Tag Archives: Coalitions

Carlos Pereira, Mariana Batista, Sérgio Praça and Felix Lopez – How Presidents Monitor Coalitions in Brazil’s Multiparty Presidential Regime

This is a guest post from Mariana Batista based on her recent article with Carlos Pereira, Sérgio Praça and Felix Lopez,  ‘Watchdogs in Our Midst: How Presidents Monitor Coalitions in Brazil’s Multiparty Presidential Regime’, published in the Fall edition of Latin American Politics and Society. The full article can be found here.

In “Watchdogs in Our Midst: How Presidents Monitor Coalitions in Brazil’s Multiparty Presidential Regime” we analyze coalition politics from the perspective of what happens after government formation or what are the president’s strategies to manage “a government of strangers” (Heclo, 2011).

We argue that presidents in multiparty settings deal with the fundamental dilemma of delegating power to coalition partners while minimizing the risk of policy drift. Cabinet positions are the main currency of coalition politics and a fundamental part of coalition formation and survival. However, when trusting cabinet positions to coalition partners, the president runs the risk of being expropriated by their cabinet. There are some mechanisms to minimize the risk of expropriation in coalition governments such as coalition agreements, inner cabinets, centralized screening, and legislative oversight. In our article, we explore the strategy to reduce policy drift based on the appointment of junior ministers.

Junior ministers are the second in command in a ministry and may act as watchdogs on behalf of the president. When presidents cannot “choose whom to trust” (Martinez-Gallardo and Schleiter, 2015) they still may use their appointment powers to appoint a junior minister loyal to their preferences. By doing so, the president will have eyes and ears inside the ministry, even though a coalition partner is in control. This is a powerful way to “keep tabs on partners” (Thies, 2001) while holding the coalition together.

The role of junior ministers in the monitoring of coalition partners is a topic explored in parliamentary regimes, but not in the presidential setting where the president is the one at the top of the hierarchy. To analyze the presidents’ appointment strategies we focus in Brazil as a case study in the period from 1995 to 2010, exploring the partnering between ministers and junior ministers. We consider a junior minister a watchdog when the junior minister is not aligned to the minister. This may happen when the president appoints a junior minister from a different party or when the junior minister is a career bureaucrat. In these situations we expect the junior minister to be loyal to the president and to report on the ministers’ doings.

Figure 1 shows that presidents have the options of appointing 1) ministers from their party (PP), 2) from a coalition partner (CP), or 3) non-partisan ministers (NP). Non-partisan ministers are aligned with the president’s preferences by definition. However, partisan ministers have policy preferences of their own that may jeopardize the president’s agenda. For this reason, these are the ones that the president considers to monitor. Figure 1 shows that partisan ministers are monitored with the appointment of junior ministers. However, ministers from the other coalition parties are monitored more frequently.

Figure 1: Portfolio Allocation and Monitoring Through Junior Ministers, 1995–2010

Considering that appointing a watchdog is a direct control over the minister, the president will not implement this strategy indiscriminately as shown above. We expect that watchdogs will be used only when the costs of the delegation are high. We argue that these costs may be captured by three variables: ideological distance as a proxy for preference distance, portfolio salience, and the coalescence rate as a proxy for the degree of the coalition agreement.

We expected that the greater the ideological distance, the greater the probability of appointing a watchdog because ideological distance would represent preference divergence between the president and the minister. Knowing that the minister is not to be trusted, the president would appoint a hostile junior minister to keep control from the inside. Also, we expected that the most important ministries would be monitored closely with the appointment of watchdogs because the stakes are high. So, the greater the portfolio salience, the greater the probability of a watchdog. Lastly, we expected that the greater the coalescence rate, the smaller the probability of a watchdog because the coalescence would be a measure of the degree of the coalition agreement. This is especially important in presidential systems because there is evidence that coalitions reach some very different arrangements regarding the distribution of portfolios and the amount of power coalition partners will have in government (Amorim Neto, 2006). We expected that the greater this agreement, the smaller the incentives for coalition monitoring.

The results indicate that only ideological distance is important to explain the appointment of a watchdog junior minister, indicating that when policy preferences between the president and the minister are not aligned, the president will try to minimize agency losses and risks of policy drift by appointing a trusted junior minister. Figure 2 shows this relationship.

Figure 2: Predicted Probability Logistic Regression (with Controls): Ideological Distance (95 percent CIs

For a minister from the president’s party (an ideological distance of 0), the predicted probability of a watchdog junior minister to be appointed is 0.64. The predicted probability increases to 0.81 when the ideological distance between the minister and the president is 2, and to 0.95 when the ideological distance reaches 4.5, the maximum value in our distribution.

Analyzing the president’s monitoring strategies concerning coalition partners in Brazil, our main result is that the greater the ideological distance, the greater the probability of monitoring. Although we specifically investigate the political dilemma that Brazilian presidents have faced deciding how to monitor coalition partners, we hope that the particular results presented could travel well and extend to other multiparty presidential regimes elsewhere. We also expect that this discussion will increase interest in what happens after coalition formation or how coalitions actually govern in presidential systems.


Amorim Neto, Octavio. Presidencialismo e governabilidade nas Américas. FGV Editora, 2006.

Heclo, Hugh. A government of strangers: Executive politics in Washington. Brookings Institution Press, 2011.

Martínez-Gallardo, Cecilia, and Petra Schleiter. “Choosing whom to trust: Agency risks and cabinet partisanship in presidential democracies.” Comparative Political Studies 48.2 (2015): 231-264.

Thies, Michael F. “Keeping tabs on partners: The logic of delegation in coalition governments.” American Journal of Political Science (2001): 580-598.

Marisa Kellam – Why Pre-Electoral Coalitions in Presidential Systems?

This is a guest post by Marisa Kellam, Associate Professor, Waseda University. It is based on her recent article in the British Journal of Political Science.

Presidential politics goes hand in hand with coalitional politics in Latin America, especially in South America. As recently reported in this blog, presidents in the region often depend on the support of other parties to win election and to govern.

In this post, I will focus on pre-electoral coalitions. [1] To give some recent examples: President Bachelet in Chile, President Santos in Colombia, and former Brazilian President Dilma Rousseff all ran for re-election with a multiparty electoral alliance.  Multiple parties also supported Argentine President Macri’s candidacy. In recent elections in other countries, the incumbent party candidate defeated an opposition pre-electoral coalition, such as in Ecuador’s recent election, Bolivia’s 2014 presidential contest, or the 2013 presidential election in Venezuela.

In fact, pre-electoral coalitions in presidential elections have been a feature of Latin American democracy since the third-wave, and even before. Yet the conventional wisdom has been that these coalitions were “not binding past election day.” [2] However, increasing attention to post-electoral coalition formation in comparative presidentialism research has led to new findings that winning pre-electoral coalitions usually go on to form post-electoral governing coalitions. [3] Does the strong empirical correspondence between pre-electoral coalitions and post-electoral governments call for a revision of the conventional wisdom? My recently published article in the British Journal of Political Science speaks to this puzzle.

Why do parties form pre-electoral coalitions in presidential systems? From the perspective of a presidential candidate, it would seem to be an easy answer—the more in my camp, the merrier—that is, unless she must give something in return. When considering potential partner parties, we might assume that the presidential candidate offers them government positions—just as presidents offer coalition partners in government negotiations—except that pre-electoral agreements involve only promises not actual offers.

Although I set out to overturn the conventional wisdom on pre-electoral coalitions, I found no convincing argument to support a contrary claim that presidential candidates’ promises to distribute government positions and resources to other parties are credible commitments in presidential systems. Presidents alone control cabinet appointments—even their own parties cannot hold presidents immediately accountable. Moreover, presidents’ partners will not necessarily punish them for breaking their pre-electoral commitments. A party that wants access to resources under the president’s control is unlikely to make a loud complaint, much less to pull out of the government completely.  And if parties do not reveal the extent to which presidents fail to honor agreements to share spoils, then neither presidents nor their parties will pay a reputational cost. This isn’t to say that presidents will break their promises; it is only to make the point that candidates’ pre-electoral promises to share spoils are “cheap talk” and party leaders know this.

I find it useful to contrast these behind the door negotiations with presidential candidates’ public campaigns.  A presidential candidate and her political party pays an immediate reputational cost if she publically campaigns on a policy compromise made with another party in order to gain its support. True, a president is not bound to her campaign platforms. But even so, the pre-electoral policy agreement reveals information about her policy positions.  A pre-electoral policy agreement also gives the president and her partners a shared mandate, or common purpose, after the election. And if the president ends up reneging on that policy agreement later, the coalition partner would likely refuse to go along, consistent with their own electoral incentives and policy motivations.

These differences between patronage promises and campaign platforms provide some insight as to why parties join pre-electoral coalitions to support other parties’ candidates in presidential systems. Political parties join pre-electoral coalitions in pursuit of policy goals, but not as part of an office-seeking strategy.  To provide empirical evidence to support this argument, I compare characteristics of the parties that joined pre-electoral coalitions with those that did not.

More specifically, I compare the probability of participation in pre-electoral coalitions of programmatic parties to that of particularistic parties.  Particularistic parties are those that experts classify as having no discernible policy position on the standard, left-right macroeconomic dimension of politics; instead, particularistic parties focus on the distribution of “pork” and patronage or serve single-interests.  According to my reasoning, if these parties do not have policy goals then they should be less likely to join pre-electoral coalitions (unless one of their own members is on the president-VP ticket). Programmatic parties, in contrast, may use pre-electoral coalitions to identify and help elect presidential candidates who are closest to them in terms of policy.

I analyzed coalitions formed, and not formed, in 77 elections held in 11 Latin American countries.  I found that programmatic political parties (i.e. policy-seeking parties) were more likely than particularistic political parties (i.e. office-seeking parties) to join pre-electoral coalitions in support of another party’s presidential candidate.  As expected, I also found that the greater the ideological distance between a programmatic party and the party of a presidential candidate, the less likely they are to join that candidate’s coalition.

While the once conventional thinking that presidents have little incentive to form governing coalitions has been overturned, this does not imply that the conventional wisdom regarding electoral coalitions should also be cast aside. As I have discussed, pre-electoral coalition bargaining differs from post-electoral government negotiations, with important implications for presidential politics in multiparty systems.

In conducting this research, I realized that the reason why parties join pre-electoral coalitions in presidential systems is less obvious than it appeared at first glance. Even if pre-electoral coalitions are not binding commitments to govern together after the election, the coalition formation process itself informs political parties about their respective policy positions and creates a shared mandate.



[[1]] On a side note, the prefix “pre” seems unnecessary to me, but I use it nonetheless because the term pre-electoral coalitions is widely used in the literature.

[2] Mainwaring, Scott, and Mathew Shugart. 1997. Presidentialism and Democracy in Latin America. Cambridge UP, p. 397.

[3] As discussed previously in this blog, see Freudenreich, Johannes. 2016. “The Formation of Cabinet Coalitions in Presidential Systems.” Latin American Politics and Society 58(4): 80-102.  In my own work-in-progress with Cecilia Martinez-Gallardo, we also find a strong empirical relationship between pre- and post-electoral coalitions in Latin America.

Fernando Meireles – Latin American presidents and their oversized government coalitions

This is a guest post by Fernando Meireles, Ph.D candidate in Political Science at Federal University of Minas Gerais (Brazil). E-mail: fmeireles@ufmg.br

In many countries, presidents have a difficult time governing because their parties lack a legislative majority. In fact, because of the combination of separate elections for executive and legislative branches with multiparty systems, this situation is far from uncommon: during the last two decades in all 18 Latin American countries with presidential systems, only 26% of the time has the president’s party had a majority in the lower house. Due to this constraint, as a vast amount of research now highlights, minority presidents usually form multiparty government coalitions by including other parties in their cabinets. Again, only four Latin American presidential countries in the last twenty years were not governed by a multiparty coalition at some point since the 1980s.

However, the need to craft a legislative majority alone does not explain why presidents frequently include more parties in their governments than necessary to obtain a minimum winning coalition – forming what I call an oversized government coalition. The distribution of this type of coalition in Latin America is shown in the graph below. As can be seen, it is not a rare phenomenon.

If government coalitions are costly to maintain, as presidents have to keep tabs on their coalition partners to ensure they are not exploiting their portfolios to their own advantage – not to mention the fact that by splitting spoils and resources between coalition partners, the president’s own party is worse off – then why are these oversized coalitions prevalent in some Latin American countries?

In a recent article in Brazilian Political Science Review, I tackled this puzzle by analyzing the emergence of oversized government coalitions in all 18 presidential countries in Latin America[1], followed by a case study focusing on Brazil, spanning from 1979 to 2012. To this end, I gathered data on cabinet composition[2] from several sources to calculate the size of each government coalition in the sample: if a coalition had at least one party that could be removed without hampering the majority status of the government in the lower house in a given year, I classified it as an oversized coalition.

Specifically, I examined three main factors that, according to previous research, should incentivize presidents to include more parties in their coalitions than necessary to ensure majority support: 1) the motivation party leaders have to maximize votes, which would make joining the government attractive to opposition parties (vote-seeking); 2) the motivation presidents have to avoid coalition defections to implement their policy agendas (policy-seeking); and 3) the institutional context, considering the effects of bicameralism, qualified majority rules, and party system format on government coalition size.

The results support some of the hypothesis suggested by the literature. First, presidents are more prone to form oversized coalitions at the beginning of their terms, which shows that the proximity to the election affects Latin American presidents’ decision to form, and opposition parties to accept being part of, large coalitions – as others studies argue, this is mainly due to parties defecting from a coalition to present themselves as opposition when elections are approaching. Second, party fragmentation also has a positive effect on the emergence of oversized coalitions, consistent with the hypothesis that presidents might include additional parties in their coalitions anticipating legislative defections. Yet on the other hand, presidential approval, party discipline, and ideological polarization do not have the same positive effects on the probability of an oversized coalition being formed.

The factor that has the most impact on the occurrence of oversized coalitions, however, is the legislative powers of the president. As the literature points out, legislative decrees and urgency bills could be used by skilled presidents to coordinate their coalitions, facilitating horizontal bargaining between coalition partners. The comparative results show that this is the case in Latin America: the difference in the predicted probability of a president with maximum legislative powers in the sample forming an oversized coalition and another with minimum powers is about 32 percent points.

By exploring the Brazilian case in more depth, I also found that bicameralism dynamics and qualified majority rules impact the emergence of oversized coalitions. With two chambers elected through different electoral rules, parties in Brazil are often unable to secure the same seat share in both houses; to make things worse for presidents, party switching is still widespread in the country. In this context, as my results uncovered, differences in the number of seats controlled by the government in the Chamber of Deputies and the Senate positively affect the emergence of oversized coalitions. Finally, as some bills require supermajorities to be approved, such as constitutional amendments, reformist presidents also tend to form and maintain larger coalitions: the maximum value in this variable predicts increases by up to 10 percentage points on the probability of an oversized coalition being formed.

Taken together, these results show a more nuanced picture of why and how presidents form multiparty government coalitions in Latin America: often, obtaining a legislative majority is not enough to implement their legislative agendas, and so they might resort to a complementary strategy: to form larger coalitions. And presidents with greater legislative power, at the beginning of their terms or facing fragmented party systems, are in the best position to pursue such a strategy. In this way, both electoral and programmatic factors, as well as the institutional context, become key to understand variations in the size and the composition of government coalitions in presidential countries.


[1] These countries are Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, El Salvador, Ecuador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Dominican Republic, Uruguay, and Venezuela.

[2] The criteria employed to identify a government coalition is the party affiliation of the ministers of the principal ministerial portfolios in each country – taking into account that ministers are not always recruited due to their connections or their congressional influence, and that in some cases they are not recognized by their parties as legitimate representatives of the same.

Scott Morgenstern – Coalitions and presidentialism

This is a guest post by Scott Morgenstern from the Department of Political Science at the University of Pittsburgh


At least since Shugart and Carey, debates moved the simple distinction between presidentialism and parliamentarism towards a focus on the factors that distinguish among presidential regimes. Presidential power (formal and partisan) has been a central focus of these newer debates, and the shape and functions of coalitions has also become a central interest. How do these two variables interact?

In this post, I first provide a short discussion of the definition of coalitions and then focus on just two themes. First, while it is simple to notice that strong presidents have less need for congressional coalitions, it may be less evident that the type of opposition also matters to this equation. This leads to the conclusion that the impact of presidential power on the coalition politics is dependent on the legislative type. Second, not all presidential powers have the same impact on coalition formation, and thus broad indices of power may not be indicative of the relationship. The sub-conclusion of that discussion is that the partial veto is particularly detrimental to legislative coalitions.

In presidential systems, the term “coalition” has at least three meanings. It can imply that parties come together during an election to support a particular candidate, b) that the cabinet is composed of a multiple parties, or c) that legislative parties join together to support a particular policy. Cabinet coalitions are still an understudied phenomenon in presidential regimes, but an important theme in the studies (e.g. Amorim Neto) is whether the president is able to buy (or rent) support by offering a party membership in the cabinet. In some countries (eg Chile) this works better than in others (eg Brazil), but a general conclusion is that cabinet membership augments, but does not guarantee legislative support. For that reason, in this note I focus on the propensity of presidents to negotiate with legislative parties on policy coalitions.

In a paper with Gary Cox (2001), I argued that presidents have alternative strategies, and they choose them based on legislative “types.” If the legislature is “workable” then presidents should seek to build coalitions and make compromises to achieve their policy goals. Imperial presidents, however, have no need to negotiate with the legislature. This suggests that presidential powers, at least in part, are inversely related to coalition-building.

The relation is incomplete, however, because legislative types are also actors in the relationship. Parties (or their component legislators) are interested in joining coalitions for either access to power, resources and perks, or to influence policy. As Strom (1990) and others have shown (see also Laver & Shepsle 1990), parties may have more influence on policy from outside the coalition.

A number of studies (note the recent paper and post to this blog by the Doyle and Elgie) have sought ways to measure the totality of presidential powers. Clearly strong powers generally influence the likelihood of a president needing the cooperation of the legislature. President who can impose their will by decree, for example, have little need to compromise with recalcitrant legislators. Other powers that are common parts of power indices, however, have less (or at least less direct) impact on the likelihood of inter-branch cooperation.

As intended by the US founders, the veto power and its override provisions, for example, can encourage cooperation. Veto provisions vary in terms of override provisions and what Aleman and Tsebelis (2004) call “amendatory” provisions. Many countries (and US governors) also have some form of partial veto, and its provisions can destroy incentives for inter-branch cooperation. While it may help a president to limit pork from some budgets, it has a sharp negative impact on inter-partisan deal-making. If the president can undo a logroll deal that led two parties agree to support a bill, then the parties will not join in the first place. Omnibus budget bills in the United States, for example, join legislators from across the political spectrum, each supporting the bill in exchange for particularistic payoffs. If the president can excise the pieces of the bill that he or she finds objectionable, legislators will be reluctant to join the coalition.

The use of the partial veto presents interesting opportunities for further study. Llanos’ (2001) study of presidential power in Argentina, for example, suggested that while President Menem used the partial veto to improve his ideological position, he did not undo all legislative bargains. A comparative study could push this idea to illuminate the degree to which the partial veto hinders coalition formation.

Scott Morgenstern is Director of the Center for Latin American Studies at the University of Pittsburgh. His research focuses on political parties, electoral systems, and legislatures, with a regional specialization in Latin America. Among his publications are Patterns of Legislative Politics: Roll Call Voting in the United States and Latin America’s Southern Cone (Cambridge University Press, 2004), Legislative Politics in Latin America, (coeditor and contributor; Cambridge University Press, 2002), and Pathways to Power (coeditor and contributor, Pennsylvania State University Press, 2008). He is just finishing a book “Are Politics Local? The Two Dimensions of Party Nationalization around the World’. His articles have appeared in the Journal of Politics, Comparative Political Studies; Comparative Politics, Party Politics, Electoral Studies, Review of International Political Economy, and other journals. He was also the primary investigator on a grant from the USAID to produce documents related to their political party development programs.

Indonesia – With preliminary results now in, what lies ahead?

The General Election Commission (KPU) announced recently that the need for the revotes in over 1000 polling stations may delay the official results for the legislative elections, held April 9, 2014, original pegged for May 9. Nevertheless, even without the official results, political jockeying between the parties for coalition-partners and the winnable president-vice president team has commenced, and these may be as intense as the finger-pointing and blame-game that has taken place following the quick count results.

What coalitions are possible and which improbable? Coalitions are necessary given that the quick-count results show none of the parties has received the requisite 25 percent of the national vote or 20 percent of the parliamentary seats to field independently a presidential candidate for the July elections. The top three vote-getters based on the quick-count results are the PDI-P, Indonesian Democratic Party of Struggle, leading with about 19 percent of the popular vote; Golkar with 14.9 percent; and Gerindra with 12 percent. President Susilo Bambang Yudhoyono’s Democratic Party edged into the double-digit league with 10 percent of the votes.

The top three parties have all announced their presidential candidates:

  • Jakarta governor, Joko “Jokowi” Widodo for the PDI-P;
  • Aburizal Bakrie, business tycoon, for Golkar, the leading party of the Suharto era;
  • Prabowo Subianto, former military lieutenant-general, for Gerindra

Coalition-partners, then, are angling for vice-presidential nomination or cabinet positions.

Polls place Jokowi as the one to beat for the presidential elections but, of course, polls have gotten things wrong before: witness the poll expectations of 30 percent popular vote for the PDI-P for the April elections, and the corresponding pummeling expected for the Islamic parties.

Still, the PDI-P was the first to announce a coalition with NasDem, the National Democrat Party, founded by former Golkar Party member and media mogul Surya Paloh. NasDem is the only new party to be sanctioned by the General Election Commission (KPU) to contest the national elections and received 6.6 percent of the popular vote. NasDem favors former Vice President and Golkar Party member, Jusuf Kolla, as the vice-presidential nominee on the PDI-P coalition ticket, but Jokowi has maintained that nominations and cabinet posts will not be traded for coalition support. With NasDem’s support, PDI-P has passed the threshold for the nomination. Since the coalition with NasDem, the PDI-P has announced a coalition deal with the National Awakening Party, the PKB, founded by former Indonesian President Abdurrahman Wahid, although the family of “Gus Dur” has severed ties with the PKB. Talks are reportedly ongoing with another potential coalition partner, the United Development Party (PPP).

The infighting in PPP became almost a sport to watch following the quick-count results. Internal rifts in the party surged to the surface with senior party members voicing their disapproval of the close association of PPP chair, Suryadharma Ali, with Gerindra. In the face of the opposition and growing pressure to step down, Suryadharma Ali pre-empted his detractors by firing them from the party; he also declared his party’s support for Gerindra’s Prabowo Subianto. The party elders followed with their own announcement of Suryadharma Ali’s dismissal from the party. Fortunately for party members, the tit-for-tat retaliations have ended and the party announced a mending of the rifts; less fortunately for Gerindra, as part of the reconciliation, the PPP withdrew endorsement of Pubrabo Subianto and will decide on a candidate to support at the national meeting in early May.

Both Gerindra and Golkar have yet to announce coalition partners, although Gerindra has been in talks with the Islamic Parties, including the Prosperous Justice Party, PKS, as well as the Hanura Party. The PKS was caught in a sex-and-corruption scandal in 2013 that has seen its president jailed and other party elders at risk for similar penalties. Yet, the party lost only about 1 percent of popular support from the previous election. The PKS maintains that Islamic parties need to support a presidential candidate with a “high level of piety.”

Meanwhile, the possible tie-up between Gerindra and Hanura may spell trouble for Golkar and its presidential nominee, Aburizal Bakrie. Golkar has insisted that it will offer up only a presidential candidate, not a running mate; however, with PDI-P’s increasing coalition partners, the list of potential partners is quickly diminishing. Correspondingly, prospects that rivals will oust Aburizal Bakrie for control of the party is increasing.

What coalitions are improbable? At the least, it is clear that a coalition of Islamic parties is not in the cards, notwithstanding the hopes of the many prominent clerics who gathered together to push for that coalition. Likewise, although President Susilo Bambang Yudhoyono’s Democrat Party is pushing ahead to name its own presidential nominee, it is not clear that it will be able to bring together a coalition large enough to support that nomination. It may have to face up to being in the opposition.