In the summer of 2010, Prof. Matthew S. Shugart of the University of California, San Diego, served as an academic consultant for IDI's Forum for Political Reform in Israel. In this article, he explores Israel's problems of governance based on a comparison with other parliamentary democracies and proposes a series of reforms that would be beneficial in strengthening Israel's political system.
Many of the problems of governance in Israel arise from a rather unusual and "extreme" set of existing institutions. In comparison with other parliamentary democracies, Israel has a low "electoral connection," between the voters and both their representatives in the Knesset as individuals and their government, as a collective. Institutions that impel Israel politics in this direction include the single national district, the use of closed party lists, the requirement for an investiture vote, and a provision permitting the Knesset to dissolve itself. Beneficial reforms would include: districting, "semi-open lists, abolishing the investiture vote, making the no-confidence fully "constructive," and not permitting self-dissolution of the Knesset. These reforms would tend to enhance the role of voters in holding both individual MKs, and the executive as a whole, accountable. A specific proposal for semi-closed lists, called Intraparty D'Hondt, is advanced in an appendix.
In comparing Israeli political institutions with their counterparts in other established democracies, certain characteristics jump out at the observer as being highly unusual, if not unique. A premise of this essay is that many of the problems of governance for which Israeli democracy is criticized stem from these unusual institutional features, specifically characteristics of the electoral system and the executive-legislative structure. In the area of the electoral system, the use of a fully closed-list electoral system and a single nationwide electoral district greatly diminish the individual accountability of politicians before the electorate that they are to represent. In the area of executive-legislative structure, of course Israel's parliamentary democracy is broadly consistent with the most prevalent model used by the most well established democracies in the world. Nonetheless, its specific model of parliamentarism empowers the Knesset majority to a greater degree than many other models of parliamentary government would do; when combined with the high fragmentation of the party system, this empowerment of the Knesset means, in practice, a high degree of leverage to the smaller and more "sectarian" parties. In turn, this leverage for smaller parties weakens collective accountability of the executive to the nation's elected representatives. When a democracy has substantial weaknesses in both individual accountability and collective accountability, it has a damaging (and unusual) combination.
The relative weakness of both individual accountability and collective accountability inhibits the development of the "electoral connection" between voters and their government. In a parliamentary democracy, the only direct electoral connection is between voters and legislators; by definition, there is no direct vote for the head of government. Because the executive in a parliamentary system emerges from within, and remains accountable to, the parliament, it is crucial that the parties and legislators be responsive to the electorate. To put it simply, if parties and legislators fail at their responsibility to represent voters, then parliamentary democracy fails. So, improving the representative function of the parliamentary electoral system is fundamental for improving the overall performance of a parliamentary democracy.
Yet it is not enough for the electoral system and political parties to function well. Even if there is no direct relationship between the vote and executive formation—the latter resulting from post-electoral bargaining (unless there is a single electoral majority)—there can be little doubt that voters expect the executive-formation process to depend on their choice at election time. After all, under parliamentary democracy, the only mechanism by which voters can influence the direction in which the executive leads their country is by their votes in parliamentary elections. Thus when I use the term, electoral connection, I mean the relationship of both the legislature and the executive to how voters express their preferences at election time. Certain reforms to both the legislative electoral system and the executive-legislative relationship can enhance the individual and collective accountability that is necessary for parliamentary government to serve the interests of the citizens for which it governs.
In the overview that follows, I will discuss the electoral system first, followed by the executive-legislative structure. The reason for this order of discussion is not accidental. The electoral system is fundamental, in that it represents the first link in the "delegation chain" of parliamentary democracy.Among works of political science that conceive of democracy as a chain of delegation with voters as the ultimate “principal,” see Palmer (1995), Strøm (2000), and Shugart (2006). If the electoral system is not working in a satisfactory manner to transmit voter preferences into the legislative body, then the rest of parliamentary democracy is unlikely to be satisfactory. To the extent that the electoral system could be reformed to improve either individual accountability (through changes in either the list format or districting, or both), political parties could be induced to be better transmitters of popular preferences over the direction of the nation's government. Moreover, if electoral reform could reduce fragmentation (by encouraging some smaller parties to combine with other parties) the second link in the delegation chain of parliamentary democracy—from parliament to cabinet—would also function better. It is precisely because of deficiencies in this second link that various reforms—including the direct-election experiment of the 1990s—have been proposed to change the way the executive is selected and held accountable. After discussing the electoral system I will turn our attention towards potential changes to the rules of government-legislative relations that could be adopted while remaining firmly within the parliamentary tradition. I will then also review, with disfavor, some reform ideas that have surfaced that go outside of the parliamentary tradition.
I. Electoral system
The current Israeli electoral system, of closed-list proportional representation in a single 120-seat district with a 2% threshold, is one of the most "extreme" in the world (Shugart 2001). It is certainly the most extreme of any that has been in long-term use in a nation with clear democratic credentials. Rivals, albeit with higher thresholds, would be the 450-seat national districts, with closed-list PR, adopted in both Ukraine and the Russian Federation in recent years. Ukraine has a 3% threshold and the Russian Federation 7%; the latter is arguably no longer a democracy and neither has any considerable record of sustained democratic governance. The Dutch system is often mentioned in the same breath with Israel's, but that is only partly accurate, as the Dutch lists are not closedVoters in the Netherlands are actually required to cast a candidate preference vote, although candidates must obtain a fairly high quota of such votes to be elected out of the order established by the party before the election. See Andeweg (2005). and there is a (minor) territorial component to the system.Voters in the Netherlands are actually required to cast a candidate preference vote, although candidates must obtain a fairly high quota of such votes to be elected out of the order established by the party before the election. See Andeweg (2005).
It is useful to think of electoral systems according to two dimensions of representation: interparty and intraparty.For an elaboration of these concepts, see Shugart (2001, 2005). The interparty dimension has been the dominant focus of political-science research on electoral systems. It refers to the impact of the electoral system on the number of competing parties, the relationship of seats to votes (i.e. proportionality), and related indicators.The “state of the art” on the interparty dimension is represented by Taagepera (2007). The intraparty dimension is less studied, but refers to the impact of the electoral system on the incentives that politicians have to "cultivate a personal vote" (Cain, et al, 1987, Carey and Shugart 1995), which basically means whether they obtain votes because of their individual characteristics or performance record, or whether instead they win or lose office on account of voters' evaluation of their parties.
The current Israeli system is about as "extreme" as an electoral system could be—on both dimensions. With a single district of 120 seats, and a very low threshold, the interparty dimension is extremely proportional, permitting the representation of a very large number of parties, many of which are very small. Because the executive-legislative structure is parliamentary, the presence of many small parties makes coalition formation and maintenance challenging, and reduces the connection between the voters and the government that forms out of post-election bargaining.
The current Israeli system is also extreme on the intraparty dimension. The use of closed party lists means that incumbent members or candidates cannot be directly rewarded or punished by voters for their performance or suitability for office. In Israel, connections between voters and legislators are weakened, due to the fact that the three or four largest parties are electing large blocs of legislators, all of whose seats are safe at the general election unless the party as a whole has an unusually bad result.Below I shall suggest that the existence of primary elections to select candidates in some parties is not sufficient to mitigate this accountability problem.
In some countries that use closed-list systems, such as Spain, lists are short, because district magnitude tends to be relatively low.Spain has an average district magnitude of 6.7. Even the two largest districts (Barcelona and Madrid), which elect over 30 legislators each, have districts less than thirty percent the size of Israel’s. These two disricts represent less than 20% of all the seats in the Spanish Congress of Deputies. See Hopkin (2005). Lower district magnitude facilitates voter knowledge of candidates (Shugart, Validini, and Suominen 2005) because the number of candidates, per party and district, is small. Of course, lower district magnitudes also reduce proportionality,In fact, the Spanish system is about as majoritarian as a system could be and still be conventionally labeled as “proportional representation.” and thus there is to some degree a tradeoff on the two dimensions. Proportionality can be retained, alongside smaller districts, only through the use of a two-tier system, with the second tier compensating for disproportionality.
The other means to increase candidates' and legislators' connections to constituents is through the adoption of preference votes within party lists. In other words, replace closed lists with either open lists (in which preference votes solely determine the order in which candidates are elected) or semi-open lists (a hybrid format, also known as flexible lists, in which party-provided ranks and preference votes both enter into the final ranking of lists). Let us now take up each of these potential reforms—districting and preference votes—in turn.
Given the extreme effects, on both the interparty and intraparty dimensions, of the current 120-seat district, it would be desirable to reduce the district magnitude down to a range of 6-10 seats per district. Smaller district magnitudes imply greater incentives for legislators (and candidates) to "go visible," by which I mean establish relations with voters. Increased visibility of legislators and candidates implies personal reputations affecting the voter choice to a greater degree, and as a result, the smaller district magnitude would tend to change the sorts of candidates that parties select. More visible legislators, concerned about their personal reputations as good legislators, imply an enhanced electoral connection.
Assuming 120 seats in the Knesset, having magnitudes in the range of 6–10 implies drawing boundaries for 10-20 districts. Exactly how these districts should be drawn would need to be specified in legislation. One possibility that would not require drawing new district boundaries would be to use the seventeen existing administrative districts as electoral districts.The use of these existing subdivisions is inspired by the Report of the President of Israel’s Commission for the Examination of the Structure of the Government of Israel. With 120 Knesset seats, such a districting arrangement would result in an average district magnitude of seven, which is conveniently in the 6-10 range suggested above. However, given population distribution, these seventeen districts would actually vary somewhat in magnitude, with a likely range of about 4–10. While many districted PR systems employ a much greater range of magnitudes, variance in the magnitude of districts has significant political consequences. For instance, even independent of distributions of voter preferences, some districts would be plausibly represented by only two or three parties, while others could be represented by many more.Additionally, Monroe and Rose (2003) showed that magnitude variance results in partisan bias in favor of parties that have their strongholds in rural areas. However, the systems where this effect is most pronounced are those that have a much greater variance than the likely 4–10 if Israel uses its seventeen administrative districts as electoral districts.
A potential solution to the drawbacks of magnitude variance in a districted system is to adopt a two-tier system, so that votes cast in small-magnitude districts for parties that are unable to win there are not "wasted." Under a two-tier system, there would be a small number of compensatory mandates allocated at the national level to correct for disproportionality arising from the smaller basic districts. Two-tier systems, while necessarily more complex, potentially provide some of the benefits of local representation without sacrificing overall proportionality (see Elklit and Roberts, 1996).
In a two-tier system, how many of the 120 seats would have to be allocated as compensatory seats, rather than in districts? The share of compensatory seats need not be large, given the use of proportional representation already in the basic districts. In fact, it could be as low as ten percent of the total (12) and still provide a high degree of proportionality.Obviously having a compensatory tier means reducing the magnitude of the basic districts (assuming a fixed assembly size). However, the aggregate effect is small. If there were twelve compensatory seats, average magnitude in the basic tier would be about 6.4 (108/17) instead of 7.1.
Of course, the inclusion of compensatory seats would limit the tendency of the smaller basic districts to bring about an overall reduction in the number of parties, and would thus have less impact on the ultimate electoral connection between voters' choice of party and the formation of governments. It is likely to move somewhat in the direction of improving the electoral connection, however, as smaller parties that cannot win seats (or can win few) at the district level may decline in popularity after a move to a two-tier PR system. Some of them might eventually merge with larger parties, to the extent that being able to win district-level seats is perceived as an important characteristic of representation.It is not possible to predict how much being able to win district seats would matter in voters’ evaluation of parties, or aspiring candidates’ decisions about what party to join. It would depend perhaps on changed norms of legislator roles. For instance, does it emerge that voters expect legislators to attend to local interests when the electoral system, for the first time in Israel, defines constituencies geographically? The answer really cannot be known in advance. A safe assumption, therefore, is that a two-tier proportional system would result in relatively minor impact on the number of parties, and that most of its impact would be felt instead on the intraparty dimension. A change in list-type might increase this effect, by allowing what are now separate parties to co-exist within lists, with their candidates competing for preference votes. So let us now turn our attention to the intraparty dimension, with a consideration of preference voting.
List type: Introducing preference votes
The recommendation for districted PR (with or without a compensation tier), and the above sketch of its effects, stands independent of any change in the type of lists. That is, introducing districts would be beneficial with a retention of closed lists, and districting would be advantageous with an adoption of preference voting. However, even if the single 120-seat district is retained, opening up the lists to allow preference voting would enhance the electoral connection, by allowing voters to favor some candidates over others on a party list.
There are three basic types of list: closed, open, and hybrids often known as semi-open or "flexible" lists. With closed lists, as currently used in Israel, there is no preference voting for individual candidates. The voter must accept or reject the list as a whole, with the ranking agreed upon through internal party procedures (candidate selection committees, primaries, or whatever mechanism a given party uses). Under open lists, voters cast preference votes for candidates and if fully open, the list order is determined solely by how many preference votes each candidate receives at the general election. Hybrid, semi-open lists entail a mix of the two principles. The party ranks candidates on the list ahead of the election, as with closed lists, yet voters cast preference votes. The list order prevails except when some candidates have achieved a quota or threshold of preference votes, as stipulated in the electoral law. That is, some candidates may be elected out of the original list order, and even moved from unelectable to electable list ranks, on account of being popular enough to attract large numbers of preference votes.
Going from fully closed to fully open lists would be a drastic reform, and it is not recommended for a single large district.The largest district with fully open lists appears to be Colombia’s Senate (100 seats; parties may present lists that are either closed or open, but most choose open). One Brazilian federal congressional district and the statewide assembly districts in three Brazilian states use open lists and a district magnitude of 70 or more. However, semi-open lists could be workable even with the current extreme district magnitude. Under most semi-open list systems, the law stipulates a threshold of votes that an individual candidate must obtain in order to guarantee election. Candidates that meet this threshold are elected first, and then any remaining seats won by the party are filled based on their pre-electoral list order.
Experience with semi-open lists. The most common types of party list employed in the proportional representation (PR) systems of Europe are variants on the semi-open list. The closed list, as used in Israel, is actually uncommon among the longer established democracies. In fact, the only West European democracies to use closed lists without a "personal" element of any sort are Italy, Portugal, and Spain. Italy has had a closed list (and actually non-proportional) system only since 2005,The PR lists used from the end of World War II through 1992 were fully open. and Portugal and Spain are younger democracies than Israel. The longstanding PR systems of Europe use either semi-open lists (e.g. Austria, Belgium, Netherlands, Sweden) or fully open lists (e.g. Finland, Luxembourg, Switzerland).Germany uses closed lists for the half of its seats that are elected from party lists, but the country’s “mixed-member” system contains an important “personal” element via the election of the other half of the Bundestag from single-seat districts. Norway’s lists are almost closed, in that the rules make it exceedingly difficult for voters to change the order of lists, and apparently no candidate has ever been elected out of the order established by a party on its list.
One should not exaggerate the actual openness (or flexibility) of most semi-open lists. In actual practice, most candidates in most semi-open systems who are elected would have made it into parliament based on the order established by the party in any case. Many parties may have their candidates elected in a different order from that of the pre-electoral ranking, but it is usually the case that relatively few candidates who were ranked too low by the party enter parliament on the basis of preference votes.
The relative balance of pre-electoral list order and preference votes on the final order of election depends on the specific rules chosen. Many of the rules in use in Europe are actually not very "flexible" in their effect. For instance, in the Netherlands a candidate requires 25% of the electoral quota in order to be guaranteed a seat regardless of party-list rank. This is a high amount, and thus few candidates are elected this way: in three elections from 1998 through 2003, only one or two per election were elected who otherwise would not have been (Andeweg 2005). Similarly in Belgium since the First World War, only 1% of MPs have owed their election to preference votes rather than to the rank their party gave them, despite the fact that more voters are using preference votes over time: 16% in 1919 but 60% in 1999 (De Winter 2006: 421-2). These two examples suggest that it is possible for semi-open lists to be little different in practice from closed lists. If electoral reformers want to increase opportunities for preference votes to matter in deciding the election of popular, but lower-ranked, candidates, it is possible to adjust the rules to make this feasible.
A good example comes from the Czech Republic, which has used semi-open lists since the fall of the communist regime. Rules on preference votes have been altered twice. The threshold of preference votes for a specific candidate to be guaranteed election, regardless of pre-election rank, was 10% of the party's vote in 1996 and 1998. In 2002 it was dropped to 7%. These rules resulted in some candidates being elected due to their preference votes: twelve (out of 200 seats total) in 2002, but only six in 2006. For the 2010 election, the rules were changed: now a voter may cast up to four preference votes (instead of two, as in 2002 and 2006), and a candidate needs only 5% of the list's total vote to guarantee election. This modest-seeming change had a large impact on the election of candidates based on preference votes: In 2010 there were 46 so elected. In addition, only 86 incumbents were re-elected (down from 115 in 2006) and 44 women were elected (up from 31 the previous election), 14 of them due to their preference votes.The data on Czech elections cited here come from an ongoing research project of mine, and from personal correspondence with Roman Chytilek (Masarykova University, Brno, Czech Republic).
Even some rules on the allocation of seats by preference votes that might seem highly flexible turn out actually to be only minimally so, in that relatively few candidates turn out to be elected who would not have made it into parliament on their party-provided rank. An example of such a case comes from Slovakia, which is a good comparative case for Israel because of its use of a single national district of 150 seats (since 1998; regional districts were used in 1994, the first election for an independent Slovak parliament). The rules are that a candidate can be elected with 3% of his or her list's total preference votes. This flexible-seeming rule actually resulted in no candidate being elected solely on preference votes in 1998 and just one in 2002. The number so elected increased to 7 in 2006 and 11 in 2010, suggesting that voters may tend to use the preference vote more as they become more familiar with it (and presumably as candidates campaign more for preference votes).
So we have seen how important the specific rules of a semi-open list system can be: in principle, semi-open lists can mean little or no difference from what a closed-list system would have meant, or they can allow some candidates to vault over co-partisan candidates with more favorable party-given ranks, and make it into parliament. Even when few candidates are elected based on preference votes, the opportunity of voters to change the list order may impact party nomination decisions, in that a party has a stronger incentive to nominate appealing candidates when doing so may bring the party additional voters who want to cast a preference vote for some candidate. The appendix offers an overview of some additional issues related to semi-open lists, and sketches a proposed rule, which I call Intraparty D'Hondt, after the allocation rule, already used in Israel at the interparty level.The Bader-Ofer method used in Israel is identical in effect to the D'Hondt divisors The D'Hondt divisors, under this proposal. would also be used to allocate seats according to party-provided rank or candidate preference votes. The Intraparty D'Hondt rule would offer a more realistic chance of election based on preference votes for candidates with substantial popularity, while simultaneously ensuring that the more voters who cast list votes without indicating a candidate preference the more seats are elected according to list order. In this sense, the proposal is a true hybrid of open and closed lists, rather than one that leans strongly towards closed lists, as do many actual semi-open lists, as the above review suggested.
Semi-open lists vs. primary elections. Given the use of primaries by several Israeli parties to select candidates and establish their ranks, it is worth considering how lists that allow preference votes relate to primaries. The latter already give voters a chance to affect the lists, even in a closed-list system, so is there additional value in having semi-open (or open) lists? The short answer is that there is no necessary correspondence between the mechanism a party uses to select candidates and the type of list used in the election. For instance, Finland uses primaries despite also using fully open lists (Hazan and Rahat 2010: 49). Israel's use of primaries is a sort of "bypass" (Rahat 2008) around the problems of closed lists, but not a solution to those problems.
By their very nature, primaries involve a different electorate (or "selectorate") in the process of nominating and ranking candidates than does the use of a non-closed list in the general election. Turnout in primaries is almost always lower than in general elections, and there is also no guarantee that voters who participate in a party's primary have sufficient loyalty to the party to vote for its list in the general election. On the one hand, some loyal party voters do not bother to turn out to vote in the primary, and hence forfeit a chance to participate in the selection and ranking of candidates. On the other hand, some voters may turn out for the primary and then vote for a different party in the general election, perhaps because they are dissatisfied with the ranking that resulted from the primary, or another party lures them away with a better slate, or other reasons. (See Rahat 2010, for more on these points.)
Even if semi-open lists are adopted, it does not mean that primaries need to be abolished. The candidates who are on the list must still be selected in the first place (even in cases where a list is completely open). In semi-open lists, the candidates must still be ranked prior to the election. The difference that list type makes is whether the primary is the final word not only on which candidates are on the list, but also which candidates are in the "safe," "marginal," and "hopeless" list positions. With primaries and closed lists, the set of voters who turn out in the primary determine electability of all but the most marginally ranked candidates. With semi-open lists, there remains an opportunity for the voters who choose the party in the general election also to have a say in the final order of the list. So whether to use primaries and whether to use a list type other than closed are actually entirely separate dimension of electoral-system choice.
List type as a party option
Nearly all PR democracies impose a standard list type on all parties. However, the question of the order in which candidates are elected from a list, and how candidates cultivate ties with voters (if at all) are necessarily intraparty considerations. Therefore it is possible to give individual parties a choice of list type. There appear to be only two countries where parties may choose what type of list to present. In Denmark, a fully closed list is not an option, although some of the options restrict the ability of candidates' preference votes to change the party order rather more sharply than others (Elklit 2005). In Colombia, since an electoral reform effective in 2006, parties may present either a closed or an open list. Most have chosen open lists; those that have chosen closed lists have been (perhaps surprisingly) the more "personalized" elements of the party system: those parties that are essentially the campaign vehicle for either the first-listed candidate or for a leader who is not a candidate in the district (Pachón and Shugart 2010).
If Israeli parties were given the choice of a semi-open or closed list, it is likely that the larger aggregative parties would opt for semi-open lists, but that very small parties and those strongly controlled by a single leader or small clique of leaders (e.g. Shas and Yisrael Beteinu) would continue to opt for the closed list. Whether to require a semi-open list of all parties, or allow a choice, is a normative question. If one desires that most Knesset members have incentive to cultivate an electoral connection to voters, then one should make the semi-open list a systemic mandate. If, on the other hand, one favors party autonomy and a competitive environment in which the extent of the personal electoral connection itself becomes a matter over which parties compete, then giving parties the option may be justifiable. Providing a party option may also be a pragmatic way of making passage of a reform to the list type more politically feasible.
II. Executive-legislative structural reform
As reviewed in the previous section, a change in electoral system that retained PR, but with either a districting feature, a preference vote, or both, would help with giving voters more say in the process of government formation. It would do so by possibly encouraging some smaller parties to merge with larger parties and by encouraging parties to be more responsive to voter preferences when making up their lists for the general election. However, a reform of the electoral system—unless it is a drastic change away from proportionality—is likely insufficient to increase the electoral connection between voters and the ultimate process of forming a government after the election. The reason is that there are likely still to be many parties that must bargain to form a government, and once formed, governments would continue to be vulnerable to the potential "blackmail" by smaller sectarian parties seeking to use their leverage over government survival to extract concessions. Thus we must also turn our attention to potential reforms of the executive's relation to parliament. Before detailing the sorts of reforms that might be considered, it is worth thinking through some theoretical issues in parliamentary democracy.
The parliamentary form of democracy has a tension built into it. On the one hand, it is a system that, by definition, means that the executive is not chosen directly by voters, but rather through the intermediation of legislators and the parties to which they belong. On the other hand, especially in modern times, voters expect there to be an electoral connection between them and their government. If the election results in a single party (or electoral alliance) holding a majority of seats, then the tension is resolved, as the process of electing a parliament will have been essentially equivalent to electing the prime minister. However, the more fragmented the party system is, the less there is an electoral connection between the voters and the executive. In Israel, with one of the greatest degrees of party-system fragmentation in the world, the formation of a government rests to an unusually high degree on post-electoral bargaining between (and within) political parties, instead of a relatively straightforward electoral connection.
If the objective is to enhance the electoral connection between voters and the executive, then it would help to have rules that privilege the executive candidates in the immediate post-electoral bargaining and that privilege the sitting prime minister in the inter-electoral period. By "executive candidates," I mean the leaders of the parties that are the realistic contenders to become prime minister following the election. Privileging them means giving them enhanced formal powers that do not undermine the fundamental principle of parliamentary democracy—that the government depends on the confidence of the parliamentary majority.Below we will consider reforms that step outside of the parliamentary model. Let us consider first the rules for forming a government immediately after elections—the investiture process—and then the rules for replacing a government between elections.
Investiture: Forming a new government
A parliamentary system must have rules, whether formalized in a constitution or statute or existing only in convention, on how a government is formed. This is most important after an election, if there is no single party (or pre-electoral alliance) with a majority, as there may be no clear electoral mandate for one or another party to lead that government, if the election was close. Yet because a defining feature of parliamentary democracy is that voters do not vote directly for executive, but rather that the executive emerges from the elected parliament, it is important to devise rules of government formation that respect the electorate's preferences. Below we will consider rules for forming a new government midterm in the case in which the incumbent government has lost the support of parliament.
The rules for forming a government after an election may, or may not, require a formal parliamentary vote of "investiture." A comparison of rules on executive-legislative relations (Bergman et al 2003) reveals that Israel has highly parliament-centered rules. These may tend to enhance the leverage of smaller parties whose support is needed to secure a vote in parliament, for instance on the inauguration of a new government. It may seem "natural" that a majority vote in parliament would be required to approve a new government in a parliamentary democracy, as is the case in Israel.As provided in the Basic Law: The Government (2001), in article 12(d): When a Government has been formed, it shall present itself to the Knesset, shall announce the basic lines of its policy, its composition and the distribution of functions among the Ministers, and shall ask for an expression of confidence. The Government is constituted when the Knesset has expressed confidence in it, and the Ministers shall thereupon assume office. However, often this is not so.
What difference does it make whether there is a majority required to vote for a government (or prime minister)? The higher the threshold required for approval, the more potential there is for small sectarian parties that might be able to deliver a decisive block of votes to hold out and demand concessions in exchange for their public commitment to the new government. The highest threshold consistent with the parliamentary model would be an absolute majority voting in favor. This high a threshold is actually mandated in Germany and Spain, but not elsewhere. And neither Germany nor Spain has had the experience Israel has of numerous sectarian parties.The post-war German Federal Republic has so far not required more than two parties to form a governing majority. In Spain, there are regional parties that could be comparable to sectarian parties, but there is an important difference: when there is no majority party, one or more regional parties have been willing to vote for the largest party to form the national government in exchange for benefits for their region in Spain’s semi-federal framework. Other parliamentary democracies, including Israel, require a simple majorityAs in Belgium, Greece, Italy, and Luxembourg. (i.e. more votes in favor than against, but not necessarily half plus one), or less than an absolute majority voting against (as in Sweden since 1975).
By limiting the role of the parliamentary majority in the inauguration of the government, the electoral connection between the voters and the government can be enhanced because the leverage of the smallest minority parties is reduced when their votes are not required to invest a government with the legitimacy to rule. The role of the parliamentary majority, including small parties needed to make it up, is most limited by simply assuming an executive candidate who has met certain criteria in assembling a government (discussed below) has the confidence of parliament until there is an active assertion by parliament that, in fact, that candidate and his or her government lack such confidence. Put another way, the question is whether the burden of proof rests with the proposed new prime minister—proven by winning an investiture vote—or whether the burden of proof rests with an opposition to show lack confidence in the government emerging out of the just-elected parliament.
So far we have concerned ourselves with the ease of forming a government following an election. But who should form it, and how is a potential government proposed to parliament, if a formal vote is required? And, if no formal investiture vote is required, what are the criteria for determining that a government has been formed? There is a very strong norm across parliamentary democracies that the political leader who is to be tapped to try and form a government following an election is the candidate put forth by the largest party. However, few if any parliamentary democracies have a formal rule stipulating a first right to the largest party. The formal proposal power is often in the hands of the head of state (monarch or president), although in many parliamentary democracies even this step is completely informal.
What would be the consequences of a formal rule stipulating first right to the leader of the largest party? Such a rule could go some distance towards strengthening the electoral connection, as it may have a "psychological" effect on the voters, making them more likely to choose between the larger parties on the basis of their preferred prime minister, rather than vote for a sectarian party. That is, voters who are "on the fence" between a large party and a sectarian party might be more likely to choose the former if they are not indifferent among the executive candidates of the two (or three) largest parties.
It is important to bear in mind that there will be situations in which the leader of the largest party is not positioned to form a government, owing to the overall balance of partisan preferences within the Knesset. It is thus important that there be a "safety valve" to allow another party leader to form the government in cases where the largest party lacks sufficient allies to do so. One way to do this is the following: allow the leader of the largest party to become prime minister, without a formal vote in the Knesset, after a stipulated bargaining period (say, 28 days) unless an absolute majority has voted in favor of an alternative party leader prior to the completion of the formation period. By such a vote, parliament would have demonstrated that the actual "winner of the election" was the leader of the second (or third, etc.) largest party, given that winning an election in a parliamentary democracy means having a working relationship with the people's elected representatives in parliament and not simply having won the plurality. However, this process of electing an alternative to the leader of the largest party would take place only after the largest party had failed to prove that it in fact has "won the election" by securing support in the Knesset. The burden of proof, in this case, would have rested with the parliamentary majority to prove that it preferred a leader other than the leader of the largest party.
Changing governments between elections: Votes of no confidence and dissolution
During the inter-electoral period, the stability of government—and thus the electoral connection with the prime minister inaugurated following the most recent election—is most enhanced through the use of the "constructive vote of no-confidence." Israel already has a variant of the constructive vote of no confidence; however, it is a less complete form than is found in some other parliamentary democracies, including Germany, Hungary, and Spain. In these latter countries, a parliamentary vote can depose a prime minister and cabinet only by electing an alternative prime minister.Consider the wording in the Hungarian constitution, in Article 39A(1): A motion of no-confidence in the Prime Minister may be initiated by a written petition, which includes the nomination for a candidate for the office of Prime Minister, … Should, on the basis of this motion, the majority of the Members of Parliament withdraw their confidence, then the candidate nominated for Prime Minister in the motion shall be considered to have been elected. There is no provision for a period of bargaining over the formation of a new government, because the task of assembling a government has been delegated to the majority through the constructive no-confidence vote. Nor is there any discretion for the head of state. The provisions in Germany and Spain are essentially the same on these points. The Israeli variant is less strict, in that a vote of no confidence against the incumbent prime minister only nominates an alternative leader to attempt to form a government, rather than formally electing the replacement as prime minister.As provided in the Basic Law: The Government (2001), in article 27(b): An expression of no confidence in the Government will be by a decision adopted by the majority of the Members of Knesset to request that the President assign the task of forming a Government to a certain Knesset member who gave his written consent thereto. The key phrase there is, “assign the task of forming a government,” rather than “elect as prime minister,” or some such wording. The distinction is important, as merely nominating an alternative leader means that there still must be a process of government formation before that leader takes office as prime minister—and he or she might fail in the effort. Moreover, if any potential new government, following a successful no-confidence vote, requires a formal investiture vote before it can take office, then we are back to the situation of maximum leverage for small sectarian parties. They can vote against the incumbent prime minister by proposing an alternative candidate, but then hold out for concessions before actually granting their votes to the new potential prime minister.Even if actual successful no-confidence votes are rare, the institutional context sets the bargaining parameters. That is, small parties have the leverage described on account of the ability to threaten to join the opposition in a vote to re-start the government-bargaining process. With a fully constructive vote, no party will join a no-confidence motion unless it is sincere about preferring the alternative, and the negotiations over a cabinet now occur on terms more favorable to that alternative, who is already the new prime minister by virtue of having been elected as such by a majority.
The most important advantage of a fully constructive no-confidence vote is that it prevents the ousting of the incumbent government on a purely negative coalition, such as extremes of left and right agreeing that they dislike the government even if they would never cooperate and agree on what government should replace the one ousted. In a situation of oppositions on either flank of the government, the constructive vote may permit a form of separated powers otherwise unthinkable in a parliamentary system: a government opposed by a majority in parliament but nonetheless able to remain in power. The government would thus be "stable" despite lacking parliamentary support.
In all other situations, a constructive vote is unlikely to make any difference in government duration, because in most situations if there is a majority opposed to the incumbent it is likely that there is also a majority that prefers a single plausible alternative. Such a situation would occur if a centrist bloc supporting the current government changed sides, withdrawing support and offering it instead to the main opposition party. In such a situation, a constructive no-confidence vote and a "regular" (negative) no confidence vote would produce the same outcome. This is an argument in favor of the constructive vote, inasmuch as it does not undermine the core principle of parliamentary government—that the executive must have a working majority in parliament—but it nonetheless enhances the position of the incumbent in situations in which there is not a absolute and publicly articulated majority in favor of an alternative. It also, as noted, enhances the bargaining power of a replacement prime minister, who has received an affirmative majority vote at the same moment that the preceding government was voted out.
What if the incumbent government has lost the confidence of parliament but no alternative government exists that could command this confidence? Most parliamentary systems have rules permitting an early dissolution, to refer the political conflict back to the electorate. However, this not a defining feature of parliamentarism. Norway, for example, has no provision for early elections. In countries where there is a provision for an early election, the decision to dissolve parliament often rests with a head of state. Alternatively, the right to dissolve may be granted to the prime minister, as a prerogative of holding the office.
It may seem "obvious" that another mechanism for an early election in a parliamentary democracy would be for parliament to vote to dissolve itself. Yet, there appears to be only one case of parliamentary democracy that actually has such a provision, and that is Israel. In none of the parliamentary systems surveyed by Bergman, et al, (2003) is there a provision for a vote by parliament to call an early election. In these European countries recourse to the electorate is always a prerogative of the executive, whether the head of state or the head of government. So here, in addition to the investiture rules discussed above, is another area in which Israel's democracy is extremely parliament-centered, even by comparison to other parliamentary systems.
A drawback of allowing parliament to dissolve itself, by majority vote, is the potential for increased leverage for smaller parties. They can defect from the existing majority and join with the opposition to call a new election, without first having to go through any formal no-confidence procedures against the incumbent government. As with no-confidence votes themselves, a vote to dissolve the parliament need not actually be passed in order for the existence of the option to be a credible threat, thereby affecting the bargaining leverage of the various parties.
If, on the other hand, the parliament can not be dissolved in this way, then a dissolution prior to the scheduled end of a term will only result in the event of some cabinet crisis, or when the executive dissolves parliament, if rules permit that path to early elections. Dissolution as an act by the cabinet or prime minister, unlike parliamentary self-dissolution, is common in parliamentary democracies. However, it is often criticized as giving the executive undue leverage, allowing the incumbent prime minister to time elections to his or her maximum electoral advantage. Because of these concerns, there has been some movement towards "fixed-term" parliaments (such as Norway has had for many years), including in some Canadian provinces, the Scottish parliament, and as a result of the recent coalition agreement in the UK following the 2010 election. The potential drawbacks of dissolution by the executive are considerably lessened in a case like Israel, due to the existence of coalition governments that contain several parties. Due to such coalitions, and their dependence on post-electoral bargaining, elections are rarely about the "reelection" of the government, as they are in Westminster-type parliamentary systems. However, it is the same tendency towards multi-party governments that makes parliamentary self-dissolution potentially undesirable.
If our goal is to enhance the electoral connection in a multiparty parliamentary democracies, it is better to require that the formal no-confidence procedures precede any recourse to dissolution, for the following reason. The existence of a cabinet crisis occasioned by the tabling (if not actual passage) of a no-confidence motion, or by the withdrawal of some party from the current coalition, provides valuable information to the electorate in the event of an early election.Through 2010, only six elections in Israeli history have been held at the maximum constitutional inter-electoral period. All the rest have been “early,” eight of these by a law passed by the Knesset. (These numbers are based on personal correspondence with Ofer Kenig.) It thus increases the opportunity for the electorate to use the subsequent election as a means to shape the parameters of the next round of parliamentary bargaining, compared to an early election called without this information. In this sense, preventing parliamentary self-dissolution is complimentary to the other provisions discussed above, such as a first right in post-election government formation to the largest party and the constructive vote of no-confidence.
Considering all of these points, it is advisable to allow the prime minister to propose to the head of state that the latter dissolve the Knesset, but not to make dissolution a unilateral power of either the prime minister or the Knesset itself. In this way, dissolution is a potential threat that the prime minister holds over an unruly majority, but the president could veto a request for dissolution as a check on abuse of the power. This combination of prime minister's proposal of dissolution, but president's final decision, would be likely to be operative only in a situation in which a majority is not consistently backing the incumbent, but in which there also exists no majority prepared to vote "constructively" for an alternative. It would also be a provision that a replacement prime minister, following a (constructive) vote of no confidence, might use if the majority voting him or her into office had revealed itself to be insincere despite having joined the coalition in favor of replacement. These are precisely the sorts of situations in which recourse to the electorate most preserves the core parliamentary principle that the electoral connection to the government runs via the elected representatives of the people.
More radical reforms?
There are those who advocate reforms that go even farther in strengthening the position of the executive. We need to consider whether such reforms are suitable for Israel. Suggestions include having a second round among the top two party leaders if no party obtains some threshold number of seats, or requiring a super-majority of the Knesset to vote no confidence.
Either of these reforms, whether adopted separately or jointly, would result in a fundamental break with the structure of parliamentary democracy. Under parliamentary democracy, the government is that party or combination of parties that can command the confidence of the elected representatives; it remains in office only so long as a majority does not vote to remove it (or for early elections, under current Israeli procedures). Either proposed reform thus ends the parliamentary system of government, either by allowing a popular vote to determine who heads the government, or by allowing a government to remain in office even if a majority of the legislature has lost confidence in it.
It is important to recognize that there appears to be no precedent anywhere in the world for either proposal. Thus there are no models to draw on. However, we now have a good deal of evidence about the consequences of "presidentializing" institutions—those that separate the executive from the legislature either in its origin (how it is formed) or its survival (how its term is ended, or whether the term is fixed). Presidentializing reforms result in weaker ties between the executive and the legislature, a markedly lower level of political experience of those typically chosen to be executive candidates, different electoral constituencies for executive and legislature, and a higher propensity for the executive to deviate from its campaign commitments (Samuels and Shugart 2010).
We can define an institutional reform as presidentializing if it either allows a popular vote, separate from that for legislators, to determine who will lead a government, or generates a "separation of powers" by which the executive and legislative branches may coexist despite the absence of a working political relationship between them. Thus either proposal falls in the category of "presidentializing." Let us consider each in turn.
A Runoff to Determine Who Forms the Government. Under one proposal, a failure of any party to surpass a threshold of one third of Knesset seats would result in a second round of popular voting between the leaders of the two largest parties. The winner of the runoff would then form the government.
This proposed change would not be notably different in its likely effects from those of the prior (and by most accounts highly unsatisfactory) experience in Israel with a directly elected prime minister. The executive–legislative deadlocks of that period, and the legislative fragmentation, can be attributed in part—but only in part—to the separate vote for executive and legislature. That is, voters could split tickets, favoring a prime-ministerial candidate of one party but a legislative list of another. In response to that experience, some proposals now would have a single vote in the first round for legislative list (with the PM candidate heading the list). Nonetheless, it would be mistaken to assume that the presence of a single vote for the first round of elections would be sufficient to induce parties to aggregate onto larger lists to try to prevent a runoff, as seems to be the intention behind the proposal. In fact, it is very likely that both big and small parties would "play for the runoff." That is, big parties would see an opportunity to win the eventual runoff by nominating a popular candidate, potentially even one without strong connections to the party organization. This is the core of what can be called the "presidentializtion" of parties—maximizing votes by selecting strong and popular personalities. Small parties, on the other hand, would have every incentive to maximize their Knesset representation by running separate lists (as they do now, and did under direct election), and then bargain to deliver their support to one of the parties left standing in the runoff.
As noted previously, there is no precedent for a system like this proposal, so we cannot say for certain what would result. But any institutional reform that leaves open the possibility of a popular vote being decisive for the selection of the head of government provides strong incentives for parties to become "presidentialized."
An Extraordinary Majority to Terminate the Government. A second reform would require a majority greater than 61 MKs to terminate the government's tenure via a vote of no confidence. Like the idea of a runoff system for originating an executive, any reform allowing a government to survive independently of the legislative majority (that is, half the members, plus one) is a fundamental break with parliamentarism. It raises the prospects that a government might remain in office despite being unable to obtain legislative approval of its policy program—a feature that is common in presidential systems. It greatly increases the risk that the government might deviate from the policy commitments made either at the election or in a post-election coalition agreement, without the executive's party or its supporting legislative coalition being able to do anything about such a breach of commitment.
Either proposed reform would move the political system in a "presidential" direction, bringing about some of the disadvantages of the presidential form of government. If both were adopted together, it would be almost as if an actual presidential system were adopted. While presidential systems have worked in various countries, their record taken as a whole has been one of worse executive–legislative relations, less political experience in the executive, and lower accountability, relative to the record of parliamentary democracy, including the record in Israel.
In comparison with other parliamentary democracies, Israel has a low "electoral connection," between the voters and both their representatives in the Knesset as individuals and their government, as a collective. The roots of this low electoral connection may be found in both a very extreme electoral system—a single nationwide district with closed party lists—and a highly parliament-centered process for forming government and determining when to have an early election. In a parliamentary democracy, the electoral connection between voters and government necessarily runs through the parliament. Nonetheless, there are reforms to political institutions that can be taken to enhance the electoral connection and, by doing so, to improve the overall legitimacy of the democratic system.
Reforms to the electoral system could include the adoption of districting (with or without national compensation) and a move to semi-open lists. Either reform, but especially both taken together, would increase the "visibility" of individual MKs and candidates. Because they would be elected from smaller district lists and/or with the assistance of preference votes cast for them as individuals, they would have a far stronger incentive to connect with voters and be responsive to voter concerns.
Reforms to the executive–legislative structure could include removing the requirement of a formal vote of investiture to form a new government after an election, and adopting a fully constructive vote of no-confidence for those cases when a new government may be needed to replace the incumbent government between elections. By mandating that the right to form a new government after an election be granted to the leader of the largest party, the electoral connection is strengthened because these rules may create a "psychological effect" among the voters that encourages them to use their vote as a means to indicate a preference over prime minister candidates. By allowing this leader to become prime minister without a formal vote in favor, the leverage of small parties would be reduced. In combination these rules shifts the burden of proof from the prime-minister-designate to the opposition, to the strategic benefit of the former. Similarly, mandating that a vote of no confidence be constructive, meaning that its passage automatically elects a new prime minister, ensures that smaller parties defect from a governing coalition only when they sincerely prefer the alternative, for which they must openly vote. Abolishing the right of the Knesset to dissolve itself also strengthens the electoral connection, because it has the same basic effect as the other changes discussed here to executive-legislative relations: a reduction in the leverage of small sectarian parties.
Taken together, the sets of reforms suggested here are intended to empower voters by making individual representatives accountable to their voters, through districting and/or preference votes, while also making the government more collectively accountable to the electorate through strengthening the role of the prime-ministerial candidates.
AN INTRAPARTY D'HONDT FORMULA FOR ALLOCATING SEATS TO CANDIDATESI gratefully acknowledge the assistance of several noted scholars of electoral laws, who helped think through the issues raised here: John Carey, Gary Cox, Richard Katz, Arend Lijphart, David Samuels, Michael Thies, and especially Rein Taagepera. Of course, none of them deserve any blame for errors. An earlier version of this memo was prepared as an appendix to the paper, “Deepening Democracy by Renovating Political Practices: The Struggle for Electoral Reform in Colombia,” which I co-authored with Erika Moreno and Luis Fajardo. A revised version of the original paper appears in Christopher Welna and Gustavo Gallón, eds., Peace, Democracy, and Human Rights in Colombia (Notre Dame University Press, 2007).
This appendix offers a proposal for implementing the principle of semi-open (flexible) lists that differs from the standard formulas in use. Most actual semi-open employ a quota or threshold of preference votes that candidates must meet in order to guarantee their election independent of their party-provided rank. As explained below, the use of a quota may have some undesirable features. Alternatively, one could use a divisor method, such as D'Hondt, as most of the world's proportional systems (including Israel) do when allocating seats to parties. That is, this proposal would apply a common interparty allocation rule also for the intraparty dimension—the allocation of seats to candidates on a list.
In most cases, the rules of intraparty allocation in semi-open list systems establish a quota of preference votes, defined either as a share of the party's vote in the district or of the total vote cast for all parties in the district. Candidates who meet this quota are guaranteed election, regardless of their party-provided rank. For any seats not filled through individual quotas, the party rank prevails. (See Katz, 1986, for details.) In practice, semi-open list systems tend to set fairly high quotas for upsetting the list order. Thus most seats are assigned in the order established by the party. Of course, the quota can be set at a lower level, making it easier for preference votes to matter in the allocation of seats (such as the example from the Czech Republic cited in the main body of this essay).
Rather than use a quota-based method, one can use a divisor method of intraparty allocation instead. One such method is the common D'Hondt divisor sequence.For an overview of quota and divisor methods, with respect to interparty allocation, see Taagepera (2007): 29–34. Just as D'Hondt allocates seats to parties by first dividing each party's votes by the numbers 1, 2, 3..., we can allocate seats within a party the same way, applying the divisors to list votes and individual candidate votes. The intraparty vote shares to which the divisors are applied then are the number of votes cast solely for the list, which we will call vL, and the individual vote totals of any candidates. It deserves emphasis that this proposal works only if voters have the option of casting a vote for the party without indicating a candidate preference.
To explain this method, we will first look at a hypothetical allocation in a modest-sized district, which keeps it manageable to follow the steps involved. Then we will see an example applied to a nationwide district that is more comparable to the current Israeli system. The first two illustrations, in a hypothetical six-seat district, are given in Table A1. Both examples depict situations in which one or more candidates ranked low by the party gained considerable preference votes. It should be noted that this is the only situation in which it will matter that a system takes preference votes into account as well as list votes, for if voters tend to give their preference votes to the top-ranked candidates, then the same candidates are elected either way. The implication in such a case would be that parties have good information about the popularity of their candidates when constructing lists, or that voters simply follow party cues, understanding high-ranked candidates to be quality candidates.
In both examples shown in Table A1, the party has obtained 1000 votes and four seats, with six candidates having been nominated. In the first example, 300 of the party's voters cast their vote for the party list with no preference vote, and the two candidates that the party ranked fifth and sixth were in fact its two most individually popular candidates. Under the D'Hondt divisors, the first seat goes to the list (and, therefore, to the candidate with the first rank). The candidate who was ranked fifth on the list has a total number of preference votes greater than half the votes given for the list without preference (vL), so that candidate wins the second seat. No candidate has more than vL/2, so the second candidate on the list gets the third seat. The fourth seat goes to the candidate with the next highest preference vote total, unless this total is less than vL/3. In this example, that is the candidate whom the party had ranked sixth, who obtained 143 preference votes.
In the second example, twice as many votes were cast for the party were cast without a candidate preference as in the first example: 600 out of 1000. Only one candidate received a preference vote total greater than the fourth quotient derived from applying the D'Hondt divisors to the votes cast for the list without preference. Thus the first three candidates on the list are elected as ranked by the party, and one popular candidate ranked low by the party is elected ahead of the candidate that the party had ranked fourth. Each of three seats allocated to the list cost 200 votes, and the fourth seat could not go to a candidate with less than 150 preference votes (vL/4).While, in this example, a candidate was elected who had only 11 preference votes, it must be borne in mind that fully 600 voters (out of 1000) had voted for the list order, thereby delegating to the party the right to select some candidates.
It is a well known effect of D'Hondt that it favors larger components over smaller in seat allocation. That effect applies whether those components are defined at the intraparty or interparty level. Thus if, as will usually be the case, the largest intraparty component of the vote is votes cast solely for the list, rather than for any one candidate, then the list as ranked before the election will be entitled to most of the seats. No candidate can win with preference votes less than vL/s (where s is the number of candidates elected by the party), but usually the minimum share needed to win would be much higher than vL/s. The larger the percentage of list-only votes, the greater the number of seats allocated that way, and correspondingly the greater number of preference votes a candidate would need to win independent of list rank.
Finally, it might be useful to compare how the intraparty D'Hondt would differ from an actual allocation in an existing quota-based semi-open list system. Here we will consider one party in Slovakia, where the 150-seat national district makes the overall electoral system quite comparable to a potential Israeli semi-open system (if districting is not adopted). The two methods will often result in the same candidates winning seats, but there are, of course, also situations in which they will differ. One example of a party for which it would make a difference, shown in Table A2, demonstrates the conditions under which intraparty D'Hondt would be arguably superior. This case is the Movement for a Democratic Slovakia (HZDS), which won 36 seats and was the largest party in 2002; its 24% of seats puts it about on par with the largest party in Israel's 2009 election.
The candidate votes and ranks come from the national list of the Movement for a Democratic Slovakia (HZDS) in 2002
The table includes just under half of all the candidates on the party's 150-candidate list, and orders them according to the preference votes they actually obtained. The table shows these votes, as well as their share out of the total number of preference votes cast for HZDS candidates. It indicates which candidates were actually elected. All 36 elected candidates would have won even if the pre-electoral rank were the only criterion; preferences votes actually did not affect who was elected. Only four candidates even surpassed the 3% of preference votes needed to guarantee their election, regardless of rank; in any case, these were also the candidates with the top four positions on the party list. Clearly, allocation in this party was effectively no different from a closed list.However, the order in which the 36 were elected (not shown) differed slightly from their rank on the list
The table shows how the allocation would have proceeded under Intraparty D'Hondt, using an estimated 22% of votes cast solely for the party list.The Slovak electoral authorities do not report the actual number of voters who cast a list-only vote or preference votes. They simply report totals for each category; given that each voter may cast up to four preference votes, knowing the latter number is insufficient to know how many cast what number of such votes. Separate columns show which candidates would have been elected off the party list, according to the pre-electoral rank, and which ones would have won a seat based on their own preference votes. The application of D'Hondt divisors to the estimated list-only vote total of 371,763 and to each individual candidate's preference-vote total results in the list being entitled to 23 seats, while 13 candidates had sufficient votes to win on their own. In the actual allocation there were no candidates who won solely based on preference votes, despite two of the candidates whom the party ranked outside of the top 36 having top-ten preference-vote totals; one of these candidates was ranked 60th on the party list. Under intraparty D'Hondt, candidates who were this popular would win, in spite of their pre-electoral list rank. The two actually elected candidates whom these two would have supplanted were ranked 35th and 36th by the party, but 26th and 60th by the voters. The reason the candidate ranked 26th by the voters would not have won a seat under Intraparty D'Hondt is that, as noted above, only 13 candidates were able to win on their preference votes. Only these 13 had personal vote shares that were greater than one 36th of the total number of party votes—the minimum under Intraparty D'Hondt for a candidate to be elected by preference votes. Contrast this with the actual quota-based allocation used in Slovakia, where 3% of preference votes is need to guarantee a win. Only four candidates reached this quota, and because they were ranked in the top four by the party, they would have been elected in any case. Two other candidates with preference ranks in the top ten, but who did not reach 3% of the party's total preference votes, did not win seats due to their low party-given rank.
Of course, what we have seen in Table A2 here is only one example from an actual allocation that proceeded under different rules. Nonetheless, it illustrates that the Intraparty D'Hondt rule is about as balanced a compromise between the principles of closed-list and open-list voting that could be offered. Unlike most quota-based methods, it ensures that popular candidates can win based on their preference votes, while setting a threshold on election by preference votes that varies according to how many seats the party wins and how many voters choose to cast a list-only vote. For these reasons, intraparty D'Hondt is worth considering as a compromise between the principles of closed-list and open-list proportional representation.
Rudy B. Andeweg. 2005. "The Netherlands: The Sanctity of Proportionality." In The Politics of Electoral Systems, ed. By Michael Gallagher and Paul Mitchell. Oxford: Oxford University Press.
Bergman, Torbjörn, Wolfgang C. Müller, Kaare Strøm, and Magnus Blomgren. 2003. "Democratic Delegation and Accountability: Cross-national Patterns," in Delegation and Accountability in Parliamentary Democracies, ed. by Kaare Strøm, Wolfgang C. Müller, Torbjörn Bergman. Oxford: Oxford University Press.
Cain, Bruce, John Ferejohn, and Morris Fiorina. 1987. The Personal Vote: Constituency Service and Electoral Independence. Cambridge, Mass.: Harvard University Press.
De Winter, Lieven. 2006. "Belgium: Empowering Voters or Party Elites?" In The Politics of Electoral Systems, ed. By Michael Gallagher and Paul Mitchell. Oxford: Oxford University Press.
Elklit, Jorgen, and Nigel S. Roberts 1996. "A Category of its Own: Four PR Two-Tier Compensatory Member Electoral Systems." European Journal of Political Research 30(2):217-240.
Elklit, Jørgen. 2005. "Denmark: Simplicity Embedded in Complexity (or Is it the Other Way Around?), in The Politics of Electoral Systems, ed. By Michael Gallagher and Paul Mitchell. Oxford: Oxford University Press.
Hazan, Reuven Y., and Gideon Rahat. 2010. Democracy within Parties: Candidate Selection and Their Political Consequences. Oxford: Oxford University Press.
Hopkin, Jonathan. 2005. "Spain: Proportional Representation with Majoritarian Outcomes." In The Politics of Electoral Systems, ed. By Michael Gallagher and Paul Mitchell. Oxford: Oxford University Press.
Katz, Richard S. 1986. "Intraparty Preference Voting." In Bernard Grofman and Arend Lijphart, eds., Electoral Laws and their Political Consequences. New York: Agathon Press.
Monroe, Burt L., and Amanda G. Rose. 2002. "Electoral Systems and Unimagined Consequences: Partisan Effects of Districted Proportional Representation." American Journal of Political Science 46, 1 (January): 67–89.
Mónica Pachón and Matthew S. Shugart. 2010. "Electoral Reform And The Mirror Image Of Interparty And Intraparty Competition: The Adoption Of Party Lists In Colombia," Electoral Studies. In Press.
Matthew S.R. Palmer, "Toward an Economics of Comparative Political Organization: Examining Ministerial Responsibility," Journal of Law, Economics, and Organization.
Rahat, Gideon. 2008. "Trial and Error: Electoral Reform through Bypass and its Repeal." Israel Affairs 14:1 (January): 103–17.
Rahat, Gideon. 2010. "The Political Consequences of Candidate Selection to the 18th Knesset." In The Elections in Israel—2009, ed. by Asher Arian, Alan Arian, and Michal Shamir. Transaction Publishers, 2011.
Samuels, David J., and Matthew S. Shugart. 2010. Presidents, Parties, and Prime Ministers: How the Separation of Powers Affects Party Organization and Behavior. New York: Cambridge University Press.
Strøm, Kaare. 2000. "Delegation and Accountability in Parliamentary Democracies,"
European Journal of Political Research 37, 3 (May): 261-289.
Shugart, Matthew S. 2001. , "Electoral Efficiency and the Move to Mixed-Member Systems." Electoral Studies 20, 2 (June, 2001): 173-193.
Shugart, Matthew S. 2005. "Comparative Electoral Systems Research: The Maturation of a Field and New Challenges Ahead," in The Politics of Electoral Systems, ed. By Michael Gallagher and Paul Mitchell. Oxford: Oxford University Press.
Shugart, Matthew S. 2006. "Comparative Executive-Legislative Relations", in The Oxford Handbook of Political Institutions, edited by R.A.W. Rhodes, Sarah Binder, and Bert Rockman. Oxford: Oxford University Press.
Shugart, Matthew S., and John M. Carey 1995. "Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas." Electoral Studies 14, 4 (December, 1995): 417–39.
Shugart, Matthew S., Melody Ellis Valdini, and Kati Suominen, "Looking for Locals: Voter Information Demands and Personal Vote-Earning Attributes of Legislators under Proportional Representation." American Journal of Political Science 29, 2 (April, 2005): 437–449.
Taagepera, Rein. 2007. Predicting Party Sizes: The Logic of Simple Electoral Systems. Oxford: Oxford University Press.
Matthew S. Shugart is The author is Professor of Political Science, University of California, San Diego. He was a Visiting Professor and Lady Davis Fellow at the Hebrew University of Jerusalem (May–July, 2010) and at Consultant at IDI during the same period.