Electoral Competition and Portfolio Allocation across Regimes Types

Transcription

1 Electoral Competition and Portfolio Allocation across Regimes Types Royce Carroll Rice University March 2009 Updated: December 2009 An earlier version of this paper was presented at the 2009 Midwest Political Science Association

2 Abstract In this paper, I extend the empirical study of portfolio allocation studies to account for the potential of electoral constraints across institutional variation. The study includes a crossregional sample of parties in all democratic governments since 1990, including presidential regimes. With this sample, I present an integrated analysis comparable in approach to the previous studies of office distribution, with a focus on the three key concepts from the literature: proportionality, agenda setting power and bargaining power. With this data, I consider the impact of electoral environment on post-electoral bargaining across regime types, extending the work of Carroll and Cox (2007) in examining the potential constraining effect of pre-electoral alliances among parties on the portfolio allocation process. I show that once regimes having presidents with appointment powers are separated out, there is no formateur effect in parliamentary regimes. In a group of semi-presidential regimes and parliamentary regimes with party-affiliated heads of state, a formateur bonus in partisan cabinet members is observed that is about half that of full presidential regimes. In terms of the effects of electoral cooperation, I show that within each regime type, 1) post-electoral governments exhibit a larger formateur effect in the presidential and semi-presidential cases and 2) a smaller voting weight effect in parliamentary and semi-presidential cases and 3) governments based on pre-electoral alliances allocate portfolios more in proportion to seats and that formateur and voting weight effects are reduced or eliminated.

3 Governing coalitions must determine how to reward partners for their contribution. Perhaps the most commonly studied form of reward within a winning coalition is the allocation of cabinet portfolios. Whether they are interested primarily in controlling a policy jurisdiction or accessing the resources of a ministry for particularistic purposes, office seeking parties view the control of cabinet ministries as a primary objective. In this paper, I extend the empirical study of portfolio allocation studies to account for the potential of electoral constraints across institutional variation. The study includes a cross-regional sample of parties in all democratic governments since 1990, including presidential regimes. With this sample, I present an integrated analysis comparable in approach to the previous studies of office distribution, with a focus on the three key concepts from the literature: proportionality, agenda setting power and bargaining power. With this data, I consider the impact of electoral environment on post-electoral bargaining across regime types, extending the work of Carroll and Cox (2007) in examining the potential constraining effect of pre-electoral alliances among parties on the portfolio allocation process. Theories on how portfolios should be divided among governing partners have pointed to three factors seat contributions, voting power, and agenda-setting (formateur) advantages. I show that once regimes having presidents with appointment powers are separated out, there is no formateur effect in parliamentary regimes. In a group of semipresidential regimes and parliamentary regimes with party-affiliated heads of state, a formateur bonus in partisan cabinet members is observed that is about half that of full presidential regimes. In terms of the effects of electoral cooperation, I show that within each regime type, 1) post-electoral governments exhibit a larger formateur effect in the presidential and semi-presidential cases and 2) a smaller voting weight effect in parliamentary

4 and semi-presidential cases and 3) governments based on pre-electoral alliances allocate portfolios more in proportion to seats and that formateur and voting weight effects are reduced or eliminated. Fairness and proportionality Gamson (1961) suggested that parties forming a government will expect others to demand from a coalition a share of the payoff proportional to the amount of resources which they contribute to [that] coalition (p. 376). For Gamson, the relevant resources are the seats in the assembly contributed by each party in government, such that portfolios should be proportionally allocated. Work by Browne and Franklin (1973), Browne and Frendreis (1980), Schofield and Laver (1985), Warwick and Druckman (2001, 2006), Druckman and Warwick (2005) strongly supported this expectation. 1 This pattern, often called Gamson s Law, retains a status as one of the most reliable results in political behavior (Laver 1998). Several possible explanations for a fairness outcome might exist. Warwick and Druckman conjecture that parties do not exploit their agenda-setting power in any one situation because they are involved in longer-term relationships (2006, 660). Carroll and Cox elaborate on this intuition, arguing that seat-based allocations are at least partially explained by the electoral environment, specifically that pre-electoral partners have an incentive to pursue such allocations. Although their empirical focus is on formal public pacts among parties, the intuition applies to more informal situations where parties are cooperative during election due to a strong expectation of interdependence. 1 Warwick and Druckman s work also tests whether weighting portfolios by salience would show that proportional allocations in Western Europe were illusory. They find that proportionality was upheld, even with weights.

5 Bargaining Power Strictly seat-based allocations are, however, not consistent with theories of bargaining. Bargaining theory literature emphasizes pivotalness, based on the notion of bargaining weight or voting weight. A party s bargaining weight depends on how many alternative majority coalitions for which the party is necessary. Bargaining models expect that a party s portfolio payoff will be higher when it can pivot between more alternative minimal winning coalitions (Schofield 1976; Ansolabehere et al. 2005; Snyder, Ting and Ansolabehere 2005). Post-electoral environments present parties with varying degrees of pivotalness in the formation of coalitions. This depends not just on their seats, but on the possible combinations of seats that can form majorities. To clarify, consider a parliament with three parties holding 44%, 34% and 22% of the seats, respectively. Were the largest and smallest parties to form a coalition government, a proportional (seat-based) allocation of the cabinet would give the largest party 2/3 of the portfolios, as it contributes 2/3 of the coalition s total seats. In bargaining theory, the resource each party brings is ability to form a majority coalition. In this example, each party can form a majority with any other single party, and thus contribute equally in this resource. Therefore, bargaining theory suggests that the two parties expected payoffs will also be equal (Schofield 1976; Laver and Schofield 1985; Snyder, Ting and Ansolabehere 2005, Ansolabehere et al 2005), despite that one contributes far fewer seats. These models predict that small parties who are equally necessary to the majority can propose and receive a number of portfolios equal to that of the larger partner who has the same weight in government formation. Distributions consistent with bargaining power that are significantly different from seat-based allocations are not observed

6 with regularity, but for present purposes the aim is to establish conditions where they are more likely to emerge. Agenda-setting power In addition to bargaining power in terms of attaining majorities, an agenda-setting argument suggests that the party in a position to propose the coalition usually called the formateur party in parliamentary regimes 2 will take shares of portfolios beyond either their bargaining weights or seat shares. The theory behind the formateur bonus derives from Baron and Ferejohn s (1989) model of legislative bargaining. The formateur in their model has a monopoly right to propose governments and portfolio allocations and can thus make a single take-it-or-leave-it offer. Rejecting this offer means waiting until the formateur party completes its subsequent bargaining before another chance at forming a government would emerge, entailing delay costs. One key assumptions behind the Baron and Ferejohn result is that the formateur has a monopoly right to propose a government and that the probability of retaining formateur status does not decline, which would be undermined if it were possible for an alternate coalition to override the formateur s charge: Morelli s (1999) model without this assumption predicts no formateur bonus. The other assumption is that is that in the event of a rejected offer, any party including the initial formateur can be chosen to make a new offer with the same probabilities that led to the initial selection. Weakening this assumption too would likely reduce any formateur effect. Empirical evidence of a formateur effect has not been found in pooled samples of parliamentary governments when controlling for governing parties seat contribution to the coalition, at least when the prime minister s position is equally weighted. Snyder, Ting and 2 Almost always the prospective prime minister.

7 Ansolabehere (2005) show however that a formateur bonus is present only if one does not control for each governing party s seat share. Portfolio Allocation across Regime types As noted, most extant studies of portfolio allocations deal with parliamentary regimes. Many of these include some Western European regimes with presidential influence while Druckman and Roberts (2005) include countries in Eastern Europe where regime type might play a role, although this is not their focus. Regimes with a prime-minister and president with cabinet appointment power are, in general, similar to parliamentary regimes. However, the concept of formateur is potentially different in such regimes if a president s power is in a position to privilege certain potential coalitions over others. As a result, the leverage of a formateur in such regimes could be greater, given that the president might restrict the probability of alternate formateurs forming alternate majorities (cf. Amorim Neto and Strøm 2006). This most clearly applies to premier-presidential systems (Shugart 2005) but is potentially relevant in parliamentary republics where presidents have to the party system. Some scholars, most notably Amorim Neto (2006), have discussed variation in portfolio allocations across presidential regimes, pointing to the specific powers and economic circumstances. 3 Indeed, as Figure 1 shows, there is a great deal of variance in the proportionality of portfolio allocations in presidential regimes. From this perspective, in a presidential regime, if controlling the government is all that matters, the president s 3 Amorim Neto evaluates the proportionality of portfolio allocations with a slightly different focus than that in the parliamentary government literature, with the emphasis on the degree to which legislative parties are generally integrated into the executive (including single-party governments). The present focus is to determine the extent to which parties portfolio payoffs among governing parties are proportional to their internal contributions, given that parties have agreed to join a multi-party government, rather than the degree of coalition building more generally.

8 advantage in proposing a portfolio allocation is absolute. Since a president s status does not depend on legislative support, a strict interpretation of the formateur advantage predicts that that the president should take all portfolios for co-partisans 4. However, we might instead assume, as does Amorim Neto (2006), that statutory majorities have a separate, if conditional, value in presidential regimes (Altman 2000, Amorim Neto 2002, Amorim Neto and Santos 2001, Cheibub, et al. 2004)). 5 The need to secure a legislative majority for passage of legislation is valuable enough to limit a president s freedom to make appointments without compromising with legislative parties. At the same time, legislative parties clearly lack the leverage they have in parliamentary regimes where their cooperation is necessary to the survival of the government. The importance of voting weight is most apparent in the case of parliamentary regimes where, from a purely office seeking perspective, a straightforward and consistent payoff is applied to majority status. In presidential regimes, one might imagine that since the The formateur advantage, by contrast, is more difficult to reconcile with a pure parliamentary environment in practice. Formateurs do not typically possess monopoly proposal power as long as alternative majorities can overturn their position. To the extent that formateurs are in a strong position, this is likely to be largely endogenous to the bargaining environment, given that formateurs obtain their status based on a lack of credible alternatives (Bäck and Dumont 2008). In other words, the party of the formateur may only have advantages already accounted for by its seat share (or perhaps voting weight). In presidential regimes, the president always retains the power to appoint the cabinet regardless of the response from prospective partners, such that the "formateur" advantage is 4 Or, alternatively, non-partisans loyal to the president. This issue is discussed below. 5 Moreover, there is additional legislative value in obtaining posts that can be controlled directly by the assembly majority (Carroll, Cox and Pachón 2006).

9 subject mainly to incentives to form more stable legislative majorities, which in turn depend on political circumstances. Electoral competition and cooperation Bargaining theory assumes that parties exist only in the legislature and use postelectoral information to make distributional decisions, but several issues interfere with this view. One of these is that parties are often constrained by the electoral environment. This is particularly true with regard to electoral pacts, which clearly influence government formation (Golder 2006b, Debus 2008). Commitments to govern together suggest not only the possibility of determining partners before elections, but agreeing to distributions as well. Electoral pacts, negotiated prior to the seat allocation, alter the post-electoral bargaining environment and introduce additional opportunities for parties to contribute resources. The logic of office-seeking parties attempting to extract the largest possible share of portfolios they can obtain given their bargaining position is stronger if negotiations to build the governing coalition are entirely post-electoral, and therefore unconstrained by previous agreements. The more important a party is to producing the majority (given its relative size and alternatives), the more it should demand in return for this contribution. If parties are opportunistic in this fashion, it is reasonable to expect post-electoral behavior be associated with pivotal parties obtaining office benefits exceeding their seat shares in government, as envisioned by Snyder Ansolabehere and Ting (2005). If pre-electoral alliances are mutually desirable because they enhance the probability of success (Golder 2006), and are costly (in terms of reputation, compromises in campaign, etc.), parties may want an allocation rule that preempts opportunistic behavior by pivotal parties or formateurs. Carroll and Cox (2007) argue that if pre-electoral coalitions negotiate

10 agreements that will regulate the government formation process their partners are already known and some commitment is made then partners have reason to restrict somewhat the full exploitation of post-electoral bargaining advantages. In order for the alliance to be successful, they must collectively win the election (be able to form a government). If a partner s payoff will depend only on post-electoral circumstances, there is no assurance that contributing more (costly) effort to the alliance will bring any marginal increase in voting weight (for the individual party). However, if partners agree ex ante to be rewarded for their output the seats they win -- all partners have an incentive to join and contribute as vote/seat-maximizers to the coalition s electoral effort, thereby enhancing the coalition's probability of reaching its majority goal. To put it another way, if some parties do form a pact, each member party s investment in campaigning for a collective electoral win produces an increased probability of a coalitional majority (a public good for the coalition as a whole). By agreeing ex ante to a proportional division of office spoils, based on seats without regard to pivotalness, the coalition can motivate cooperative campaigning and investment in the coalition s success. In effect, parties in electoral pacts trade away their practical ability to be pivot among alternate coalitions in exchange for the benefits of an increased chance to be in government. Thus, portfolio allocations should be more proportional (to seat share) in governments based on pre-election pacts than in other governments, holding constant other factors, while bargaining weight should not be important. Most of the literature on pre-election coalitions is explicitly or implicitly tied to parliamentary regimes. In presidential regimes, electoral coalitions are less studied and deserve more elaboration to evaluate the applicability of this argument. In presidential systems, the president plays the role of formateur in the sense of making an offer to attract

11 support to a governing coalition. The key difference is the value of the legislative coalition, which, in presidential regimes does not include upholding the government. Thus, along with the separation of survival between the executive and legislative branches is the separation of government formation (as defined in parliamentary literature) from coalition formation (as defined by works such as Amorim Neto 2006). The need of even the most powerful presidents to achieve some measure of legislative support counters this tendency as noted above (Deheza 1997,. Yet, there is a sense in which the promise of ministries can buy support for a government, since multi-party presidential regimes in practice rely on the votes of other parties support bases in winning the presidency. In the process, a presidential coalition can dramatically reduce the amount of post-electoral bargaining that actually occurs since the identity of cabinet partners is pre-determined. More importantly, pre-electoral bargaining reduces somewhat the president s ability to keep a large proportion of portfolios for his or her own party, by promising a greater proportion (perhaps specific ministries). In general, then, pre-electoral bargaining should increase proportionality as parties who supported presidential candidates receive cabinet portfolios in exchange for their preelectoral support. These arguments lead me to three empirical predictions regarding the effects of preelectoral negotiation: 1. A governing party s share of portfolios will be more proportional to its contribution of seats to the governing coalition when the coalition has agreed to govern jointly before the election. 2. A party s contribution of voting weight to a governing coalition will produce a bonus in its portfolio allocation that is, relative to what the party would get

12 based on its seat share only when the coalition has not negotiated before the election (in governments formed through post-electoral bargaining). 3. A party s proposer advantage will produce a bonus in its portfolio share only when the coalition has not negotiated before the election in governments formed through post-electoral bargaining. Data I use an original dataset of portfolio allocations in all countries with multi-party governments considered at least Partly Free by Freedom House during elections between 1990 and 2007 (48 countries in all). I restrict the sample in two ways to improve comparability and limit the effect of unmeasured effects. Following Carroll and Cox (2007), I restrict the sample to majority governments, since several of the arguments are predicated on majorities, to avoid the influence of support parties, and because minority governments are less comparable across regime types. For now, I also only focus on initial post-electoral governments to avoid differences in the definition of new governments. 6 I create three regime-type categories. The pure" parliamentary category is composed of regimes where the chief executive is subject to the confidence of the assembly and is not appointed by a partisan political actor. The presidential category includes all cases where the chief executive directly elected and is not subject to a parliamentary confidence. The semi-presidential term is used loosely here to denote the mixed category that includes directly and several indirectly-elected heads of state with the power to appoint a prime minister and the cabinet. Specifically, it includes all cases Shugart (2005) considers premierpresidential along with those parliamentary regimes Siaroff (2003) considers to have these 6 Both of these issues will be addressed in future versions using more sophisticated models.

13 appointment powers. 7 This is rough to be sure: de facto presidential influence in the cabinet appointment process varies considerably and is usually thought to be minimal, but these sources provide a reasonable categorical cutoff for cases where such influence on the formateur s bargaining position might be possible. The concepts discussed above are operationalized as the following variables. Seat contribution is simply a party s seats divided by the total seats of all government parties. Voting weight contribution is measured as a party s voting weight measured as the minimum integer weights calculated with Strauss (2003) algorithm, following Ansolabehere et al (2005) 8 divided by the combined voting weight of all governing parties. Formateur is a dummy variable indicating whether the party was regarded as having formed the government, almost always becoming the prime minister. The president is coded as formateur in presidential regimes. Pre-electoral alliances are defined here as governments whose members had committed to governing prior to election campaign in some explicit fashion. I have compiled a database of instances of explicit electoral cooperation from parties. The coding of electoral alliances relied on case literature or news sources acquired by the author. The coding criteria are similar to that of Golder (2006) and Carroll and Cox (2007). If a set of parties during an election campaign made an explicit public statement of intent to govern jointly, they were coded as a pre-electoral coalition (PEC). Cooperation was ruled out either when public pre-electoral statements clearly indicated a lack of commitment, multiple coalition options or nonalignment. Cases where alliances are reported to be technical, 7 These are Albania, Bulgaria, Croatia, Finland, France, Italy, Lithuania, Poland, Portugal, Romania, Slovakia, and Ukraine. 8 See Laver et al (2008) for an extensive discussion of this measure s use and properties.

14 localized and otherwise non-binding for governing purposes are not included, nor are cases of implicit cooperation or merely expected post-electoral partnerships. For present purposes of evaluating previous results in light of institutional variation, I follow the basic approach of Carroll and Cox (2007) and Warwick and Druckman (2006) in using an ordinary least squares regression as follows: Portfolio share = Seat + 2 Weight + 3 Formateur +. A proportional allocation implies a 1 indistinguishable from 1, with 0 of 0, while 2 and 3 reflect any benefits, above that predicted by seat contribution, obtained by post-electoral bargaining advantages of having a greater voting weight and the ability to propose the government, respectively. Interactions of seats, weight and formateur status with the categorical dummies are used to establish the conditioning effects of regime type and alliance each allocation predictor. Results Table 1 presents regression results designed to differentiate the contribution/advantage effects by regime type, pooling regime types and including an interaction with each regime-type dummy. The coefficients are plotted in Figure 2. The results are generally intuitive. First, in parliamentary regimes, the seat contribution coefficient is not entirely consistent with a Gamson s Law prediction of 1, a less proportionate result than that seen in previous literature though the dominant predictor nonetheless. The seat coefficient in the semi-presidential category is indistinguishable from the parliamentary group while it is lower in the presidential group (.6). Voting weight provides a significant if small additional effect in parliamentary regimes, but has no effect in the other two groups. Consistent with previous findings, there is no formateur effect in

15 parliamentary regimes generally. Not surprisingly, the formateur effect in presidential regimes is large, at about.2., though perhaps smaller than what one might expect. More interesting is that a formateur bonus of about.07 exists in the semi-presidential category. This is consistent with the possibility that formateurs in systems with presidential cabinet appointments are more likely to have some advantaged position in making proposals. [Table 1 Here] The effect of pre-electoral coalitions within each regime type is examined in Table 2. Here the PEC dummy indicating that the government was based on a pre-electoral pact is interacted with the Seat, Weight and Formateur variables. In general, the consistency in the direction of each effect across the samples is noteworthy. The basic differences for each effect between the PEC and post-electoral governments are quite similar across regimes types. The parliamentary and semi-presidential results in this sample are consistent with the expectations about pre-electoral coalitions. The coefficient on Seat is statistically indistinct from 1 in pre-electoral coalitions, but significantly lower in post-electoral governments. The presidential sample, while far less proportional on average, is also more seat-based in the pre-electoral coalitions (.75 versus.49). The effect of voting weight is also largely consistent with the electoral constraints expectations. There is a positive and significant effect for Weight in both the parliamentary and semi-presidential groups for post-electoral governments. The weight effect is not present in pre-electoral governments in parliamentary regimes. The voting weight effect is also reduced to a mean of about zero in semi-presidential PEC governments, though the degree of variance is high enough that the difference is not statistically significant. This pattern is similar in presidential regimes (positive in post-electoral governments, no effect in PEC governments), though not significant in either case.

16 [Table 2 Here] Finally, there is no formateur effect in parliamentary post-electoral governments, but a negative effect in the PEC governments. In the semi-presidential sample, there is formateur effect among post-electoral governments, an effect reduced to zero in the preelectoral governments. In presidential regimes, the formateur effect (or presidential bonus) is quite strong in post-electoral governments (.26), but significantly weaker in pre-electoral coalitions (.13). One possible explanation for this is that presidents in the PEC governments are simply appointing more independents rather than their own partisans. Model 4 tests this question by rerunning the same sample using an alternate measure of portfolio share in which the percent held by non-partisans is added to that of the president s party, rather than excluded. With this operationalization where the presidential share is both partisans and technocrats the results are similar: the presidential bonus is (naturally) stronger, but remains substantially lower in the PEC sample. Conclusions The pres has higher formateur maybe not as much higher as one would expect. Initial allocations in presidential regimes with electoral alliances are constrained in the reflection of post-electoral power, either in pure appointment power or voting weight. Post-electoral bargaining is often a foregone conclusion. I particularly highlight the importance of this in presidential regimes, since these institutions are sometimes accused of providing an insufficient basis for intra-branch cooperation. Beyond the fact that forming statutory majorities provides some incentive for forming stable coalitions, much of the bargaining in presidential regimes is not post-electoral at all: governing partners tend to begin cooperation earlier built around electoral goals.

17 Presidential regimes might be insulated from pre-electoral partners depleting their cabinet appointment resources because the number of ministries in these democracies is more easily inflated to satisfy partners demands due to less institutionalized ministerial jurisdictions -- thereby allowing partners what they perceive to be a fair allocation in absolute numbers even while allowing the president a larger share.

18 References Amorim Neto, O "Presidential Cabinets, Electoral Cycles and Coalition Discipline in Brazil." In Legislative Politics in Latin America, ed. Scott Morgenstern and Benito Nacif. New York: Cambridge University Press, Amorim Neto, O "Breaking the Parliamentary Chain of Delegation: Presidents and Non-Partisan Cabinet Members in European Democracies." British Journal of Political Science 36 : Amorim Neto, O., G.W. Cox, and M.D. McCubbins "Agenda Power in Brazil's Camara Dos Deputados, " World Politics 55 (4): Amorim Neto, O., and F. Santos "The Executive Connection: Patronage and Party Discipline in Brazil." Party Politics 7 (2). Amorim Neto, O., and K. Strøm "The Presidential Calculus: Executive Policy Making and Cabinet Formation in the Americas." Comparative Political Studies 39 (4): 415. Ansolabehere, S., J.M. Snyder, A.B. Strauss, and M.M. Ting "Voting Weights and Formateur Advantages in the Formation of Coalition Governments." American Journal of Political Science 49 (3): Baron, D.P., and J.A. Ferejohn "Bargaining in Legislatures." American Political Science Review 83 (4): Carroll, R., and G.W. Cox "The Logic of Gamson's Law: Pre-Election Coalitions and Portfolio Allocations." American Journal of Political Science 51 (2): Carroll, R., G.W. Cox, and M. Pachon "How Parties Create Electoral Democracy, Chapter 2." Legislative Studies Quarterly 31 (2): Cheibub, J.A., A. Przeworski, and S.M. Saiegh "Government Coalitions and Legislative Success Under Presidentialism and Parliamentarism." British Journal of Political Science 34 (04): Debus, Marc Pre-Electoral Commitments and Government Formation. Public Choice 138 Deheza, G.I "Gobiernos de coalicion en el sistema presidencial: America del Sur." PhD Dissertation. European University Institute, Florence.

19 Druckman, J.N., and A. Roberts "Context and Coalition-Bargaining - Comparing Portfolio Allocation in Eastern and Western Europe." Party Politics 11 (5): Druckman, J.N., and P.V. Warwick "The Missing Piece: Measuring Portfolio Salience in Western European Parliamentary Democracies." European Journal of Political Research 44 (1): Gamson, W.A "A Theory of Coalition Formation." American Sociological Review 26 (3): Golder, S.N "Pre-Electoral Coalitions in Comparative Perspective: A Test of Existing Hypotheses." Electoral Studies 24 (4): Golder, S.N "Pre-Electoral Coalition Formation in Parliamentary Democracies." British Journal of Political Science 36 : Golder, S.N The Logic of Pre-Electoral Coalition Formation. Columbus: Ohio State University Press. Laver, M., and N. Schofield Multiparty Government. Oxford University Press New York. Morelli, M "Demand Competition and Policy Compromise in Legislative Bargaining." American Political Science Review 93 (4): Morelli, M., and M. Montero "The Demand Bargaining Set: General Characterization and Application to Majority Games." Games and Economic Behavior 42 (1): Schofield, N "Kernel and Payoffs in European Government Coalitions." Public Choice 26 : Schofield, N., and M. Laver "Bargaining Theory and Portfolio Payoffs in European Coalition Governments " British Journal of Political Science 15 (APR): Warwick, P.V., and J.N. Druckman "Portfolio Salience and the Proportionality of Payoffs in Coalition Governments." British Journal of Political Science 31 (04):

20 Figure 1: Seat Contribution and Portfolio Allocations across Regime Type, Portfolio Share.6.4 Share Seat Contribution Parliamentary Presidential Semi-Presidential

21 95% Confidence Intervals Figure 2: Voting Weight, Formateur and Seat effects in Parliamentary, Semipresidential and Presidential Regimes, , OLS Coefficients Parliamentary Semi-Pres. Presidential Formateur Voting Weight Contribution Seat Contribution Formateur Voting Weight Contribution Seat Contribution Formateur Voting Weight Contribution Seat Contribution ( 0.07, 0.27) 0.12 ( 0.00, 0.25) 0.57 ( 0.42, 0.73) 0.07 (-0.04, 0.19) (-0.24, 0.20) 0.85 ( 0.58, 1.13) 0.05 ( 0.02, 0.08) 0.21 ( 0.13, 0.29) 0.57 ( 0.50, 0.64)

22 Table 1: Portfolio Distribution across Regime Types , OLS PORTFOLIO SHARE Presidential (0.0128) Semi-Presidential (0.0150) Seat Contribution 0.741*** (0.0412) Semi. Seat Contribution (0.126) Pres Seat Contribution ** (0.0781) Voting Weight 0.157*** (0.0475) Semi Voting Weight (0.0839) Pres Voting Weight ** (0.0503) Formateur (0.0176) Semi-Presidential Formateur ** (0.0322) Presidential Formateur 0.192*** (0.0237) Constant *** ( ) Observations 442 R-squared Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1

23 Table 2: Influence of Pre-electoral Pacts on Portfolio Distributions, in Parliamentary, Semi-presidential and Presidential Regimes, , OLS (1) (2) (3) (4) Parliamentary Regimes Semi- Presidential (parliamentary with partisan president) Presidential Regimes (partisan portfolios only) Presidential (independents as president s share) PEC ** (0.0195) (0.0190) (0.0335) (0.0365) Seat Contribution 0.789*** 0.627*** 0.497*** 0.353*** (0.0538) (0.0687) (0.0731) (0.0797) Seat Contribution PEC 0.355*** 0.391** 0.256* 0.356** (0.127) (0.162) (0.150) (0.164) Voting Weight Contr ** 0.149** (0.0558) (0.0728) (0.0764) (0.0833) Voting Weight Contr. PEC ** (0.0940) (0.180) (0.148) (0.161) Formateur *** 0.264*** 0.436*** (0.0150) (0.0233) (0.0301) (0.0328) Formateur PEC ** ** ** *** (0.0573) (0.0506) (0.0631) (0.0687) Constant *** *** *** *** ( ) (0.0109) (0.0163) (0.0177) Observations R-squared Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1