should interpretations of principles vary between countries or constitutional provisions be outdated because normative preferences have changed, 10 the measurement of the principle-dimension solely based on constitutional texts would probably be a fallacious strategy. However, the evidence shows that constitutional embeddedness has an impact (even if not consistently so for technical details) and also that the content of constitutional principles can usually be linked to ‘electoral system reality’ in a meaningful way with little to no evidence that constitutional provisions are outdated or of negligible importance.
Conclusion
The survey of constitutions has disclosed interesting findings that call for a partial reassessment of some common claims in electoral system research. On the one hand, the exploration of constitutional embeddedness highlights that to fully capture the normative principle-dimension of electoral systems, solely examining constitutions will not suffice, as only an overall minority of 26 constitutions include an overarching normative principle for the electoral system. On the other hand, the analysis has shown that neglecting the principle-dimension could lead to serious misconceptions when inferring normative goals from technical design and evaluating electoral systems based on false assumptions. It has also revealed that constitutional embeddedness appears to positively affect the stability of the electoral system. It is thus important to consider constitutional embeddedness for methodological and substantial purposes. Moreover, this analysis highlights how the combination of a parsimonious general source and information about individual cases is very fruitful for theory development. In that spirit, the closing paragraphs highlight paths for future research as they emerge from the empirical analysis.
Regarding the principle-dimension of the electoral system, future research should not only consider the means–ends connection between design and principle, but focus on the dominance of the proportionality principle on the constitutional level. The key question is whether this dominance is due to (idiosyncratic or regional) historic developments or whether the proportionality principle is considered to be more worthy of constitutional entrenchment than other principles. Answering this question would go a long way toward explaining patterns of electoral reform and might address the hypothesis that, eventually, only PR electoral systems will remain (Colomer, 2004; but see Harfst, 2013).
Furthermore, mixed electoral systems remain an intriguing topic as, similar to the question of whether they present a superior middle ground or a poor compromise, the assessment of constitutions suggests that genuinely mixed normative principles exist but are few in number. Hence, other mixed systems may come without any consistent principle upon which they could be evaluated unambiguously. Similarly, specific technical elements such as legal thresholds are not seen as violating the proportionality principle in some countries yet are clearly linked to the majority principle in others. The task of disentangling the principle-dimension and the technical design of electoral system calls for research that assesses such seeming inconsistencies.
Finally, the most crucial question for future research is that of whether and when constitutional principles make a difference. Does it matter that the electoral system is embedded in the constitution? An initial assessment has shown that the effects for the specifics of any electoral system appear to be contingent upon contextual factors. At the same time, constitutional embeddedness appears to be leading towards greater stability of the overall electoral system. These preliminary findings demand increased theoretical and empirical scrutiny. A closer investigation of the consequences of constitutional embeddedness could also help to assess whether the role of the electoral system is always as consequential as commonly assumed. Especially for new democracies whose founders may be stuck between aiming for institutional stability and a sufficient level of institutional flexibility, research on these questions should be of great use.
Footnotes
Acknowledgements
I wish to thank Eric Linhart, Jasper Schwampe, the four anonymous reviewers of the journal and the editors for valuable advice and many insightful suggestions. I would also like to thank Roland Krifft and Joshua Vogel for research assistance. Finally, I am especially indebted to Lotta Hübner for continued assistance in putting together the constitution dataset.
Funding
This manuscript has been written in the context of the research project ‘The Comparison of Electoral Systems’ Functions with a Special Focus on Mixed Systems’ supported by the German Research Foundation (DFG).
Notes
Author biography
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
