Submitted in partial fulfillment of candidacy requirements
A significant debate in comparative politics concerns the relationship between political institutions and party systems. Current theories only offer basic models of how regime types—the ideal types of presidentialism, parliamentarism, semi-presidentialism, and president-parliamentarism, for example, as well as their hybrids—interact with electoral systems and social cleavages to produce party systems. These models usually take the form of distinguishing between presidential and parliamentary systems, where presidential systems are generally held to reduce the effective number of legislative parties relative to parliamentary ones, dependent upon the electoral cycle and the permissiveness of the presidential electoral system. Surely, however, this is not the only difference in regime type that matters: it is now well acknowledged that significant differences exist within regimes broadly classified as presidential or parliamentary. Further, there are the less common ideal types of semi-presidential and president-parliamentary regimes to consider, not to mention the many hybrids that do not exactly approximate any one ideal type.
The lack of theorizing on this matter is surprising because the type of regime in which political competition takes place certainly affects the strategic calculations of voters and political elites alike. The decision to vote strategically is not only a function of the electoral system but also of the way in which political power is structured within a regime. Similarly, the strategic choices faced by party leaders about how to fight elections—when, if, and what kind of alliances with other parties should be formed—are at least partially a function of the regime type under which elections are held. In other words, this paper agrees with existing theories that both the propensity for and the qualitative kinds of strategic behavior that actors engage in vary with the type of political regime, as different regimes confront actors with coordination problems of varying difficulty that impose varying penalties for coordination failures. However, this paper argues that existing theories fail to account for significant cross-national and cross-temporal variation in party systems because they only model the variable of regime type as a dichotomy between parliamentarism and presidentialism. The paper extends existing models by hypothesizing that the strategic incentives for coordination in electoral contests vary with the distribution of political power within a regime: the more that a regime concentrates political power in one political body such as the executive or the legislature, the greater the incentive for actors to construct strategic alliances in the electoral contest for that body, and thus the greater consolidatory pressure on the party system surrounding it.
This paper demonstrates the short-comings of existing theories by examining Israel’s experience with a novel political regime from 1996-2001, a natural experiment where regime type is allowed to vary but social cleavages and electoral system are held relatively constant. It provides further evidence to this effect by comparing Israel’s executive party system with the executive party systems of France and Peru, polities that are somewhat similar in social cleavages and that employ the same electoral system in the selection of their executive, yet operate under different political regimes. After illustrating the problems with existing models of party systems, the paper will develop an alternative approach that rejects the conventional presidential-parliamentary dichotomy for a more detailed classification of regimes by the distribution of political power (political constraints) within each regime.
Accordingly, the paper first reviews the literature relating political institutions and social cleavages to party systems. It then derives testable hypotheses from this literature about how party systems should vary when variables such as social cleavages are controlled for and other variables such as regime type are allowed to vary. It turns to the Israeli case to test these hypotheses. Finally, it concludes that existing models need further refinement and outlines the logic of such an alternative (more refined) model; in doing so, it suggests how strategic coordination might vary from regime to regime.
Academic Literature: Party Systems This section of the paper analyzes the academic literature surrounding the dependent variables of the national legislative and executive party systems. By national legislative party system, I refer primarily to the effective number of parties winning seats in the national legislature, a statistic that characterizes a party system by both the number of competitors and their relative sizes.1 Similarly, by national executive party system, I refer to the number of candidates (or the number of parties) running for the national executive office, if the executive is elected separately from the legislature. Usually, but not always, this executive is termed a president. In brief, three independent variables have been utilized to explain both cross-national and cross-temporal variance in party systems: social cleavages, electoral systems, and—more recently—regime type. The latter two variables could clearly be subsumed by a more broadly defined variable of political institutions. As illustrated in Figure 1, a reasonable consensus has developed in the literature that party systems are best understood by studying the interaction between political institutions and social cleavages. The focus of this paper is on the former, although it does not of course ignore the latter.
Figure 1. Independent Variables in the Party Systems Literature. Accordingly, I shall trace the development of the literature relating the political institutional independent variables to party systems. While much ink has been spilled in comparative politics investigating the relationship of political institutions to legislative party systems, much less focus has been given to executive party systems. I shall focus on both legislative and executive party systems here.
Cox (1997) argues that democratic politics can be thought of as a series of choices by which a government is chosen from the citizenry, where coordination problems obtain at each stage of the choice. The goal of actors is to coordinate so as to obtain their most preferred outcomes. At issue is how the political institutions structuring the choice process influence the selection of the various arms of government, such as the executive and the legislature: the electoral alliances that are encouraged in these contests and the number of competitors that are accordingly sustainable in equilibrium. In this manner, political institutions influence two groups of actors: voters, in terms of their propensity to vote strategically, and elites, in terms of their strategic decisions about entry (i.e., for candidates, whether or not to contest races; for elites more generally, which candidates to support).
Electoral systems have traditionally been the most important political institution examined in this respect. The most important components of an electoral system are the electoral formula and district magnitude (Taagepera and Shugart, 1989): proportionality increases with district magnitude and the permissiveness of the electoral formula. The more disproportional an electoral system, the fewer the effective number of parliamentary parties (Lijphart 1999, 165-8). Furthermore, as disproportionality of an electoral system increases, the incentives for strategic coordination (both voting and entry) increase, and vice versa. Duverger’s famous law about electoral district-level ‘party’ systems,2 as re-formulated by Cox (1997), is that many electoral systems will in equilibrium support an upper bound of M+1 candidates or lists in the district, where M is the district magnitude (or the number of candidates advancing to the second round in the case of top-M majority run-off).3 These equilibrium upper bounds result from strategic voting on one hand and strategic entry decisions on the other, in addition to purely mechanical effects.4 With respect to strategic voting, voters will abandon their preferred trailing candidate for a more viable if less preferred one in order to make their votes count. With respect to strategic entry, anticipation of strategic voting will deter candidates if clear expectations about the viability of candidacies exist, such as that provided by party labels; thus we get the Duvergerian argument that minor party candidates will be deterred (Cox 1997, 170-1).5 Note that the discussion here refers to the number of candidates, lists, or party labels sustainable at the district level. Note also that the upper bound requirement allows social cleavages to play an important role, as noted earlier: social cleavages initially determine the number of competitors, and only if this ‘natural’ number is greater than carrying capacity of the system will the electoral rules act as a constraint.6 Thus, at the district level, strategic voting and entry decisions in one-seat districts with plurality rule will tend to produce no more than two viable candidates; in multi-seat districts with proportional representation, they will tend to produce M+1 viable candidates.7 In sum, then, in a single electoral district for any given office, different electoral systems will promote greater or lesser degrees of strategic coordination and thus will support in equilibrium a varying upper bound on the number of candidates or parties. Electoral systems act as an intervening variable between the independent variable of social cleavages and the dependent variable of ‘party’ system (candidate- or party label- carrying capacity) in a district.
How, though, is competition at the district level linked to competition at the national one? Regression analyses of cross-national variation in legislative party systems have found that electoral laws do have consequences (Rae 1967; Taagepara and Shugart 1989; Lijphart 1990) and that there is a significant interaction effect between social heterogeneity and electoral structure (Ordeshook and Shvetsova 1994) at the national level. These studies lend empirical support to the general Duvergerian argument that both electoral systems and social structures must factor into our theories of national-level party systems. However, Duverger’s Law is incomplete as a theoretical generalization at the national level, despite its empirical merits: Cox argues that the Duvergerian logic by which district-level patterns of competition are aggregated into a national pattern is unclear (1997, 275). To complete the argument, he suggests that we consider the end goal of political competition: that is, the ability to govern, which requires parties to gain control of the executive, however it is constituted. The selection of the national executive, like the selection of district-level legislative candidates, poses citizens, legislators, and political elites with a coordination problem, which encourages the formation of cross-district strategic alliances. As argued above, the electoral system governing the executive choice procedure sets particular upper bounds on the number of viable candidates for the executive office. The key question is how the strategic alliances formed in this contest will affect cross-district legislative alliances—the legislative party system at a national level—and vice versa. Thus, in accounting for national-level party systems, we must ask about other kinds of political institutions than electoral systems. Specifically, we must consider where governing power rests in a polity: how the executive is constituted and the balance of power between the executive and the legislature. This variable of regime type interacts with the district-level relationship between social cleavages, electoral systems, and party systems discussed above to give rise to national party systems.
The paper will first examine the literature that addresses how the quest to govern impacts the national legislative party system before turning to the reverse relationship. Most studies that address this issue take as their baseline parliamentary regimes, where governing power (the executive) arises from the legislature. Controlling the executive involves, loosely speaking, controlling a majority of legislative seats, so executive and legislative party systems are effectively one and the same: the quest for the legislature is the quest for governing power. Accordingly, the focus here is on cross-national variance in electoral systems and social factors, which affects the incentives of parties to ally in legislative elections with the goal of winning enough seats to form a government.8 Other studies ask if presidential regimes, where the executive is elected separately from the legislature and governing power distributed between the two bodies in some manner, might produce a deviation from this baseline, all else being equal. Powell (1982) and Lijphart (1994) investigate the effect of presidential executives on legislative fractionalization and find them to discourage fractionalization. Both employ a simple dummy variable for presidential regimes in their analyses. Shugart and Carey (1992) agree that parliamentary and presidential regimes differ, but point out that two factors—the electoral rules by which the presidential executive is elected and the timing of elections— are crucial.9 Mainwaring and Shugart (1997) primarily study Latin American presidential regimes and concur that the electoral cycle and presidential electoral system have a significant effect on the effective number of legislative parties. Summarizing the findings of these studies, Lijphart argues that “presidential systems can have an indirect but strong effect on the effective number of parliamentary parties” (1999, 155). Cox 1997 offers a more integrated theoretical perspective on the matter. He suggests that the number of legislative parties at the national level is a joint product of legislative and executive electoral rules, which both interact with the social cleavage structure. He develops a model along these lines, testing his theoretical hypothesis that the national legislative party system depends interactively on the degree of integration of executive and legislative elections and the strength of the executive choice procedure (Cox 1997, 204).10 A scaled proximity variable is used to capture the degree of concurrence between legislative and executive elections; the effective number of presidential candidates serves as a variable reflecting the interaction between social cleavages and presidential electoral rules, given his argument that the presidential election in a polity indirectly (via the presidential party system) affects the legislative party system.11 Cox’s findings confirm earlier results that the effective number of legislative parties depends on the interaction of social heterogeneity and electoral permissiveness as well as regime type (in the form of the existence, nature, and timing of presidential elections).
Second, the literature examining the reverse relationship—the effect of the national legislative party system upon the executive party system—is much spottier. Shugart and Carey’s data show that non-concurrent elections of a majoritarian president and an assembly reduce the number of presidential candidates relative to concurrent ones, their argument that electoral cycles have little effect in this case notwithstanding (1992, 220 and 223). Non-concurrent elections of a plurality president and an assembly allow separate party systems to form, while ‘mutual contamination’ from concurrent elections leads to an increased number of presidential candidates (an imperfect two-party presidential system) relative to what we would expect in the absence of the assembly election (1992, 240). Cox does not really address this issue, arguing that while the legislative party system almost certainly impacts the presidential one, “the direction of influence is primarily from executive to legislative elections” (1997, 212). Making such an assumption facilitates the testing of his model. While Shugart and Carey present some data supporting the conclusion that the causal arrows do run in both directions, from legislative to executive as well as executive to legislative party systems, not much work has been done on this issue.
To foreshadow a later section of the paper, it is worth noting that the existing literature, whether empirical or theoretical, does not offer comprehensive hypotheses about the impact of regime type on national-level party systems. The impact of presidentialism, broadly defined, is tested by comparing polities with presidential executives to those without; similarly, more detailed investigations into the impact of presidentialism have focused on both the electoral system by which the executive is chosen and the electoral cycle. These studies do not really ask how the institutional structure of a state matters beyond the broad dichotomization of polities into parliamentary and presidential regimes. It is now well-acknowledged that there are great differences in the substantive functioning of polities broadly classed as presidential (Shugart and Carey 1992). Likewise, the notion that polities broadly classed as parliamentary function according to the same logic has increasingly come under attack (Tsebelis 1995).12 Shugart and Carey (1992) in fact propose a classification system embracing four ‘ideal types’ of regimes: pure presidentialism, pure parliamentarism, president-parliamentary, and premier-presidential. While some polities can be classed as one of these four ideal types, they argue, many cannot, being best thought of as hybrids combining features of the ideal types.13 Further, Shugart and Carey codify presidential powers in order to reveal the great variation in the functioning of even purely presidential regimes. Mainwaring and Shugart (1997) echo this point. On the parliamentary front, much ado has been made about the differences between parliamentary regimes that experience single-party majority government and those that tend towards coalition government, where the latter category can be further sub-divided by the types of coalition government often experienced (Strom 1990; Lijphart 1999).
In short, the comparative politics literature recognizes that making a dichotomous distinction between parliamentary and presidential regimes types is both theoretically and empirically untenable: even in the realm of ideal types, we need to think in a more nuanced fashion. Moreover, significant variations exist within ideal types. At the same time, however, comparative politics theories about the impact of regime type on national party systems have not sufficiently addressed these concerns.
Testing Existing Theories This paper agrees with earlier studies that the interaction between regime type, electoral system, and social structure of a polity gives rise to its national-level party system. It differs from prior studies in its hypothesis about how regime type matters. The following sections of the paper will demonstrate that existing theories cannot account for either cross-temporal variation in legislative party systems, controlling for social cleavages and the legislative electoral system, or for a significant amount of cross-national variation in executive party systems, controlling for the executive electoral system and social cleavages. Given the failure of existing theories, I will argue in this paper for a more nuanced theoretical treatment of regime type. My proposition, which will be developed in section four, is that regimes that diffuse executive power, all else being equal, seem to provide fewer incentives for strategic behavior among both voters and elites than those that concentrate executive power. The former types of regimes thus sustain higher equilibrium numbers of candidates or party labels than do the latter. However, this paper will first derive and test hypotheses drawn from the existing literature before developing an alternative approach.
With respect to cross-temporal variation, existing theories make predictions about the effect of the introduction or removal of a separately elected (presidential) executive, all else being equal, on the national legislative party system. It is worth re-iterating that existing theories of party systems can only classify regimes as either parliamentary or presidential, and thus model regime type primarily as a dichotomous variable.
Proposition 1: The addition of a separately elected executive, i.e. the shift from a parliamentary to a presidential regime, will have a consolidatory effect on the legislative party system provided that the legislative and executive elections are reasonably concurrent and the executive electoral system is reasonably restrictive relative to the legislative electoral system, all else being equal.
That is, given reasonably concurrent elections and a relatively restrictive executive electoral system, strategic coordination in the contest for the executive will promote greater strategic coordination in the legislative electoral contest than was promoted in the past by the parliamentary regime. This is Cox’s (1997) argument that the impact of a presidential regime on the legislative party system will be mediated via the presidential party system. Executive electoral systems are almost always restrictive due to their small magnitude (only one candidate can capture the office) and thus almost always promote a reasonable amount of strategic behavior, particularly when compared to the permissive legislative electoral systems that obtain in many polities with parliamentary regimes. The incentives for strategic coordination in the executive contest do vary, however, due to the differing electoral formulas that can be used to govern the executive choice procedure. Thus, existing theories also make predictions about the national executive party system that is likely to arise following the introduction of a directly elected executive (cross-temporal variation), as well as about how executive party systems across polities should vary with the executive electoral system employed, all else being equal (cross-national variation). The following proposition can accordingly be viewed as accounting for either cross-temporal or cross-national variation.
Proposition 2: Holding social cleavages constant, a directly elected executive selected using a permissive electoral system such a majority run-off electoral formula will have a higher carrying capacity and thus will sustain more candidates in equilibrium in the executive contest than less permissive ones such as a plurality electoral formula, and vice versa.
By extension, we can make an even more specific prediction.
Proposition 3: Holding social cleavages, the congruence of executive and legislative elections, and executive electoral systems constant, there will be no variance in cross-national executive party systems.
In other words, there should be no observed variation in the executive party systems of regimes with separately elected executives (presidential regimes) if social cleavages, the congruence of executive and legislative elections, and the executive electoral systems are all controlled for. Note that similar propositions to 2 and 3 could be drawn up with respect to national legislative party systems; however, this paper will only address cross-temporal variation in legislative party systems.
To test these propositions drawn from the existing literature on party systems, this paper will conduct a case study that it will then attempt to place in comparative perspective. Israel’s 1996-2001 experience with a new set of political institutions is in many respects a natural experiment with which to test the above hypotheses. Social cleavages and the electoral system are controlled for by the experiment, while the regime type is allowed to vary. The existing academic literature agrees for the most part with popular conceptions at the time of the reform that a presidential executive was added to existing Israeli institutions: that is, that the Israeli regime was changed from a parliamentary to a presidential one in 1996, with the implications for Israel’s legislative and executive party systems as outlined above. However, as I will shortly discuss, Israel’s political regime from 1996-2001 was in fact a unique institutional arrangement best characterized as a president-parliamentary regime according to Shugart and Carey’s criteria. Will existing theories nevertheless prove to be accurate in accounting for Israel’s experience, as they do not recognize such fine-grained distinctions between regime types? As suggested above, the Israeli case will instead demonstrate that existing theories fall short of an acceptable explanatory mark: that we must do more than dichotomize regimes into presidential and parliamentary types. Instead of the addition of a separately elected executive promoting consolidation in the legislative party system, legislative party fragmentation was observed; this legislative party system fragmentation occurred despite the executive contest being characterized by a perfect two-party system. Existing models such as Cox’s, as outlined above, would predict that strategic alliances formed for the ‘presidential’ contest would carry over to the legislative one, exerting a downward pressure on the carrying capacity of the legislative party system. Yet this did not happen in Israel. Other anomalies that lead me to reject the prior hypotheses will also be recounted.
It is worth briefly elaborating upon the experimental controls before continuing. First, along with the regime change, minute adjustments were made to Israel’s electoral system: the electoral threshold for representation in the Knesset was raised by 0.5% and major parties began selecting their candidates via party primaries. However, it seems acceptable to view the legislative electoral system as effectively constant across the two regimes due to the minor nature of the change in the threshold. At any rate, the data do not show the electoral threshold having the predicted effect of consolidating the legislative party system. Further, I do not believe that the party primaries can alone account for the changes in Israel’s legislative party system, nor for the surprising development of its executive party system, although I do not provide evidence for this claim in the paper; future work should re-visit this issue. Second, of course Israeli society changed in the six-year period at hand; nevertheless, it seems that six years is too short a period for radical social re-alignments to take place, at least to the extent that social changes alone could account for the developments in the party systems. Israel’s next few elections will provide a natural test of the social versus political change hypotheses: with subsequent elections being held under its original pre-1996 political institutions, if social change drove the party system developments from 1996-2001 to be recounted below, then the same patterns should persist despite the change in political institutions.
This paper will accordingly compare Israel’s legislative party system under two different political regimes: the parliamentary regime that it utilized for most of its history and the novel regime type that it adopted from 1996-2001. It will then consider the developments in the executive party system under the new regime, for which there is no earlier correlate. Finally, the paper will attempt to place the developments in Israel in comparative perspective in order to further test the hypotheses drawn from the existing literature regarding the link between regime types and party systems. To do so, it will briefly compare the French, Israeli, and Peruvian national-level executive party systems.
The Case of Israel Pre-Reform Israel: 1949-1992 Israel has long been a unitary and highly centralized state, and its electoral system one of the purest and most democratic. It is also one of the few democracies without a written constitution, the United Kingdom being the other prominent example. This section of the paper will discuss Israel’s pre-1992 political institutions and party system.
The democratic institutions structuring the Israeli polity were given legal status by a multi-document constitution consisting of individual Basic Laws that were formulated and adopted over time by the Knesset, a compromise that resolved disagreements about the fundamental nature of the Israeli state (Hazan 1997, 329-30).14 Israel was established as a parliamentary system, where the executive (the prime minister and cabinet) were drawn from and responsible to the 120-member legislature. The Basic Law: The Knesset established two of the four key features of the electoral system, a single nation-wide district and an extremely proportional PR electoral formula for the parliamentary elections.15 Section 83 of the Knesset Election Law of 1969 established a third key feature, the closed list system (Bogdanor 1993, 84). Fourth and finally, the threshold for representation in the legislature was set at an extremely low one percent of the national vote. The Knesset elected Israel’s president, its head of state.
These institutional features ensured scholarly interest in Israel, as its electoral system was one of the most proportional in the world. For example, Lijphart found its average disproportionality (the deviation between vote and seat shares) from 1945 to 1980 to be 1.1%, and from 1945-96 to be 2.27%.16 For the latter period, only The Netherlands, Demnark, and Sweden had lower disproportionality scores, with 1.30% (The Netherlands) being the lowest. By way of comparison, the mean disproportionality of all 36 democracies investigated by Lijphart was 8.26% (1993, 115; 1999, 162-3). Furthermore, only The Netherlands and Israel used a nation-wide constituency for their elections. Note that the existence of a single legislative electoral district for the polity as a whole obviates the need to link the district-level party system with the national-level one in the simple sense that the district and polity are one and the same. However, actors still face coordination problems in the form of gaining control of the executive, particularly given the large district magnitude; coordination merely takes place within the district instead of both within and across districts.
Partially due to this permissive electoral system and partially due to its many social cleavages, Israel had a multi-party system characterized by centrifugal competition during the period of 1949-92. Israel has never had a majority party, although from 1949-77, Mapai (the forerunner of Labor) controlled the pivotal point in the party system; many academics consider it to have been a dominant party during this period, as it (from 1949 to 1965) and its ‘Alignment’ of smaller, ideologically sympathetic parties (from 1965 to 1977) obtained between 37.5% and 46.7% of the Knesset seats (Diskin and Diskin 1995, 34-5). In 1977, a coalition of several right-wing groups led by Herut emerged as Israel’s largest political party, the Likud, and dealt the Alignment a stunning defeat. From 1977 to the early 1990s, the Israeli party system polarized into competition between the left (Labor plus smaller Jewish and Arab left-wing parties) and right (Likud plus smaller right-wing parties), with small religious parties holding the pivotal center position. These parties enjoyed a greater affinity with Likud and thus tended to back the right after 1977, but were willing to enter into coalitions with Labor (Ibid., 36-9). Over the entire period, the effective number of political parties in the Israeli polity was 4.42 (Ibid., 34),17 placing Israel in the same party system class as countries such as the Netherlands, Denmark, Belgium, Italy, and Finland (Lijphart 1999, 76).
The 1992 Reforms Institutional reform has been a popular topic of discussion in Israel since the state’s founding. Public dissatisfaction increased greatly in the late 1980s and early 1990s as a result of the “negotiating excesses that occurred in forming and maintaining the government coalition in the 12th Knesset” as well as during the March 1990 coalition crisis (Doron and Kay 1995, 299-300). Specifically, voters were unhappy due to the perceived lack of government accountability, stalemated policy initiatives, and disproportionate power exercised by the small religious parties. The 1992 reforms were supposed to address these issues. It is worth noting that while public justifications for the reform proposals largely relied upon these concerns about the political system, such concerns fall far short of explaining why reforms took the specific form they did. But the search for an explanation of Israel’s institutional changes must be taken up at a later time and by another paper; this section of the paper describes the reforms made to the Israeli electoral system and regime.
During the final days of the 12th Knesset (1988-1992), several institutional changes were made. First, the most significant of these was the March 1992 amendment of the Basic Law: The Government, the result of which was that Israel became the first country to directly elect its prime minister in the 1996 elections for the 14th Knesset. Israeli politics were conducted under this political regime from 1996 until 2001, when the Knesset voted to restore the one-vote parliamentary system of government that had operated prior to the 1996 elections.18 Note that although the regime change did not go into effect until 1996, some scholars argue that it nevertheless impacted the 1992 elections, as politicians began to anticipate the forthcoming direct election of the prime minister (Hazan 1997, 331; Shamir and Arian 1995, 10). Due to both theoretical and empirical difficulties with this argument, I will only focus on the 1996 and 1999 legislative elections under the new regime.19 Second, an electoral reform that increased the national threshold for representation from 1% to 1.5% of the valid votes cast was voted upon at the same time; it went into immediate effect, in time for the 1992 elections to the 13th Knesset. Third, changes in the candidate selection procedures of the major parties—the introduction of party primaries—were also made in 1992. I will first focus upon the electoral system reforms and then discuss the change in regime type.
Israel’s main electoral reform was relatively straightforward. Increasing the national threshold for representation from 1% to 1.5% of the valid votes cast made the Israeli electoral system less proportional, although clearly not by much. Parties are now faced with a greater hurdle to overcome in order to gain representation in the Knesset. As mentioned earlier, this straightforward and relatively minor reform is not the focus of this paper. Such a marginal adjustment in the electoral threshold cannot be considered a significant change to the electoral system. Nor will this paper focus on the effects of the introduction of party primaries by both Labor and Likud; interested readers should see Rahat and Sher-Hadar 1999 for a discussion.
The main focus of this paper is the change in regime. This reform—the addition of a separately elected executive, analogous to a president despite still technically being termed a prime minister—was designed to complement the electoral reform in weakening the power of small parties, producing more stable governments, and ultimately leading to a less consensual policy process.20 Opinion amongst academic commentators on Israeli politics, however, was divided as to whether this or the opposite effect—the strengthening of small parties— would occur.21 Arian and Shamir (1999) summarize the changes made by the constitutional reform. Amendments to the Basic Law stipulated that the prime minister was to be directly selected by the electorate using a majoritarian double-ballot electoral system, similar to that used in France. That is, a second round run-off between the two candidates with the highest number of votes would take place if no candidate received a majority in the first round. The prime minister alone was given the power to form and to head the cabinet. Both the prime minister and the Knesset were given powers of dissolution, though: by a special vote of 80 members, the Knesset could dismiss the prime minister and force a new election for the prime ministership; likewise, the prime minister could dissolve the Knesset with the president’s approval, forcing new elections for both the assembly and the prime ministership; finally, the Knesset could bring about double elections either by a vote of no confidence or by failing to pass the national budget, both of which required a majority vote in favor. Thus, prime ministerial and legislative elections sometimes but not always coincided. The Knesset continued to be elected using a closed list system of proportional representation in one national constituency, although the threshold for representation increased to 1.5% as previously discussed.
At the most general level, the addition of a directly and separately elected executive moved Israel away from a parliamentary regime to one of a unique format that was neither presidential nor semi-presidential.22 Shugart and Carey argue that the defining characteristic of presidentialism is the separate sources of origin (separate popular elections) and survival (separate fixed terms) of the legislature and executive (1992, 18-22). Conversely, parliamentarism is a fused power system, where the executive originates in the legislature and survives for only as long as it maintains the legislature’s confidence. Israel’s 1996-2001 regime does not clearly fit into either category: the direct election of the prime minister gave the executive branch of government a separate source of origin from the legislative branch, but the survival of the executive branch still depended upon legislative confidence, and vice versa (either branch could dissolve the other). Nor is Israel’s now-defunct regime semi-presidential, as is the French 5th Republic, where a directly elected president coexists with a government headed by a premier reliant upon parliamentary confidence; this kind of regime is usually described as alternating between parliamentary and presidential phases, depending upon whether or not the legislative majority supports the president (Duverger 1980).23 Obviously, if we amalgamate Israel’s prime minister to the French president, we are left without the second executive of semi-presidentialism who heads a government responsible to the legislature. So what type of regime did operate in Israel from 1996-2001?
Shugart and Carey discuss a fourth hybrid regime, which they label “president-parliamentary” (1992, 24-7). Again amalgamating the prime minister to the ‘president,’ here both the president and the parliament have authority over the composition of cabinets. That is, the president both appoints and dismisses cabinet ministers, who are subject to parliamentary confidence. This gives both branches equal authority to dismiss members of the cabinet, unlike the other regime types where this authority is asymmetrically distributed. There is also a lack of independent survival of legislative and executive powers. This seems to approximate the Israeli regime of 1996-2001. Lijphart (1999, 199-224) has more difficulty dealing with it, describing it as “uncharted territory.” He ends up characterizing the regime as presidential, however, unlike Shugart and Carey—an odd classificatory decision. By all of his own criteria, it belongs in the class labeled “Hybrid VI,” which he argues has no real-world examples; the Hybrid VI executive is selected by voters and the one-person executive is dependent on legislative confidence, a reasonable description of the defunct Israeli system. Sartori (1997) subsumes both Israel and France within one broad semi-presidential category. Obviously, we are in a classificatory conundrum with respect to 1996-2001 Israel. I shall adopt Shugart and Carey’s classification, as it reflects my belief that the post-1996 Israeli system is a unique one—neither presidential, parliamentary, nor semi-presidential.24 Existing theories of party systems, however, focus solely upon the presence or absence of a separately elected executive (a president) and must accordingly classify the Israeli polity during this period as a presidential regime.
Institutional Reform: Results Now we arrive at the key question of the paper: what were the effects of these reforms on the Israeli party systems? Specifically, what was the impact of adding a separately elected executive and changing the regime from a parliamentary to a president-parliamentary type? First of all, it is worth noting that there are some difficulties in empirically separating out the impact of the 1996 regime change on 1996-2001 elections from the impact of the earlier 1992 reforms (Labor party primaries and the increase in the threshold), as well as from the impact of concurrent further changes in the candidate selection process (Likud party primaries). Likewise, while we can assess the 1992 threshold and candidate selection changes independently of the 1996 regime change by studying the 1992 elections, we will not be able to completely disentangle the two 1992 reforms from each other. Nevertheless, as I argued earlier, I do not think that the changes in the electoral threshold and candidate selection procedures within parties together constitute a significant change in the electoral system during the period at hand.
This section of the paper will begin with an analysis of the reforms’ effects on the legislative party system in Israel before turning to the development of a party system around the contest for the executive. First, the legislative party system. Following Lijphart (1999), I characterize the legislative party system by the effective number of parliamentary parties, an index originally developed by Laakso and Taagepera (1979). This index is calculated as follows:
Np = 1
( si 2 ),
where si is the proportion of seats of the i-th party. As Lijphart points out, this index carries the same information as Rae and Taylor’s index of fragmentation, F (1999, p. 68). Throughout, I count an electoral list as a party. That is, if several parties ran together as an official list in a particular election, I counted them as one party. While the fractionalization and volatile nature of the Israeli party system make breaking lists down into their component parts a difficult process, one opening the analyst up to charges of inconsistencies, simply counting lists is consistent and reflects the common conviction in the literature that closely allied parties can often be counted as one in characterizing a party system (1999, p. 69-71).25 This is particularly the case in Israel, where certain parties running as lists (e.g. the Alignment of the Labor Party and Mapam from 1969-84) can be considered one party by all of Lijphart’s criteria. If anything, then, my estimates of the effective number of parliamentary parties are underestimates, particularly with respect to the 1996 and 1999 elections.26 These estimates are shown in Table 1.