Strategic Political Competition in a Comparative ...

5 downloads 0 Views 833KB Size Report
I study these theoretical claims using a case study taken from the Israeli context illustrating a ... Israeli withdrawal from the Gaza strip in 2005. I then present a ...

Policy Instability in a Comparative Perspective: The Context of Heresthetic This paper studies the factors that affect the likelihood of radical policy changes. Radical policy changes are defined here as acts in which political leaders choose a policy which is contradictory to policies that they have committed to during elections and formed coalitions on their basis. The reason for such changes is that quandaries change the policy space that shapes elections and initial coalitions. This change creates incentives for political players to re-bargain over existing coalitions. As the bargaining process decreases the leaders’ survival prospects, leaders’ likelihood to opt for the ‘wrong side’ of the policy space will increase. The claim is validated using a case study and a large-N quantitative analysis of fifty-five cases from thirty-seven countries.

Keywords: Heresthetic,

Policy

Instability,

Political

Disengagement Plan, Most Different Systems

Quandaries,

Political

Competition,

Introduction Why do elected politicians in consolidated democracies opt for radical policy changes, moving from a policy they have previously committed to and form coalitions on the basis of a policy they previously negated? Such policy changes happen in diverse issue contexts and are taken by leaders in qualitatively different settings. These include established democracies such as the UK, where the British political system has seen a series of radical policy changes that shaped it as we know it today (McLean 2001). It also includes South-American countries where pro-welfare campaigning presidential candidates opted for pro-market policies once they became presidents (Stokes 2001). Another setting that saw such changes was the one of contested democracies such as Israel with its entangled and contradictory path towards conflict resolution with the Palestinians (Rosenthal and Doron 2009). Thus, policy instability purposely contrived by leaders is an essential aspect of the functioning of democracies (Schumpeter, 1952, pp. 269-273; Riker 1982), which should be studied systematically (Shepsle, 2006). Some of those who have studied this topic, use the set of concepts proposed by the heresthetic research program. The heresthetic research program deals with the study of the practice of political strategy in competitive political settings (e.g. Riker 1986; 1996; Weingast 1998; McLean 2001; Doron and Sened, 2001; Epstein and Shvetsova, 2002; Schofield 2006; Ferrie, Dupret and Legrand 2008; Dardanelli 2009; Rosenthal and Doron 2009). Theoretically, the expectation in the heresthetic research program is that leaders will make a radical policy change if they are locked in a corner of a perceived political loss (McLean 2001). The likelihood of radical policy changes manifested in the making and breaking of political coalitions (Doron and Sened, 2001) will increase when political quandaries are set on the political agenda and de-stable the existing policy space (Schofield 2006). Such quandaries yield political situations in which there is not one clear social belief pointing at what needs to be done (ibid). When quandaries emerge, the existing policy space loses its relevance and the contest for a new policy starts (ibid, pp. 67-68). Times of uncertainty yield political competition over the goods, which the government allocates (Doron & Sened, 2001, pp. 3-7). In political institutions this means bargaining over the participation in policy coalitions where all sides asses their ability to receive 2

payoffs (policy and office) from taking part in a policy coalition (ibid, 7-12). In such cases we may see competition from within the previous coalition or even within the ruling party by players who believe their threat power over the existing coalition framer increases. As the ability of the coalition's leader1 to bargain with the rest of the coalition's components decreases, her incentive to look for another coalition increases. As that threat on the framer’s survival becomes higher, the further the framer might go in terms of the changing her policy positions. Thus, the coalition framer might form a new coalition based on a new policy, radically different, from the one she campaigned on and formed after the elections. Hence, the combination of quandary and political competition for leadership positions increases the likelihood for radical policy changes. I study these theoretical claims using a case study taken from the Israeli context illustrating a similar process that happened when Ariel Sharon initiated a unilateral Israeli withdrawal from the Gaza strip in 2005. I then present a quantitative analysis supporting these claims using 55 cases from 37 countries happening during the years 1990-2006. Both these analyses validate the significant effect of quandaries and bargaining for leadership positions on radical policy changes. The next section discusses policy instability situating it in the more general topic of political instability to emphasize its uniqueness. On that basis and using concepts stemming from the heresthetic research program, I offer a theoretical reasoning for the process that yields a radical policy change. Then I illustrate this reasoning using Israel’s decision to unilaterally evacuate the Gaza Strip. The third section introduces the research design for the large N analysis. The fourth section presents the findings. The subsequent section discusses the findings and sets guidelines for proposed future research. Policy Instability in Democracies: Context and Theoretical Guidelines In order to define policy instability I will first define its inverse: policy stability. Political leaders form political institutions based on their ability to attain gains from these institutions. Such gains are based on the willingness of a sufficient number of people to cooperate with these leaders and comply to the rules and regulations set by them (Sened, 1997; Doron & Sened, 2001, pp. 55-75). Policy stability in democracies is based on the ability of an elected leadership to form and maintain a coalition that supports the 1

I use the terms coalition’s leader/premier and framer interchangeably.

3

leadership’s desired policies. As the leadership needs to bargain more so as to sustain its desired policies it will need to give in some of its policies or pay in portfolios and budgets to maintain the support over its policies. Thus, policy instability would be the deviations in policy positions and the division of goods needed to sustain a coalition working for the implementation of a policy (ibid). Policy instability is an aspect of political instability that has some interactions with other aspects of political instability, namely socio-structural instability and cabinet instability. Yet, policy instability can be differentiated from these other facets of political instability. Let us discern it from other elements of political instability. For political elites aiming to maintain a coalition that rules institutions, there has to be a sufficient number of individuals and groups who will contribute to the institutions and not go against them (Doron & Sened, ibid). This is the basis for social legitimacy to ruling institutions (ibid). A safe assumption is that such compliance correlates with macro-structural elements that characterize societies and influence group and individual behaviors towards the government and its policies (Przeworski et. al., 1996; McLean, 2001, pp.17-28; Schofield, 2006, pp.17-21). Policies unaccepted by the social beliefs’ system as beneficial to a sufficient number of people and groups will have hard times in creating a governing coalition that will promote them (Schofield, ibid). Another aspect of political instability is cabinet instability or the ability of an elected coalition to maintain its position in the country’s driver seat, regardless of the policies it takes and the social legitimacy it receives (Damgaard, 2008). Some have shown that the ability of cabinets to survive in office depends on institutional arrangements concerning cabinet termination (Persson & Tabellini, 2004; Schleiter & Morgan-Jones, 2009). Others relate to the discretion of premiers who choose to dissolve cabinets due to electoral considerations (Lupia & Strom, 1995), or to dissolution due to technical reasons rather than specifically institutional or based on discretion (Damgaard, ibid). Figure 1 illustrates the multifaceted perspective of political instability. Figure 1 about here Thus, it might be reasonable to assume that elite in power in a socially legitimized institution and a politically supported cabinet will hold on to its policy positions since both guarantee its hold in power. In that sense, policy instability reads 4

social instability and cabinet instability. Without one of those the likelihood for policy instability will decrease. However, there can be times of social instability, which will not yield policy instability. For instance, as was the case in France in 2005 of riots in immigrants’ neighborhoods against government immigration policies. Despite these riots, the government which its policy did not favor immigration and the assimilation of immigrants did not respond to these demonstrations by leading a policy change (Koff & Duprez, 2009). Moreover, assume that a policy led by a small coalition of players that have a variety of options to form a minimum winning coalition. Even if the bond they create with other parties/parliament members so as to form a cabinet breaks, they can still maintain their policy either with another party or even due to the lack of ability of a coalition of parties to unite against their policy (Schofield & Laver, 1990). Thus, cabinet instability is also not a guarantee for policy instability: new elections might be called and the cabinet might be formally dispersed but will continue to rule as an interim cabinet and then will be re-elected and continue its policy (Damgaard, ibid). Hence, policy instability is a topic differentiated from social and cabinet instability and should be studied as such. One of the research programs dealing with policy instability and takes into account these characteristics of the policy process in democracies is the heresthetic research program. The focus of this research program is on policy as an outcome of strategic interaction between purposeful players. Heresthetic (or heresthetics) is a concept coined by the late William H. Riker from a Greek word that means both being able to choose and heresy (Schofield, 2001). This concept relates to the strategic aspects of political life: making a strategic choice of de-stabilizing existing policy coalitions or maintaining them through agenda setting, strategic voting and adding dimensions to existing deliberation processes (Riker, 1986; Schofield, 2006; McLean, 2009) Two of the most salient critical evaluations of this research program are Green and Shapiro’s (1994) epistemological and conceptual analysis of rational-choice theory literature, and Mackie’s (2003) normative, theoretical and empirical examination of the heresthetic research program. Green and Shapiro’s claims relate mostly to the lack of realism in rational choice analysis and its operational conceptualization. Moreover, they relate to the lack (at the time) of empirical analyses of rational choice claims (Green and 5

Shapiro, ibid). This paper does not dispute the need for further empirical analysis of rational choice concepts (including heresthetic) as well as further efforts on conceptualization and operationalization to enhance this research program’s realism. In fact this is exactly what it tries to do. Mackie’s motivation for criticizing this literature theoretically and empirically is normative (McLean, 2009). It mainly contests Riker’s conclusions from social choice theory as overreaching, inaccurate and suffering from severe internal validity problems (Mackie, 2003, pp. 27-43). Riker’s conclusions were that democracy cannot produce any decision outcome that can seriously claim to represent the will of the majority. Rather, any decision which is set on the agenda can be accepted by a free collective. Hence, such decisions are either an outcome of arbitrary decision or a manipulation of the policy process (Riker, 1982, pp. 233-253). More recent applications of social choice theory to Political Science, show that even if in complex decision making environments where the ability to reach collective rational decisions is problematic (Austen-Smith & Banks, 2001, pp. 155-184), there can be individual preferences profiles and voting mechanisms that can produce valid collective decisions made by purposeful individuals (ibid, pp. 119-153). Hence, the inference that ‘anything can happen’ (Riker, 1980) can be relevant to a subset of collective decisions and not all such decisions (Austen-Smith & Banks, 2001, pp. 184; McLean, 2002). Thus, even for proponents of the heresthetic research program, Riker's inference is an overestimation of what can actually be deduced from this research program (McLean 2009). Hence, Mackie's normative point is redundant in terms of current affairs in the heresthetic research program (see also: Shapiro, 2011, p. 267). I agree with McLean (2009) that the importance in Mackie's work lies in locating flaws in some of Riker’s empirical narratives (Mackie, ibid, pp. 197-334). Yet, despite these flaws, the heresthetic research program is able to continuously check and update the internal validity of its claims (e.g Austen-Smith and Banks, 2001; 2005). Furthermore, it can develop its theoretical understanding of reality (e.g. Doron and Sened, 2001; Schofield, 2006). Finally, it can provide the basis for an informed, careful and reliable analysis of political decisions (e.g. McLean, 2001). Hence, the heresthetic research program continues developing its ability to supply valid and reliable tools for 6

political scientists to understand democracies. This paper is a modest effort in that direction. So as to understand a hypothetic process of policy instability yielding a radical policy change which follows the guidelines of the heresthetic research program let us define what are radical policy changes. Conceptually radical policy changes relate to situations where players move to the 'wrong side' of the policy space (McLean, 2001, p. 28). These moves are taken in contradiction to the policy commitments (usually) made in elections and are the basis of the post-electoral coalition. Mclean (2002) claims that this move "turns the table" over an existing situation, due to a perceived political loss to a player locked in an inferior political position (Shepsle, 2003). I will try to offer a hypothetical process that might yield such policy moves. The starting point of a process that yields a maneuver to the wrong side is a postelectoral coalition based on agents voted into power positions. These players are able to form a coalition to promote a policy they committed during the elections. The clear condition for coalition formation is that it is better in terms of policies and portfolio allocations than any other alternative coalition (Sened, 1995). The common assumption is that these players have a basis for compromises that might yield a low and mitigated level of instability and disagreements in governmental policymaking. Such a coalition will be formed around the uncovered set of coalition options: some friction and instability will always be between the coalition members. Yet, that level of instability will be tolerable and lower than the level of instability, which will happen in any alternative coalition these players can join (Austin-Smith and Banks, 2005, pp.134-135, 180-184). Assume that in enters the policy space (exogenously or endogenously) a new political divide accompanied by a controversy that pulls the rug under the feet of the ruling coalition. Thus, it renders the coalition policy lines irrelevant. Hence, new proposals need to be set on the table that might not coincide with the choice process that yielded the ruling coalition. Consequently, the coalition’s policy positions and probably its leadership positions are now re-open for grabs. This gives policy entrepreneurs new opportunities to gain power by raising a new coalition which its polices best fit the new situation (Schofield, 2006, pp. 11-16). Such situations include times of social quandaries 7

as a basis for radical policy changes. The main examples for such situations are the slavery debate in antebellum US (Weingast 1998); the controversies in UK politics from the 19th century onwards regarding Ireland, British class cleavages and economic crises (McLean 2001); the creation and maintenance of US independence in late 18th century history; as well as 20th century Keynesianism (Schofield 2006). Others relate to the post 9/11 legislation in the UK (Ferrie Dupret & Legrand 2008), or Rabin's decision to make peace with the PLO taking place in the context of deep changes in Israeli society and the Middle East balance of powers (Rosenthal & Doron 2009). Yet, why should quandaries yield radical policy changes? When the socio-political environment becomes uncertain (i.e. a situation of quandary) it also becomes more competitive (Doron & Sened, 2001, pp. 4-6). In competitive political environment players face a viable threat on their political survival (ibid). In such situations players look for potential political partners that can credibly commit to a cooperation that will sustain their survival in power. So as to form cooperations politicians are willing to give in budgets and resources they wish to allocate to their most desired policies. As the potential partners’ threat power over such coalitions increases those who wish to form it will be willing to pay a higher price in policy compromises and government responsibilities transferred to their potential allies (Baron & Ferejohn, 1989; Sened, 1995; 1996; Banks & Duggan, 2000). In essence, any such compromise is a policy change: politicians commit to a policy before the elections and are willing to move to another policy to maintain their access to resources. In political bargaining, as the threat of losing power is higher so the concessions in policy issues and/or budgets will increase (Sened, ibid). Thus, if the only way to stay in power or receive some latitude for policy making, is by making a radical policy change, then strategic coalition leaders will take that step. Note the combination of two crucial factors: a political quandary stemming from social and economic factors alongside the extent of the competition with its consequent risk for the leadership’s power position. Figure 2 about here This chain of events starts with a given coalition, which is the outcome of the post-electoral bargaining process. On the next phase, a change in the existing policy 8

space happens and a new bargaining process begins based on new dimensions and consequent policy options. This bargaining process threatens the position of the coalition’s leadership as it might yield a new coalition that they will not be a part off. As the multifaceted character of political instability implies, the two other aspects of political stability may mediate this process’ outcomes: socio-structural stability and cabinet stability. The last link in this ’chain’ of events would be an increased likelihood for the emergence of a radical policy change taken by the coalition’s leadership, opting for a set of policy positions previously endorsed by the rulers’ opposition. In order to make these claims more tangible I illustrate them using Israel’s former Prime Minister Ariel Sharon’s surprising decision taken in 2003, deliberated in 2004 and implemented in 2005 to evacuate the Israeli civilian settlements and military camps from the Gaza strip. An Illustration: Sharon’s 2003-2005 Disengagement Decision Israel is a multiparty parliamentary democracy where the whole country serves as one electoral district. Seats in parliament (Knesset) are allocated on the basis of proportional representation with a 2% electoral threshold to the parliament. Besides the years 19962001 elections in Israel are party based where the voters choose between closed party lists determined in a variety of intra-party selection methods. Following the elections Israel’s president (a symbolic duty) nominates the head of the faction with the highest prospects to form a coalition. In order to receive that nomination which will them to the prime-minister’s position they need to show that they have enough seats in parliament to sustain a majority. They usually need to show that they have enough parties supporting them to raise at least a minimum winning coalition (Rosenthal, 2011). On December 2003 the Israeli Prime Minister Ariel Sharon declared that he made a decision to initiate a unilateral Israeli withdrawal from the Gaza strip. This meant an evacuation of all civilian Jewish settlements in which about 7,500 Israeli citizens lived, alongside the Israeli military posts located within the Gaza strip. The surprise about that decision was that on January 2003, less than a year before Sharon made that declaration, he led his Likud (unity) party to power campaigning on the rejection of the proposal to unilaterally evacuate the Gaza strip. The 2003 elections took place within the final days of the second Palestinian Intifada, and the international 9

community was looking for ways to end it through negotiation and compromise between Israelis and Palestinians. This was a main issue in that time and the electoral campaign mostly focused on it (Rynhold and Steinberg 2004). Hence, Sharon’s campaign pledged to NOT evacuating the Gaza strip’s civilian settlements and military posts, was one of the reasons for his victory. The main challenger for Sharon’s position as a potential post-electoral coalition framer was the left-wing Labor party led by Amram Mitzna. Mitzna’s campaign committed to initiate an Israeli unilateral withdrawal from the Gaza strip. On the basis of the policy positions presented during the elections the Likud gained 38 seats in the 120 seats unicameral Knesset and Mitzna and the labor party received 19 seats2. In order to understand Sharon’s radical policy change I turn to estimate the policy positions of parties and their weights in the 16th Knesset following the 2003 elections. Figure 3 about here In order to estimate the policy positions taken by the political parties in the 2003 elections and their consequent weight in the parliament I used the 2003 elections survey data supplied by the Israel National Elections project3. Following Schofield and Sened (2006: 70-100), I chose two lists of questions from that survey so as to estimate the latitude of voters’ policy positions as a basis for the estimation of the parties’ policy positions. These policy positions are estimated with respect to Israel’s most dominant national politics’ policy dimensions: the Arab-Israeli conflict and the separation of state 2

Results can be viewed at the Kneeset’s website:

http://www.knesset.gov.il/description/eng/eng_mimshal_res16.htm preferably using Internet Explorer. 3

This study is a part of a series of electoral studies on the elections in Israel handled by Asher Araian

(deceased on 2010) and Michal Shamir. The descripton for the method of this study as presented at the survey’s website is: The study was conducted January 12-24, 2003, prior to the elections for the 16th Knesset, held on January 28 2003. Mahshov Research Institute carried out the fieldwork using Hebrew, Russian and Arabic telephone interviews. A stratified sample of Jews and Arabs by geographical areas, and random sampling within each strata. Special attention was given to the ultra-orthodox and immigrants from the former Soviet Union, so that they be proportionally represented in the sample, if needed by quota methods. N=1,234 (1,083 Jews and 151 Arabs). The sample was divided between two waves, where each wave consists of an independent representative sample of the electorate. The first wave was interviewed between January 12 and 17 2003 (N=618), and the second wave, between January 19 and 24 (N=616). The series of the INES studies can be viewed at: http://www.ines.tau.ac.il/2003.html.

10

from religion (Arian and Shamir, 2008). Using factor analysis two orthogonal components were created separately to serve as the basis for estimating the latitude of policy positions with respect to the Arab-Israeli conflict (the x-axis) and state and religion (the y-axis). The state and religion dimension is ordered so that its negative values represent policy positions that reject the separation between religious doctrines and political institutions and vice versa. The horizontal axis (x) is ordered in a way that the left side of the axis represents policy positions towards the conflict supporting Israeli appeasement policies towards the Arabs. As we move towards the right on that dimension we find positions that prefer a power driven Israeli approach towards the Arabs. The policy positions mapping divides the Israeli political space into four subspaces. If the upper end of the y-axis points at the ‘north’ of this policy map, the northwest quarter relates to being both pro-appeasement and a supporter for some level of separation of Jewish religion and state institutions. The southwest quarter represents supporters of appeasement who are reluctant to support the separation of religion and state. The southeast corner relates to positions that both reject the separation of state and religion and support a power-based approach towards the Arabs. The northeast quarter relates to policy positions that support the separation of state and religion and a powerdriven approach towards the Arabs. Parties’ overall characteristics are perceived by the voters with respect to policy and the parties’ valence. Voters are also affected by their own social backgrounds in perceiving parties both in terms of the parties’ positions and their valence (Schofield and Sened, ibid, pp. 40-55). Thus, the policy positions of voters who expressed support in one of the parties and the parties’ positions were perceived here as equivalent. This yielded the mapping of the parties’ positions as shown in figure 3. The parties not included in the survey (Am Ehad, Yahadut Hatora and the Arab parties: Balad, Raam and Hadash) were estimated along these policy lines on the basis of the policy positions noted at their manifestos4. In order to estimate the positions of parties and their abilities

4

Retrievable at http://www.knesset.gov.il/elections16/eng/lists/menu.asp (watched at December 21st 2011).

11

to form a coalition I add in their weights, or the parties’ seats’ share in Knesset (Schofield & Sened, ibid: 55-61). The policy positions taken by the Likud in these elections are located somewhat below the median position of the state and religion policy dimension and in the center of the right wing sub-space of the Arab-Israeli conflict dimension. No coalition can be formed by this side of the political map thatwill not include the Likud. Let us prove it. The parties located at the lower right wing-religious (southeast) corner of the policy space are: Likud (38) +Israel Baalia (2)+ National Unity (7)+ NRP(6)+ Shas(11)+ Yahadut (5)= 69 Without the Likud their number decreases to 69-38=31 where a quota of 61 MKs is needed for a minimum winning coalitiona. Even if the centrist Shinui (15 seats) was to join in a right-wing no-Likud coalition it would still be 31+15=46 seats which is way below the threshold. Was there any viable left-center alternative coalition? Labor (19) +Meretz (6) + Am Ehad(3)+ Shinui (15)= 43 seats. Clearly not enough. Was there an alternative that included the religious parties? 43 (left-center) + Shas (11) + Yahadut (5) = 59 Again not enough for the threshold. The only viable alternative to a Likud led right wing coalition could have been between the high valence parties: Likud (38) + Labor (19) + Shinui (15) = 72 seats. Yet, a formation of such a coalition had to be based on a policy change. This is due to the fact that at the time these parties’ policy positions on the most vibrant policy topic in Israeli politics (Arab-Israeli conflict) could not yield a stable coalition. Other hypothetical coalitions could include cooperation between the left+ religious parties+ the far right parties or left+ religious parties+ Arab parties. For the farright parties this was a non-viable coalition at the time (Doron, 2005). As the coalition with the Arab parties this was also not viable in the Israeli political sphere where the leaders of both sides could not come with a reasonable explanation to their voters at election time regarding such cooperation (Koren, 2010). Hence, with accordance to Schofield and Sened (2006: 56-61), based on its policy positions and the amount of seats the Likud in 2003 can be defined as a Structure Stable Core party: centrist enough and large enough to take part in any coalition and no coalition can be formed without it (ibid).

12

By March 2003, as the second Intifada was essentially over, the international community- backed by the Bush administration- was pushing for new peace proposals. Prime-Minister Ariel Sharon, the farmer, liked to use the corales image when he discussed the way the Israeli-Palestinian peace process was handled by the international community. In Israeli farmers’ jargon corales is a path of curved fences in which bulls destined for slaughter are sent into. In that path they move from one phase of the path to the other eventually reaching a point in which they cannot avoid the slaughter. Sharon believed that accepting the plans set on the agenda by the international community would channel Israel to a similar position (Eldar, Haaretz, 5/23/2011). Thus, Sharon decided to initiate a unilateral withdrawal from Gaza. Yes, the same policy position he campaigned against about a year before that decision. Sharon believed that this will buy some time for Israel while paying a minimal price and will decrease international pressure on it to make even further and larger concessions (Eldar, ibid). Sharon assumed that trying to please the wills of the international quartet and the Bush administration whilst keeping the same policy positions he campaigned on; would have implied maneuvering between the international community, the party, the coalition and the Israeli public opinion. Since the beginning of the 1990’s such maneuvering attempts by right-wing Israeli prime-ministers usually end up with political fights within the party, the coalition and as international pressures increase. Eventually the PM loses his position and the consequent elections. This was not a wild guess. Sharon knew it from the experience of Likud Prime Ministers in the 1990’s, Itzhak Shamir and Benjamin Netanyahu. The international community set a similar conflict resolution agenda for these Prime Ministers. They were compelled- to some extent- to follow that agenda. This created a wedge within the Likud intraparty factions and between the Likud and its far right coalition partners. That was the scenario that brought down Shamir’s government in 1992 and Netanyahu’s government in 1996, both Likud leaders facing these parallel pressures while keeping their original policy positions, losing the support of the public opinion, their coalition and their party’s leadership (Doron and Rosenthal, 2009). This is a quandary: high risk and uncertainty associated with a given choice between undesired options (Schofield and Gallego, 2011, p. 4). This is the setting for a 13

surprising heresthetic move: create (or renovate) a new option and construct the coalition needed for supporting it through a change in the policy space. Facing that quandary Sharon chose an option that was risky but more certain than the other option: the unilateral withdrawal from Gaza. This meant that Israel buys time from the international community while evacuating the least desired territory in comparison to the other occupied territories. When pressures would restart a couple of years down the road a new trick will be pulled out of the hat. These are not wild guesses. These claims were made by one of Sharon’s top aids, adv. Dov Weisglas, in a press interview given on October 2004 in one of Israel’s most important (and pro-appeasement) newspapers to one of Israel’s leading pundits (Shavit, Haaretz, October 6th 2004). Thus, in December 2003 Sharon radically changed his revealed preferences and started supporting the unilateral withdrawal option (Ben, Haaretz, 2/8/2004). As figure 3 shows Sharon’s coalition was based on the ideas he campaigned on: power-based strategy with regards to the Palestinians, and a rejection to the withdrawal from the Gaza strip. Clearly, that coalition could not have supported the new prowithdrawal position. Moreover, soon after Sharon made his declaration he saw a faction within the Likud forming against that new policy (Rynhold, 2007). Yet, Israel’s institutional structure allows prime ministers to initiate and lead policies quite flexibly as long as they control their coalition (Arian, 2005, pp. 264-265). Thus, a new coalition had to be formed seeking for replacements for the far right parties and the Likud faction which decided to act against the disengagement. For the left, Sharon was implementing their platform. Hence, there was no reason not to support Sharon. His bigger issue was to signal that his own party ‘base’ was still supporting him. Sharon decided to seek for that mandate from the Likud rank and file members who elected him to be the Likud’s leader. Yet, that support was not given and Sharon had to seek for other sources of support. He finally decided to lean on the government’s majority decisions and a continuous plea to the general public opinion while circumventing his party as a basis for legitimacy. During the implementation of the disengagement Sharon was both able to maintain a coalition and a high standing in public opinion surveys (Rynhold, 2007). An examination of figure 3 shows that Sharon moved to his electoral opponent’s position (the Labor) on security issues. The Labor having its policy implemented by the 14

Likud joined the coalition. The Labor’s support replaced the support of the National Unity party which its ministers were fired by Sharon on June 2004 since they risked his majority around the government’s table. The NRP gradually left the government starting in June finally leaving the government in October 2004 (NRG, 2/20/05). Moreover, along the process of implementation, the split between Sharon and the ‘rebel’ Likud MKs became fierce, with his finance minister (and contender for the Likud’s leadership) Benjamin Netanyahu resigning his office and supporting the rebels (Rynhold, 2007). The split within the Likud is a crucial factor here. On face value if the head of the party goes against the political party then the party is supposed to punish him and make him resign his office as the party's chairperson. Consequently, he is supposed to lose his nomination for a prime ministerial position that in a parliamentary democracy belongs to the party and not the candidate (Katz, 2001). A heresthetic agenda-setting move is based on re-constructing the policy space in order to handle the agenda setting process wisely (McLean, 2009). Here Sharon used a wedge which existed within the Likud movement (and the Israeli right) for several decades to create a within-Likud faction that would support his policy. The main agenda of Israel's right after 1967 became the issue of control over the territories Israel occupied in 1967. Yet, this agenda had two different implications associated with it. One was an ethnic policy implication that relates to these territories as a part of the Jewish people historical and religious heritage. Consequently, no one has the permission to negotiate over the territories let alone handing them to others. Another implication is that the control over the territories had a security-based military importance. Hence, the assessment of Israel's occupation has to be utility driven and if the occupation entails more risk than prospect then the territories can be waived (Rynhold, 2007). Knowing that with the ethnicity-driven part of the Likud he has no chance Sharon's public and parliamentary campaign was based on turning to the security-oriented part of the Likud. Thus, Sharon’s new coalition had the stable support of about security-oriented 20 Members of Knesset (MKs) from the Likud, the support of the labor party (additional 19 MKs) with Meretz (6 MKs), Shinui (15 MKs) some of the Arab parties (not formally included in the coalition) alongside opposition members that

15

supported the disengagement on an ad-hoc basis5. The disengagement plan was finally accepted by the Knesset as a law on February 16th 2005 and implemented in summer 2005. As the political final encore of that process Sharon and his supporters part ways with the Likud and formed a new party Kadima (forward) maintaining Sharon’s coalition with the Labor and Shinui (Spyer, 2007). The analysis shows the quandary, then the decision made by Sharon with his clear understanding that each choice will open up a strong competition against him from his coalition and his own party. Hence, this description seems to change the sequence of the theoretical dynamics proposed above. Yet, Sharon made a commitment to a policy and not implemented the policy itself. He went out to do the policy only when his original coalition completely collapsed and he found himself in a fierce fight (which he was on the verge of losing) with his party. He was compelled to raise an alternative prodisengagement coalition and only when he formed the alternative coalition Sharon could implement the disengagement. This illustration aimed to demonstrate the process of radical policy changes as brought forth by the heresthetic research program: a leader facing a quandary and then political bargaining from the old coalition and his party, contrives a policy turn to the ‘wrong side’, thereby creates a new policy coalition and promotes its policy. The next section presents the logic and main components of the research design for the large-N analysis. Large N-Analysis: Research Design The research design proposed here to study the factors that increase the likelihood for leaders moving to ‘the wrong side’ facing a quandary and intense political bargaining is a large-N research design based on the most-different systems rationale. The choice in this design strategy is different from the usual practice in the study associated with the heresthetic research program. Methodologically, this research program is characterized by a careful study of a relatively small number of cases, which have an abundance of the relevant data needed to construct and maintain such a careful analysis. Yet, the concern regarding a potential selection bias in that particular selection of cases (Aldrich and 5

See Knesset voting record on the disengagement process (Hebrew) using the following search engine in Knesset decisions: http://www.knesset.gov.il/description/eng/eng_work_mel7.htm the results are in Hebrew (alas) and are best viewed in Internet Explorer using the key word ‫( ההתנתקות‬viewed on December 22nd 2011).

16

Shepsle, 2000) yields a need to use a different research strategy. Thus, in order to test the external validity of claims stemming from this research program, I take here a large N most-different systems sampling strategy. That research strategy is based on choosing many and diverse countries and contexts while aiming at receiving a similar interaction between the variables. This type of analysis shows us the extent to which our theoretical claims are robust and do not depend on some local, area-based phenomenon (Przeworski, 2007). For the analysis, 55 cases of elected government’s reign were pooled from 37 countries on the basis of the availability of the needed historical facts relevant to the analysis and to extend the diversity of cases. The cases are drawn from 1990-2006 in consolidated democracies which their level of democracy was estimated using Polity IV. For most of the selected cases (presented in Appendix I), this was a period that stabilized them as democracies and opened the door to a usually non-violent internal political competition. It should be noted that this paper's result is limited to that historical context. Yet, geographically, socially, culturally and economically, the sample aimed at maximizing the variance in structural diversity by drawing cases from different contexts, continents and countries. Variables and Data 1. The control variables: social structures and political institutions- The social instability aspect of political instability includes the level of economic inequality in the distribution of revenues, economic welfare (Roemer, 2001), and civic liberties (Przeworski, 2006). As claimed above, the assumption is that as these variables have lower values in a specific setting then the likelihood for an exogenously set quandary in that setting might increase. Inequality was estimated by the Gini index supplied by the UN 2007/2008 Human development report (data were not available for all cases and years consequently cases with missing values were omitted from the analysis). As the values of this coefficient increase inequality increases. Economic welfare is estimated using countries' Gross Domestic Product. This indicator reflects a country's human capital, the accumulation of knowledge and the strength of its social institutions (Helpman, 2004). This was retrieved using the GDP in terms of Power Purchasing Parity (GDP-PPP weights) as supplied by the IMF as an indicator of the country’s aggregated purchasing 17

power in US dollar terms (International Monetary Fund, 2009). The civil liberties were estimated using the Freedom House database utilized in many comparative politics studies. As the index values increase, the extent of liberties granted to citizens in these countries decrease. The values for the variables included in the database were the values reported for that government’s last year in office indicating the structural setting associated with the government’s behavior. The second aspect of political instability which might interact with policy instability is cabinet instability. Following the theoretical review the variables which estimated cabinet survival and its influence on radical policy changes were the regime type, the number of parties in the legislature/parliament and the cabinet’s ability to conclude its Constitutional Inter-Election Period (CIEP) (King et. al. 1990). The variable regime type was defined with accordance to the Person and Tabellini 2001 database.6 They define presidential regimes as regimes in which the executive does not need the approval of the legislature to rule. They therefore classify semi-presidential and parliamentary regimes as a joint category. As for the number of parties for each of the cases, I used as a measure the number of parties in parliament, using Adam Carr’s election data website7 . The CIEP was estimated by a research assistant who was asked to discern the full constitutional term for a country in the sample and calculate whether an elected leader (president in presidential and semi-presidential regimes and a primeminister in parliamentary regimes) was able to rule for 75% of the designated term before going to new elections. The coding was binary with a below CIEP survival the case was coded 0 where cases with the CIEP and above were coded 1. 2. The independent variables: quandary and political competition- Following Schofield and Gallego (2010) institutional quandaries are defined as times of great uncertainty:”… a quandary is a choice situation where all possible options appear extremely unpleasant, and laden with risk and uncertainty.” (ibid, p. 4) In such times the ‘core’ of the policy 6

Viewed and down loaded from:

http://didattica.unibocconi.it/mypage/index.php?IdUte=48805&idr=4274&lingua=ita on December 7th 2011 7

Adam Carr's Election Archive was retrieved from

http://psephos.adam--‐carr.net/about/about.shtml viewed on 12/01/2010.

18

space is lost and the main political players are re-assessing their views aiming at new policy coalitions (Schofield 2006: 3-16). So as to make the analysis of quandary in a variety of cases, many of them not generous with information regarding public opinion, roll call voting and explicit records of government decisions, operationally quandary is defined here as a political issue which brings about a government crisis. In multiparty systems such a government crisis would be bargaining over the coalitional agreement, while in coalitions ruled by one party this would be an intraparty inter-faction crisis. This was coded as a binary variable where 0 meant no quandary and 1 meant quandary. As for political competition following the theoretical review I assume there are three levels of intensity in political competition: 1. Competition is strictly oppositional. Hence, the coalition is the best choice for its members 2. Competition includes intra-coalitional challengers against the leading party demanding more policies or perks to stay in the coalition. 3. Competition includes challengers from the coalition framer’s party, trying to challenge the framer’s position as premier. To discern levels two and three of competition from quandary it should be noted that quandary has to be associated with a policy topic while competition is more focused on the existence of a challenger to the leadership whether in the presence of a policy issue or without it. 3. The dependent variable: radical policy change- following McLean (2001) radical policy change was defined as a policy adopted by the head of the executive which she campaigned against during the elections. Coding Process for the Dependent and Independent Variables The coding process was done in three stages. In the first stage, the cases for analysis were randomly chosen from a large pool of cases, which vary in terms of geography, demography and culture. This first stage of choice yielded ninety cases. In the second stage a research assistant was asked to create a 10-15 lines summary for each of the cases. In that summary the research assistant needed to relate to: 1. The main policy the elected leader of the executive committed to during the elections that brought her/him to power. 19

2. The main issues that all the main political forces had to relate to publicly for an extensive period of time. 3. Crises in the ruling coalition and the issues that caused them. 4. The circumstances that led to new elections. 5. Was there a political quandary (defined as above) during that government’s tenure? The options were: yes or no. 6. Whether that leader completed at least 75% of her/his CIEP. The options were: yes or no. This summary was constructed using the Lexis-Nexus and Ebsco media databases. In the third stage two coders were asked to study the cases on the basis of the short summaries and answer two questions: 1. Where did the competition to the head of the executive come from? The options were: the opposition, the coalition and the leader’s own party. 2. Did the leader initiate a policy that contradicted the main campaign promises he/she made? The options were: yes or no. In order to maintain a final amount of cases that their relevance and meaning can be estimated with a relatively objective criterion, I filtered out cases of the database. The filtering out process aimed at a level of intercoder reliability which was sustained at a Spearman-rho correlation of (.772***) for estimating the levels of political competition and a Spearman-rho correlation of (.742***) for estimating the level of radical policy changes. The filtering out process aimed at maintaining variance in the contexts of cases to keep in line with the main purpose of the most-different systems strategy. As this research is exploratory an intercoder reliability of above (.70) is considered acceptable in quantitative content analysis (Lombard et. al. 2002). The next section presents the findings of the statistical analysis. Findings I start with the descriptive statistics of cases, run a correlation between the variables, scale the control variables using exploratory factor analysis and then run series of regression estimations of the theory. Table 1 about here

20

In the sample used as a basis for analysis, the median number of parties in Parliament was seven with 25% of the cases having more than nine parties in parliament, median GDP-PPP level was 193 billion dollars with about 25% of the cases having about a 538 and up billion dollars in their GDP-PPP capabilities. The median of civil liberties was 1.00 (the highest level) with 25% of the sample showing a lower level of political freedom and rights. The mean level of the Gini index was 34.8 with a normal distribution around that number. About 30% of the cases are presidential regimes and about 70% were parliamentary regimes. In most cases, the government completed its CIEP. The median of political competition was the expected in democracy: oppositioncoalition competition, meaning that overall the post-electoral coalition was the most desired option. Most cases experienced quandaries and radical policy changes characterized about 20% of the cases. These data are presented in table 1. Table 2 about here The next step before testing the hypotheses is to study the correlates between the specific variables. For these I used the Spearman-rho correlation for categorical variables. Some main correlations should be noted: 1. Economic welfare and economic inequality are correlated (.315*), where economic inequality also correlates with a low level of civil liberties (.460**). 2. Presidential regimes correlate with economic welfare (-.355**), with inequality (-.771**) and with a low level of civil liberties (-.588**). 3. There is a correlation between quandary and radical policy change (.463**) and between political competition with radical policy change (.416**). Yet, the correlation between quandary and political competition is low and not significant. Table 3 about Here Due to the obvious correlations within the socio-political structural variables, they were scaled using an exploratory factor analysis with the Principal Component Analysis as an extraction method and Varimax with Kaiser Normalization as the rotation method. Two factors were extracted and rotated and used as independent variables. In the context of this dataset this technique potentially solves two issues: the estimation of the construct validity of the theoretical claims, and a decrease in the concern for the 21

emergence of a biased estimation of the dependent variable due to multicollinearity. The following table shows the results of the factor analysis of the ‘structural’ variables. The variables which load on the first factor include presidential regimes, economic inequality, low level of civil liberties, economic welfare, a low correlation with number of parties and the ability to complete the CIEP. The variables which load on the second factor include parliamentary regimes, a high number of parties, high level of civil liberties and a low ability to complete the CIEP. Theoretically, this means that the two aspects of political stability namely structural and cabinet instability interact in a way which does not yield a clear separation between the two. This coincides with Person and Tabellini (2004) who claim that constitutional arrangements have clear interaction with country characteristics (ibid). Hence, when structure is estimated in this dataset it is done using two constructs that indicate that social characteristics and cabinet survival characteristics interact. Specifically they cluster around the characteristics of presidential regimes and parliamentary regimes. The estimation of the dependent variable which is binary and dichotomous is based on a binary logistic regression analysis. The results are presented in table 4. Table 4 about here The models estimated in table 4 include as a baseline for reference the two structural factors (Model I), the structural factors with quandary (Model II), the factors with political competition (Model III), the factors with quandary and political competition (Model IV) and the factors alongside the interaction between quandary and political competition (Model V). For each model the table presents the intercepts, both un-standardized (with std. dev. in parentheses) and the standardized parameters. In the lower part of the table the models' general estimates are presented. This includes the -2 log likelihood, model’s chi-square, Nagelkerke pseudo R-square and the correct classification of cases using the model estimates. Model I shows that the structural factors have a poor explanatory power for radical policy change as a dependent variable. The poor performance of that model is also illustrated in the model’s general characteristics, all low and insignificant. One can see by observing the socio-structural factors that their low and insignificant influence is consistent in all models. Quandary is included in Model II and provides an unclear 22

outcome. On the one hand the model estimates become strong and significant with 2= (15.720***), and Nagelkerke pseudo R2 = (.418). Moreover, its un-standardized B shows a positive influence on the dependent variable of (20.740). Hence, on face value quandary is an important and significant influence on the likelihood of a radical policy change. Yet, its very high standard error (8031.470) does not allow for a stable analysis of that variable and indeed its exp (b) is very low implying a very low influence on the dependent variable. Moving to model III Political competition estimated in Model III shows stronger performance than quandary with a significant exponent (b)=8.822***. The direct interpretation of that variable’s behavior is that as political competition increases across levels then the likelihood for a radical policy change increases. Despite the fact that the model characteristics demonstrate strong explanatory ability in comparison to Model I it in fact explains less the behavior of the dependent variable in comparison to Model II. That is, it has a 2= (14.32***), and Nagelkerke pseudo R2 = (.400) both lower than those demonstrated in Model II. However, it correctly classifies 88.5% of the observations in comparison to Model I and Model II which correctly classified only 80.8% of the observations. As expected when both quandary and political competition are regressed then these parameters sharply increase: 2= (25.395***), and Nagelkerke pseudo R2 = (.621). That model improves classification to 90.4%. When both variables are merged to an interaction variable, the largest significant exponent (b) independent variable is produced and equals (9.600***) implying that as both the competition increase and the situation is one of quandary the likelihood for policy change increases. Despite the strength of the interacted variable’s coefficient in comparison to the variables’ coefficients when regressed separately in the same model, Model V’s main characteristics have a lower explanatory power in comparison to Model IV. These include a 2= (24.494***), and Nagelkerke pseudo R2 = (.602) higher Initial -2 log likelihood (26.419) and the model classifies the cases as good as Model IV (90.4%). As with the theory and the case study it is clear from this analysis that the interaction between quandary and political competition affects leaders’ tendency to contrive a

23

radical policy change. Let us examine that interaction more closely using a population pyramid that appears in figure 4. Figure 4 about here The pyramid is split to five rectangular boxes. Each box represents one level of political competition. These include the three categories of competition (oppositioncoalition-premier’s party) and the two in-between averaged categories that represent the cases the coders were not able to agree upon. Each rectangular box is split to two matched squares. The left-hand square represents the observations that had no quandary and the right-hand square represents the cases that had a quandary. Within the squares the observations can have one of two categories: no radical policy change which is represented in the lower bars and a radical policy change represented in the upper bars. The frequency of these events is represented at the bottom of the figure. The top rectangular represents cases that experienced competition from the opposition, the middle rectangular represents cases with competition from the coalition and the bottom rectangular represents cases with competition from the premier’s party. The upper rectangular shows that in the presence of competition from the opposition even when there is a quandary in most cases there will not be a radical policy change. Such a change happens in the presence of competition from the opposition, in less than five of the almost 40 cases. When there is a competition from the coalition in most of the cases there are no quandaries and no radical policy changes (lower left darker color bar in the middle box). Yet, when there is a competition from the ruling party and a quandary there will be a radical policy change. Thus, the combination of quandary and a clear threat to the premier’s power from her party will yield a radical policy change.

24

Discussion The answer I gave above to the puzzle of radical policy changes was that as political systems face a quandary and leaders face intense political competition and need to bargain on valued policy positions, budgets and portfolios; they will be willing to change their coalition and the policy their coalition represents. I tested this claim using an illustrative case study of Ariel Sharon’s decision to unilaterally evacuate the Gaza strip. This analysis shows that while the pro-evacuation coalition existed from the postelectoral outset, the combination of quandary and Sharon’s belief (based on Sharon’s past experience) in the emergence of credible threat power to elements in the existing coalition and within Sharon’s party as long as the anti-evacuation coalition is kept. This combination yielded a radical policy change based on a pro-evacuation coalition. The large N analysis also shows that same finding. Both the regression model and then the population pyramid show that the combination of competition within the premier’s party and quandary are highly likely to yield a radical policy change. Methodologically this research shows that the heresthetic research program is able to give us tools for analysis of the circumstances leading to radical policy changes. Yet, in line with the criticism regarding Riker’s conclusions from social choice theory, these events are not ubiquitous. However, the impact of radical policy changes and their centrality yield a need to improve the conceptualization, measurement and operationalization of the concepts associated with the heresthetic research program. The analysis of the other aspects of political instability also deserve some consideration. This analysis shows that the social instability and cabinet instability interact in a way that creates unique settings for sub-populations of the sample. Moreover, they have proved to be unbeneficial in studying policy instability. There is no way in which I can claim that these issues are redundant in the study of policy instability. I can hypothesize (without proving it here due to lack of space) that these issues interact in a more nuanced way (than I did here) with quandary and bargaining. Put differently, it may well be that these institutions (both social and political) are an outcome of an endogenous process that may interact with policy instability. Hence, in order to fully account for their interaction with policy instability a different research strategy might be employed which will take this endogeneity into account. 25

Conclusion This paper aims to understand the reasons for policy instability using concepts developed by the heresthetic research program. Specifically, I used here a policy maneuver to the ‘wrong side’ (McLean 2001) as the dependent variable and quandary (Schofield 2006) and competition for leadership positions through political bargaining (Doron and Sened 2001) as the main explanatory variables. The theory proposed here was that quandaries change given policy spaces and re-open coalitional bargaining processes. In the end of such processes, the deviation between a new coalition set by a reigning premiere and the policy positions of the original coalition will depend on the threat power of the potential coalition partners over the premier’s political survival. As this threat will increase the premier will be willing to form a coalition on the basis of new policy lines with other partners which can less threaten her. Another option is that the premier will concede in policy terms just to survive in office even on the basis of radically different policy lines. An illustration for that dynamic was presented describing and explaining the decision taken by the former Israeli Prime minister, Ariel Sharon, on 2003-2005 to unilaterally evacuate the Gaza Strip. The large N quantitative analysis that followed the theory and illustration has shown that quandary and competition for leadership positions increase the likelihood for radical policy instability. The direct implication from this analysis is that leaders change policy positions so as to survive in power, even if they need to pursue goals they did not campaign on and for. They do that facing a quandary and facing a competition for their position stemming from their own ranks. Hence, potentially one of the tools leaders can use so as to deal with quandaries and policy instability is to contrive instability which might yield a coalition for them that will improve their survival prospects (McLean 2001; Schofield 2006; Schofiled and Gallego 2011). A more general methodological implication I draw here is that by focusing on the effects of quandaries on policy making, the tools suggested by the heresthetic research program, although rightfully challenged (Mackie 2003; McLean 2009), offer a beneficial and realistic perspective on policy instability (Schofield 2001; McLean 2009).

26

Bibliography Aldrich, J. H. and Shepsle K. A. (2000). ‘Explaining Institutional Change: Soaking, Poking and Modeling in the US Congress’, in: W. T. Bianco (ed.) Congress on display, Congress at work. Ann-Arbor: Michigan University Press. Austen-Smith, D. and Banks, J. S. (2000). Positive Political Theory I: Collective Preferences. Ann Arbor: Michigan University Press. ----- 2005. Positive Political Theory II: Strategy & Structure. Ann Arbor: Michigan University Press Banks, J. S. and J. Duggan (2000). ‘A general Bargaining Model of Legislative PolicyMaking’, Quarterly Journal of Political Science 1: 49-85 Baron, D.P. and Ferejohn, J. A. (1989). ‘Bargaining in Legislatures’, The American Political Science Review, Vol. 83(4), pp. 1181-1206. Ben, A. (8/2/2004). ‘Defeat? Maybe an Improved Draw’. Haaretz (viewed on October 28th 2011). http://www.haaretz.co.il/misc/1.944290 Damgaard, E. (2008). ’Cabinet Termination’. In: Cabinets and Coalition Bargaining: The Democratic Life Cycle in Western Europe, eds. K. Strom, W. C. Mueller and T. Bergman, Oxford: Oxford University Press, pp.301-326. Dardanelli, P. (2009). ‘Europeanization as Heresthetics: Party Competition over Self-Government for Scotland 1974—97’, Party Politics 15: 49-68. Doron, G. (2005). ’Right as Opposed to Wrong as Opposed to Left: The Spatial Location of "Right Parties" on the Israeli Political Map’, Israel Studies 10(3): 29-53 Doron, G. and Sened, I. (2001). Political Bargaining. London: Sage Publications. Duverger, M. (1952). ‘Public Opinion and Political Parties in France’, The American Political Science Review 46 (4), 1069-1078. Eldar, A. (5/23/2005). ‘Exposing the Real Face’. Haaretz. Viewed on: October 28th 2011 Epstein, L. and Shvetsova, O.V. (2002).’ Heresthetical Maneuvering on the US Supreme Court’, Journal of Theoretical Politics 14 (1): 93-122. Ferrie, J.N., Baudouin D., and Legrand, V. (2008). ’Comprendre la Délibération

27

Parlementaire: une approche praxeologique de la politique en action." [Understanding Parliamentary Deliberation: a Praxeological Approach to Political Action] Revue Française de Science Politique [French Political Science Review] 58 (5): 795-815. Helpman, E. (2004). The Mystery of Economic Growth. Cambridge, MA: Belknap Press of Harvard University Press. Huber, J. D. (1998). 'How Does Cabinet Instability Affect Political Performance? Portfolio Volatility and Health Care Cost Containment in Parliamentary Democracy’, American Political Science Review, 92(2): 577-591. Huber, J. D. and Lupia, A. (2001). ‘Cabinet Instability and Delegation in Parliamentary Democracies’, American Journal of Political Science, 54(1):1832. International Monetary Fund. (2009). World Economic and Financial Surveys. Retrieved 12/01/2009 http://www.imf.org/external/pubs/ft/weo/2008/02/weodata/index.aspx Katz, R. (2001). ‘The Problem of Candidate Selection and Models of Party Democracy’, Party Politics, 7(3): 277-296 King, G., Alt, J. E., Burns, N. E., and Laver, M. (1990). ‘A Unified Model of Cabinet Dissolution in Parliamentary Democracies’, American Journal of Political Science, 34 (3): 846-871. Koff, H. and Duprez, D. (2009). ‘The 2005 Riots in France: The International Impact of Domestic Violence’, Journal of Ethnic and Migration Studies 35(5): 713_730 Koren, D. (2010). ‘Arab Israeli Citizens in the 2009 Elections: between Israeli Citizenship and Palestinian Arab Identity’, Israel Affairs, 16(1): 124–141 Laver, M. and Schofield, N. J. (1990). Multiparty Government: The Politics of Coalition in Western Europe. Oxford: Oxford University Press. Lombard, M., Snyder-Duch, J. and Campanella-Bracken, J. (2002). ‘Content Analysis in Mass Communication: Assessment and Reporting of Intercoder Reliability’, Human Communication Research, 28(4): 587–604 Lupia, A. and Kaare, S. (1995). ‘Coalition Termination and the Strategic Timing of 28

Parliamentary Elections’, The American Political Science Review , 89 (3): 648665. Mackie, G. (2003). Democracy Defended. Cambridge: Cambridge University Press. Mainwaring, S. (1993).’ Presidentialism, Multipartism and Democracy’, Comparative Political Studies 26 (2): 198-228. McLean, I. (2001). Rational Choice & British Politics: an analysis of rhetoric and manipulation from Peel to Blair. Oxford: Oxford University Press. — (2002). William Riker and the Invention of Heresthetic(s). British Journal of Political Science 32(2): 535-58. — (2009). In Riker’s Footsteps. British Journal of Political Science 39(1): 195-210. NRG (2/20/2005). ‘The Road to Evacuation’. (Hebrew) http://www.nrg.co.il/online/1/ART/872/607.html viewed on: 11/08/11. Ordeshook, P. C. and Shvetsova, O. V. (1994). ‘Ethnic Heterogeneity, District Magnitude, and the Number of Parties’, American Journal of Political Science. 38(1): 100-123. Przeworski, A. (2006). ‘Self-Enforcing Democracy’. In: Weingast B. R. and D. Wittman (eds.) The Oxford handbook of Political Economy. NY: Oxford University Press. — (2007). ‘Is the Science of Comparative Politics Possible?’, In: Boix, C. and S. C. Stokes (eds.) Oxford Handbook of Comparative Politics, NY: Oxford University Press. Przeworski, A., Alvarez, M., Cheibub, J. A., & Limongi, F. (1996). ‘What Makes Democracies Endure?’, Journal of Democracy 7(1): 39-55. Riker, W. H. (1980). ‘Implications from the Disequilibrium of Majority Rule to the Study of Institutions’, American Political Science Review, 74: 432-446. — (1982a). Liberalism Against Populism. Illinois: Waveland Press. — (1982b). ’Two Part System and Duverger's law: an Essay on the History of Political Science’, American Political Science Review, 76(4): 753-766. — (1984). ‘The Heresthetics of Constitution-Making: The Presidency in 1787, with Comments on Determinism and Rational Choice’, American Political Science Review 78(1): 1-16. 29

—(1986). The Art of Political Manipulation. New Haven: Yale University Press. — (1990). ‘Heresthetic and Rhetoric in the Spatial Model’, In: Enelow, J. and Hinich, M. (eds.) Advances in the Spatial Theory of Voting. Cambridge: Cambridge University Press. Roemer, J. E. (2001). Political Competition. Cambridge MA: Harvard University Press. Rosenthal, M. (2011). ’Agenda Control in an Unstable Parliamentary Democracy: Evidence from the Israeli Public Sector’, Constitutional Political Economy, Published online 30 September 2011. Rosenthal M. and Doron. G. (2009). ‘Israel's 1993 Decision to Make Peace with the PLO: Or How Political Losers (this time) Became Winners’, International Negotiation, 14(3):449-474. Rynhold, J. and Steinberg, G. (2004). ‘The Peace Process and the Israeli Elections’, Israel Affairs 10(4): 181-204 Rynhold, J. (2007). ‘Peace and Security in the 2006 Election’, Israel Affairs, 13(2), pp.384–400 Schleiter, P. and Morgan-Jones, E. (2009). ‘Constitutional Power and Competing Risks: Monarchs, Presidents, Prime Ministers and the Termination of East and West European Cabinets’, American Political Science Review, 103(3): 496-512. Schofield, N. J. (2001). ‘Constitutions, Voting and Democracy’. Social Choice and Welfare, 18: 571-600. —2006. Architects of Political Change. Cambridge: Cambridge University Press. Schofield, N. J. and Sened, I. (2006). Multiparty Democracy: Elections and Legislative Politics. New York: Cambridge University Press. Schofield, N. J. and M. Gallego (with J. Jeon and U. Ozdemir) (2011). Chaos or Leadership: The Heart and Soul of Politics. Heidelberg: Springer. Schumpeter, J. A. (1952). Capitalism, Socialism and Democracy. London: Unwin University Books. Shapiro, I. (2011). The Real World of Democratic Theory. Princeton: Princeton University Press. Shavit, A. (10/6/2004). ‘Top PM aide: Gaza plan aims to freeze the peace process’,

30

Haaretz.

http://www.haaretz.com/print-edition/news/top-pm-aide-gaza-plan-

aims-to-freeze-the-peace-process-1.136686 viewed on November 10th 2011. Shepsle, K. A. (2002). Political Losers. The inaugural William H. Riker Lecture delivered at the annual meeting of the Public Choice Society. — (2006). ‘Rational Choice Institutionalism’. In Rhodes, RAW , Binder, S.A. and B. A. Rockman (eds.) The Oxford Handbook of Political Institutions. Oxford: Oxford University Press. Shumpalbi, A. (4/29/2004).’The Stomach of Likud Party Members’, YNET. http://www.ynet.co.il/articles/0,7340,L-2909351,00.html Viewed in December 22nd 2011. Stokes, S. C. (2001). Mandates and Democracy. New York: Cambridge University Press. Spyer, J. (2007). ‘Downfall of a Dominant Party: The Likud and the 2006 Election’, Israel Affairs, 13(2), pp.289–304 Tsebelis, G. (2002). Veto Players. Princeton NJ: Russell Sage Foundation.

31

Marked with yellow received 6 and above in democratization in Polity IV If there were not measurements in Polity IV for the years studied then the last observation and the current (2010) observation were averaged while controlling for whether there has been a regime breakdown in that country. Leaders

Years

Countries

Leaders

Years

Countries

Leaders

Nestor

2003-

Argentina

Geir Haarde

2006-

Iceland*

José Manuel 2002-

Kirschner

2006

2007

Years

Durão

Countries Portugal

2004

Barroso John

1998-

Howard

2001

John

2001-

Howard

2004

Viktor

1997-

Klima

2000

Wolfgang

2000-

Schussel

2007

Guy

1999-

Verhofstadt

2003

Guy

2003-

Verhofstadt

2007

Jean-Luc

1995-

Dehaene

1999

Mathieu

2001-

Kerekou

2006

Lula

de- 2002-

Silva

2006

Paul Martin

20042006 8

Australia

Davio Odsson

2003-

Iceland*

2004 Australia

Atal

Bihari 1999-

Vajpayee Austria

Austria

India

2004

Inder Kumar 1997Gujral

1998

John Bruton

1994-

India

Ireland

1997 Belgium

Bertie Ahern

1997-

Ireland

2002 Belgium

Ariel Sharon

2003-

Israel

2006 Belgium

Ehud Olmert

2006-

Israel

2008 Benin 10

Brazil

Canada

Silvio

2000-

Berlosconi

2005

Jean-Claude

1995-

Juncker

1998

Jean-Claude

1999-

Juncker

2004

Italy

António

1995-

Guterres

1999

Vladimir

1999-

Putin

2004

Vladaimir

2004-

Putin

2006

Vladimir

1994-

Meclar

1998

Mikuláš

2002-

Dzurinda

2006

Nelson

1994-

South

Mandela

1999

Africa9

Jose

Maria 1996-

Aznar

2000

Ingvar

1994-

Carlsson

1996

Luxemburg* Goran Persson Luxemburg* Chen

1998-

32

Russia8

Russia

Slovakia

Slovakia

Spain

Sweden

Sweden

2002 Shui- 2004-

Received below 6 in Democratization in POLITY IV. POLITY IV indicates that South Africa in 1994 is a regime which has collapsed and an authority in the making. Next time South Africa is measured is 2010 and receives 10 Received below 6 in Democratization in POLITY IV. 9

Portugal

Taiwan

Joseph

1997-

Canada

Jacques Jean 2000

Eddie Fenech 1998Adami

2004

Vicente Fox

2001-

Malta*

bian

2008

John Major

1992-

UK

1997

Chrétien Eduardo Frei

1994

Chile

Ruiz- 2000

Mexico

Jorge Battle

2006

2000-

Uruguay

2005

Tagle Poul Nyrup 1998Rasmussen

2001

Anders

2001-

Fogh

2005

Denmark

Wim Kok

1998-

Netherlands

2002 Denmark

Helen Clark

William

1992-

Clinton

1996

1999-

New

William

1996-

2002

Zealand

Clinton

2000

2002-

New

2005

Zealand

Olusegun

1999-

Nigeria11

Obasanjo

2003

Rasmussen Gerhard

2002-

Schroder

2005

George

1993-

Papandreou

1996

Viktor

1998-

Orban

2002

Germany

Greece

Hungary

Helen Clark

Gro

Harlem 1993-

Brundtland/

Norway

1997

Thorbjorn Jagland Ferenc

2004-

Gyurcsany

2006

Hungary

Gro

Harlem 1993-

Brundtland/

Norway

1997

Thorbjorn Jagland Estrada

1998-

Philippines

2001 Aleksander

2000-

Kwaśniewski

2005

Poland

* Not included in Polity IV 11

Nigeria was denoted as regime breakdown on 1998 and then 5 on 2003. An uneasy decision was made to leave it within the database on the basis of its relative average strength in democratization in comparison to other countries and the access to the country information needed.

33

US

US

Policy Instability: Deviation of leaders from policies they previously committeed to.

Political Instability: society, Institutions and Policy

Cabinet Instability: the ability of an elected coalition to maintain its leadership position.

Figure 1: Political Instability as a Multifaceted Concept

34

Socio-Structural Instability: social acts against government stemming from resistance to government policies and/or behaviors.

Figure 2: Factors Increasing the Likelihood for Policy Instability

The ruling coalition is the outcome of a bargaining process within a given policy space

35

A breakup of the existing policy space due to exogenous or endogenous efforts yields a new bargaining process

The rate of change and its direction in the policy space is mediated by structural factors

Radical Policy Change: choosing the opposition's policy and forming a new coalition on this basis

Weighted Party Policy Positions 2003 Kneset Elections 1 RAAM Balad 2 Hadash 3 2

0.8

0.6 Meretz 6

0.4 Shinui 15

Labor 19 Am Ehad 3

0.2

0 Arab-Israeli Conflict

-1

-0.8

-0.6 -0.4 Likud for disengagement 20

-0.2

0 -0.2

-0.4

-0.6

0.2 Israel Baaliya 2

0.4

National Unity 7 NRP 6

Shas Yahadut 11 5

-0.8

-1 State and Religion

Figure 3: Weighted Party Policy Positions 2003 Kneset Elections

36

0.6 Likud 38

0.8

Figure 4: The Interaction between Quandary and Political Competition and its Influence on Radical Policy Changes

37

Table 1: Descriptive Measures of Variables

Radical Economic

Economic

Democratic

Number ofRegime

Completion

Welfare

Inequality

Freedoms

Parties

Type

term

Valid

55

53

55

54

55

Missing

0

2

0

1

Mean

638.3833

34.8000

1.4364

Median

193.0350

34.3000

N

a

Political

Policy

Quandary

Competition

Change

55

55

55

55

0

0

0

0

0

8.2222

1.745

.8000

.5091

1.4636

.1818

1.0000

7.0000

2.0000

1.0000

1.0000

1.0000

.0000

1.00

5.00

2.00

1.00

1.00

1.00

.00

Mode

6.59

Std. Deviation

1363.68997

8.57716

.99561

5.97689

.439

.40369

.50452

.65866

.38925

Variance

1859650.327

73.568

.991

35.723

.193

.163

.255

.434

.152

25

91.2640

26.4500

1.0000

5.0000

1.0000

1.0000

.0000

1.0000

.0000

50

193.0350

34.3000

1.0000

7.0000

2.0000

1.0000

1.0000

1.0000

.0000

75

538.2360

38.9000

2.0000

9.2500

2.0000

1.0000

1.0000

2.0000

.0000

Percentiles

25.00

a

of

a. Multiple modes exist. The smallest value is shown

38

Welfare

equality

Civic

Number Regime

Liberties

of Parties Type

CIEP

Quandary

Political_

Policy

Competition

_Change

Economic Welfare Correlation Coefficient 1.000 Sig. (2-tailed)

.

N

55

Economic

Correlation Coefficient .315*

1.000

Inequality

Sig. (2-tailed)

.021

.

N

53

53

Civic Liberties

Correlation Coefficient .122

.460**

1.000

Sig. (2-tailed)

.376

.001

.

N

55

53

55

Number of

Correlation Coefficient .210

.156

.178

1.000

Parties

Sig. (2-tailed)

.127

.270

.198

.

N

54

52

Regime

Correlation Coefficient -.355**

-.771

Type

Sig. (2-tailed)

.008

.000

N

55

54 **

-.588

54 **

.036

1.000

.000

.887

.

53

55

54

55

Completion

Correlation Coefficient .003

.078

.098

.053

-.188

1.000

of term

Sig. (2-tailed)

.983

.581

.478

.701

.170

.

N

55

53

55

54

55

55

Correlation Coefficient .234

.147

.236

.005

-.156

-.127

1.000

Sig. (2-tailed)

.086

.294

.083

.973

. 254

.354

.

N

55

53

55

54

55

55

55

Quandary

Political

Correlation Coefficient .224

.101

.062

.087

.031

-.241

.194

1.000

Competition

Sig. (2-tailed)

.100

.470

.652

.531

.820

.076

.155

.

N

55

53

55

54

55

55

55

55

.038

.051

.014

-.049

-.118

.463**

.416**

1.000

Radical Change

PolicyCorrelation Coefficient .107 Sig. (2-tailed)

.437

.788

.713

.921

.721

.391

.000

.002

.

N

55

53

55

54

55

55

55

55

55

*. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed).

Table 2: Spearman-Rho Correlations between Variables

39

Table 3 : Factor Analysis of Indicators Component Indicators

Presidential

Parliamentary

Regimes

Regimes

Growth Level

.442

.108

Economic

.841

.036

.656

.244

Number of Parties .084

.984

Regime Type

-.964

.109

Completion

.442

.108

Inequality Democratic Freedoms

of term Extraction Method: Principal Component Analysis. Rotation

Method:

Normalization.

40

Varimax

with

Kaiser

Table 4: Model fit of Policy Change

Radical Policy Change Model I B Exp (SE) (b) .113 1.120 (.343)

Model II B Exp (SE) (b) -.029 .971 (.412)

B (SE) -.116 (.505)

-.050 (.371)

.951

-.214 (.360)

.807

-.309 (.444)

----

----

Political Competition Quandary  Political Competition

----

----

20.740 (8031.470) ----

1.017 E9 ----

----

----

----

----

Constant

-1.440 (.353)

.237***

Presidential Regimes Factor Parliamenta ry Regimes Factor Quandary

-2 log likelihood Model 2 Nagelkerke Pesudo R2 % Correctly Classified N

-21.246 (8031.470)

.000

Model III Exp (b) .891

Model IV Exp (b) .664

B (SE) -.468 (.624)

Exp (b) .626

.734

-.447 (.422)

.639

-.469 (.438)

.625

----

----

6.988E8

----

----

2.107 (.644) ----

8.822***

20.365 (7604.183) 1.911 (.731) ----

6.671***

----

----

----

2.262 (.753)

9.600***

-4.958 (1.257)

.007***

-24.077 (7604.183)

.000

----

-4.412 (1.158)

.012***

50.787

35.193

35.981

25.518

26.419

.126 .004

15.720*** .418

14.932*** .400

25.395*** .621

24.494*** .602

80.8

80.8

88.5

90.4

90.4

55

55

55

55

55

1. N= 55 2. Base Line Category- Policy in accordance with platform 3. Standard Error in Parentheses 4. * significance