Coalition Presidentialism in Bicameral Congresses : How does the Control of a Bicameral Majority Affect Coalition Survival ?

Presidential legislative majorities under bicameral congresses have seldom been considered in the literature on coalition cabinets. However, this combination is far from being an isolated or irrelevant topic. Bicameral congresses comprise a double round of negotiation for the executive, increasing the shadow of the unexpected. Indeed, controlling one of the two chambers may not be sufficient for a president to guarantee the approval of policies. In this case, what if a government does not control both houses? How are coalition cabinets affected by the non-control of one or both chambers? I will focus analysis on the 25 cases of coalition cabinets in Latin America since the return of democracy and present a bicameral framework. Through mvQCA, I find that, contrary to common belief, the controlling of a bicameral majority is neither necessary nor sufficient to ensure stable coalitions.

e0002 -2/27 nlike the study of coalition formation, explaining the termination/survival of coalition cabinets is quite a recent topic in the extensive literature on coalition theories.Nevertheless, following the dominant trends in political science, several scholars analyzing coalitions in parliamentary regimes have recently considered behavioural, institutional, administrative and heterogeneous conditions to predict the likelihood of coalition partners walking away, or not (CHIRU, 2015;DAMGAARD, 2008;LUPIA and STROM, 2008;SAALFELD, 2008, among others).
Literature on coalition cabinets under presidential regimes reveals that coalition dissolution has, also, been received some attention.However, the extension and diversity of these studies is still quite limited and mostly relate to institutional conditions (ALBALA, 2016).Thus, the most recurrent condition pointed to by scholars, most of whom are from Latin America, deals with one presidential feature: the principle of mandate fixity 1 (ALTMAN, 2000;CHASQUETTI, 2008;GARRIDO, 2003;MARTÍNEZ-GALLARDO, 2012).
Furthermore, no consideration has ever been bestowed to another presidential particularity: the formation and holding of presidential majorities under bicameral systems.More precisely, how the controlling -or not -of a bicameral majority affects the survival of coalition cabinets.This also goes for parliamentary studies, as very few studies, with inconsistent results, have considered the possible impact of controlling two chambers on coalition survival (DIERMEIER et al., 2007;DRUCKMAN and THIES, 2002).
This paucity is quite surprising when considering that bicameralism is not an uncommon feature, especially in Latin America.Moreover, in presidential polities, bicameralism is usually symmetric, that is both chambers are used to sharing similar powers, attributes and influence (LLANOS, 2003;NEIVA, 2006).
Thus, symmetrical bicameralism constitutes one of the most important differences between parliamentary and many Latin American presidential systems.This feature entails a two-round procedure in the policymaking process for the president, thus increasing what Lupia and Strøm (2008) call "the shadow of the unexpected".
Hence, if a president controls one of the two chambers, or even both, this may not be sufficient to ensure that policies are approved, nor to guarantee coalition survival _____________________________________________________________________________________________________ 1 Linz (1990), Cheibub (2007) and Cheibub et al. (2014) have pointed out that the fixity of the presidential mandate (i.e., the inauguration and conclusion date being constitutionally settled) constitutes one of the most salient and distinctive features of presidentialism in relation to parliamentarism.
However, recent events in Brazil have shown that even a president (Dilma Rousseff) enjoying a bicameral majority is not enough to maintain his/her coalition.
Therefore, this unprecedented topic, which at first might sound trivial, finally appears to be worthy of study.
Hence, this article enhances coalition theories on presidential regimes by including, for the first time, the bicameral majority issue.My objective is to answer to a central question: In what way is the controlling of one, two or neither chambers a 'relevant condition for the survival/dissolution of coalition cabinets in presidential regimes'?
Based on a sample of 25 past Latin American coalition cabinets in bicameral legislatures, I proceeded by testing several hypotheses using multi-value quality compared analysis (mvQCA), pointing out that the survival of a coalition agreement is not consistently related to the holding of a bicameral majority.

Bicameralism: an underexplored field
It appears that, until recently, political science has given little attention to bicameralism beyond coalition theories.This is surprising, as bicameralism is quite a common constitutional feature.As shown by Tsebelis and Money (1997), a third of all legislatures in the world comprise of two chambers.
Unfortunately, the concept of bicameralism has generally been reduced to that of a legislature composed of two chambers (HELLER, 2007;RUSSEL, 2013).Indeed, it is through this rather simplistic conception that several scholars have analyzed bicameralism, using the lens of 'veto players' theory as a potential process element that may motivate strategic behaviour.However, Tsebelis (2002) considers the two chambers as two potentially distinct veto players if -and only if -the two chambers are incongruent (i.e., the party contingents in the two chambers are not equal).Scholars who follow this line of thought emphasize that by increasing the number of players, bicameralism affects the policymaking process, creating delays or vetoes (DAHL, 2013;HALLERBERG, 2011;PEREZ-LIÑÁN and RODRIGUEZ-RAGA, 2009;VANDUSKY-ALLEN and HELLER, 2014).However, this generic conception hides considerable variations in types of bicameralism that require more specific analysis (CUTRONE and McCARTHY, 2008).
Hence, Lijphart (2012) carried out a ground-breaking typology of the different forms of bicameralism.Lijphart's approach distributes the different types of bicameralism around two main dimensions: 01.chamber symmetry/asymmetry; and 02.secondchamber legitimacy/representativeness.The first condition (the symmetry/asymmetry distinction) relates to determining whether two chambers have similar powers, attributes or influence in the policymaking process 2 .The second condition presupposes the election process of members of the upper chamber (usually called the 'Senate').
Lijphart (2012) shows that in many parliamentary systems senators are selected rather than elected 3 .
These considerations are assumed by Tsebelis and Money (1997, p. 211), who state that even weak upper chambers may veto or exercise influence whenever they are incongruent with lower chambers.
However, all these considerations apply mainly to Western parliamentary and semi-presidential regimes.In fact, despite the existence of a diversity of bicameral models under presidential polities, most share the same tendency whereby both chambers have formally comparable powers, attributes and influence, and the legitimacy of the chambers always relies on direct voting (LLANOS and NOLTE, 2006;REYNOSO, 2010;LÓPEZ, 2004).Since a president is accountable to neither of the two chambers (legitimacy having been conferred by the popular vote (CHEIBUB et al., 2014), he or she can choose which chamber to submit bills to first.In other words, bicameralism under presidentialism consists of two legislative bodies that have similar powers and influence.The major differences between the models of bicameralism are connected to the electoral process (e.g., district size for electing senators vs. deputies) and the duration of the mandate within each house 4 .
_____________________________________________________________________________________________________ 2 These considerations are distinct in Lijphart's words (2012).Indeed, although the German Bundesrat may not share the same attributes and powers as the Bundestag, its composition (the minister-president and cabinet ministers of each of the Länder) makes it very influential. 3The British House of Lords is a classic example of an unelected upper chamber. 4Generally, upper chambers have longer mandates and the conditions for membership are more restrictive.

Expected impacts of two chambers on coalition survival/ dissolution
As mentioned above, very few works on coalition theories have considered the potential impact of bicameralism on policymaking processes.The same goes for measuring the potential impact of two chambers on the survival/dissolution of coalition cabinets.Moreover, the rare studies that have done so have produced contradictory results.By continuation, I expose the few studies that have considered the impact of bicameral majorities on coalition cabinets, as well as their inconsistent findings.
The main theoretical considerations that have emerged from veto player theories are that, 'ceteris paribus', bicameralism may decrease government duration (TSEBELIS, 2002;TSEBELIS and MONEY, 1997).Incongruence between two chambers is supposed to conduce political gridlocks that can only be broken through tight compromise agreements between the two chambers (CUTRONE andMcCARTY, 2008, DAHL, 2013).Moreover, the existence of a second chamber supposes an increase in the costs for intraparty discipline, as it implies that every co-partisan in each chamber should follow the same party line irrespective of the socio-political context or congressmen's individual interests (VANDUSKY-ALLEN and HELLER, 2014).
According to this literature, bicameralism should have an impact only when there is majority incongruence between two houses.Applied to coalition cabinets, this statement assumes that by increasing the number of veto players, bicameral incongruence increases the potential for political gridlocks and tensions within coalitions.In other words, coalition cabinets that do not hold a bicameral majority would be short-lived.Hence, Druckman and Thies (2002) have found that governments that control a majority of seats in both chambers last substantially longer than those that lack majority status in one chamber.
However, a priori, these intuitive findings have been challenged by Diermeier et al. (2007), who concluded that bicameralism has no clear impact on coalition survival.
Therefore, the only two published works that dealt with determining the impact of the control -or not -of a bicameral majority on coalition survival in parliamentary regimes led to impasse.It is particularly surprising that, since then, no study has looked to settle these contradictory results.
As expressed above, it is even more surprising that no work has ever considered this question in relation to presidential regimes.Therefore, theoretical frameworks analyzing whether the controlling of a bicameral majority determines the survival of a presidential coalition remain quite limited.Thus, I opted to further the discussion of parliamentary scholars.
I followed the path of the first study on the subject by Druckman and Thies (2002) and made their argument intuitive, reasoning that the controlling of a bicameral majority makes the policymaking process easier, therefore avoiding coalition dissolution.Thus, the prime hypothesis of this work states that: H1: If a coalition cabinet in a presidential regime has a bicameral majority, then the coalition agreement would survive.

Bicameralism in Latin America
As shown in Table 01, bicameralism is a widespread configuration, being present in 52.6% (10/19 cases) of Latin American's polities.This figure is significantly higher than the 30% world average recorded by Tsebelis and Mahoney (1997).No* Total = 19 Bicameral = 10 (52.6%)Source: Constitutions of Latin American countries.Note: Venezuela and Peru used to have bicameral legislatures (until 1999 and 1992, respectively), but have since shifted to unicameral congresses.However, constitutional features are not frozen in time.Indeed, two countries (Peru and Venezuela) have recently changed from bicameral to unicameral congresses.
Conversely, as Lijphart (2012) has emphasized, no shift from a unicameral to a bicameral parliament has ever been recorded, at least at the national/federal level5 .
Finally, it is worth pointing out that in some polities the relationship between the executive and legislature has a direct bypass through the upper chamber.As shown in Table 02, in Argentina and Uruguay, for instance, the vice-president, who is a full member of the executive, also holds the position of head of the Senate (BIDEGAIN, 2017).This is supposed to confer a particular role to both the vice-president and the Senate in the policymaking process, as well as in the mediation between the executive and legislature6 .

Sample of the study: bicameralism and coalitions in Latin America
When computing every government since the third wave of democratization in the region, which began in 1979, the number of new cabinets derived from democratic processes raises to 134.Of these, 54.47% (74) were formed under bicameral polities.
However, as the object of this article focuses on pointing out a relation between _____________________________________________________________________________________________________  2017) 11 (2)  e0002 -8/27 bicameral majorities in presidential regimes and their impact on coalition cabinet survival or termination, here I only consider the coalition cabinets formed under bicameral polities.
I follow the definition of coalition cabinets set by Reniu and Albala (2012), considering that a coalition cabinet supposes the presence of distinct political forces that can count on the support of all their congressional party members.A coalition government is, above all, the result of a negotiation between two or more parties that have converged to form an agreement (whether public or not) about ensuring mutual commitment on a broad range of topics at different levels (mostly at the executive and legislative levels).Last but not least, coalition members are supposed to be responsible and accountable to both the president and the parties that form their legislative basis.
Hence, with this definition and focusing on the combination of 'coalition cabinets + bicameralism' in Latin America, and considering that I aim to identify a causal relation between bicameralism and coalition survival, I had to focus on concluded cabinets.Therefore, I have computed 25 cases of concluded coalition cabinets ruling under bicameral legislatures, which compose the sample of this study, as shown in Table 03.
The first conclusion that can be set out is the ability of Latin American presidents leading coalition cabinets to succeed in reaching bicameral majorities.
Indeed, 72% of coalition cabinets (18/25) enjoyed a bicameral majority on the inauguration day (ID) of their presidential mandate.Moreover, among the presidents who failed to secure bicameral majorities on ID, the first three Chilean presidents since the return of democracy (Aylwin, Frei and Lagos) had to deal with a particular constitutional feature inherited from Pinochet's rule: the existence of nine designated senators, mostly from the armed forces, who prevented the government from reaching a majority in the Senate.Therefore, excluding those three special cases, only four presidents-elect failed to reach a bicameral majority: De la Rúa in Argentina, Lula in Brazil (in his first term), Piñera in Chile and Pastrana in Colombia.
An interesting finding to point out from Table 03 is that every time that a president could only reach a partial legislative majority (i.e., a majority in only one chamber), the defecting house (i.e., the one in which the president could not reach a majority) was always the upper chamber.It appears, thus, that achieving a majority in the upper house is more difficult or 'expensive' than in the lower house.This finding is (2017) 11 (2)  e0002 -9/27 surprising when considering that the Senate usually has fewer members, compared to the House of Deputies.This is probably due to a different electoral rule for electing senators, which makes every seat in the Senate more costly, conferring to every individual senator a much greater value and influence than representatives of lower chambers.Notes: In the last column, I marked those cases that could not reach a majority with the value '0'; those that reached a majority in one chamber with the value '1'; and those that reached a bicameral majority with the value '2'.* = those that did not finish the mandate

Research design and method
Remembering that the objective of this article is to set out why coalition pacts survive or terminate, I will test several hypotheses picked from the mainstream literature, among which I highlight the impact of legislative majorities (whether bicameral or not) in order to find the eventual causal configurations that lead to the survival/termination of coalition pact.Moreover, as an internal condition, I added the 'number of partners' variable (H3), following the literature on veto players (ALEMÁN and TSEBELIS, 2011;TSEBELIS, 2002).These works assume that an increase in the number of players dilutes the likely concentration of vetoes, making the result of the game more insecure and unpredictable.
In other words, an increase in the number of parties in a government should increase the transaction costs within it.Conversely, a lower number of partners would make it easier to find a solution in case of disagreement among the members, thus decreasing potential dissatisfaction within the coalition (CHASQUETTI, 2008).In turn, this should impact coalition survival.
In addition, the occurrence of 'relevant' midterm elections during the mandate of the president (H4) is expected to have a great impact on coalition survival.In these cases, we anticipate that the incentives for remaining in coalition decreases (CHASQUETTI, 2008;COX, 1997).The relevant midterm elections being considered here are those that might produce parliamentary coalition realignments, leading to a weakening of presidential legislative influence.In other words, midterm parliamentary elections after which a coalition might lose a majority in one or both houses.
Finally, following Martínez-Gallardo's (2012) arguments, I added a contextual control variable (H5) that assumes that a 'favourable' socio-political context, expressed in high rates of economic growth and low levels of unemployment, has a positive impact on coalition stability.
In summary, the complementary hypothesis to be tested, together with the bicameral hypothesis, are as follows: H2: If an incumbent president cannot run for re-election, coalition survival would be affected.
H3: If there is a large number of parties in the coalition, the number of veto players increases and the lifespan of the coalition's agreement should decrease.
H4: If relevant midterm elections (legislative/governor) occur, tensions within the coalition should increase and the coalition may not survive.
H5: If a critical situation has been occurring, the incentives for government defection should increase.
I will proceed to the testing of the hypotheses through an operationalization of the conditions, following the principles of multi-value quality compared analysis (mvQCA).MvQCA is a case-oriented approach and one QCA variant (together with crisp- set and fuzzy-set QCA).It consists of establishing set-relationships between configurations of conditions (or 'independent variables') and an outcome (in other words, a 'dependent variable').Causal relationships, in QCA, are expressed in terms of 'sufficiency' and 'necessity'.In short, a condition, or a configuration of conditions, is considered as 'sufficient' if its presence is enough to produce the outcome.On the other hand, a condition is called 'necessary' when the outcome cannot, or can hardly, be produced without it (SCHNEIDER and WAGEMANN, 2012).This approach, also known as the 'configurational method', does not consist of setting probabilities for a condition 'X' to produce an outcome 'Y'9 .Instead, it confers the possibility to find, for an outcome ªYª, the conditions (X1, X2, X3…) and configurations of conditions (e.g., X1 together with X2) that lead to this outcome.In other words, by assuming equifinality, which presents all the paths producing the outcome, it presents the great advantage of combining conditions together.This advantage is, therefore, particularly interesting for our purpose here, especially when considering the relatively low number of cases in this study.
MvQCA differs from the most well-known application of QCA (i.e., crisp-set QCA), as it includes a higher level of membership complexity for the conditions, above the explicit present/absent dichotomy of crisp-set QCA, as well as the implicit calibration of fuzzy sets (CRONQVIST and BERG-SCHLOSSER, 2009).MvQCA consists of calibrating multinomial categories, as it introduces high-, medium-and low-set memberships.This complexation permits greater refinement of the analysis and a closer view of the configurational relationship.Indeed, through the dichotomization, some information may be lost due to rough or sometimes arbitrary thresholds.MvQCA permits, thus, the catching of a higher variation for cases and configurations.
Finally, here I considered that fuzzy sets were not the most suitable variant due to the dichotomous nature of the outcome (survived/breakdown) and the difficulty of proceeding with calibrations for many of the conditions being tested.Hence, mvQCA appeared as the best variant for the testing of the hypotheses.
I proceed to the operationalization of the outcome and conditions, following the QCA guidelines, from the five hypotheses set out above.Every condition is supposed to be relevant in producing the outcome, thus the operationalization follows a 'if then' _____________________________________________________________________________________________________ ( 2017) 11 (2) e0002 -13/27 logic, in which the absence of a condition is expected to be conducive to the absence of coalition 'survival'.

Operationalization of the conditions
Table 04 presents the operationalization details of the outcome and each condition, corresponding to the five hypotheses to be tested.The conditions are ranked according to their theoretical likeliness of occurring.• 1, if the coalition pact survived until the end of the president's mandate; • 0, if the coalition pact broke down.
'The bicameral majority condition (MAJ)' corresponds to H1.This condition includes, thus, the controlling, or not, of a legislative majority.Hence, the operationalization consists of: • 2, if the coalition holds a bicameral majority; • 1, if the coalition holds a majority in only one chamber; • 0, if the coalition holds no majority at all.
'The re-election condition (REELEC)', corresponding to H2, relates to the possibility of a president leading a coalition cabinet running for re -election.In other words, I will consider all the incumbent presidents that could or could not/could not anymore run for a consecutive mandate.The operationalization stands as follows: • 1, if the president could run for re-election; • 0, if the president could not or could not anymore run for re-election.
'The number of partners condition (PART)' corresponds to H3 and follows the literature on veto players and their expected effects on coalition survival.In this case, we operationalize this condition while considering the arguments in the literature regarding the most 'favourable' number of partners to avoid potential tensions.I followed Saalfeld's (2011) considerations about low, moderate and high numbers of partners, where he suggests that every additional member decreases the probability of coalition survival by 29%.A comparable work, although less systematic, has been carried out by Chasquetti (2008) for presidential regimes.The rate supposes that if the coalition has six members or more member, it should not endure.Thus, I coded this condition as follows: • 2, when the coalition is formed by the minimal number possible (i.e., two partners); • 1, when the coalition is formed by three to five partners; • 0, when the coalition is formed by six or more partners.• 1, if a midterm election occurred during a presidential mandate; • 0, if no midterm election occurred during a mandate.
'The context condition (CTXT)' or 'control' condition corresponds to H5.This condition deals with the context the government had to deal with, supposing that a favourable context would produce fewer incentives for walking away.The data are taken from the World Bank and the Economic Commission for Latin America and the Caribbean (ECLAC) annual reports on economic and social conjuncture.The condition is operationalized as follows: • 1, a favourable or normal socio-economic context; • 0, a critical or unfavourable context, marked mostly by the occurrence of an economic crisis/recession or socio-political tensions.

Results and findings
The merit of QCA is that the process is explicit and systematic, making the replication of the test feasible through transparent coding and the presentation of the cases.Therefore, these characteristics make the verification of the process easy (RIHOUX and RAGIN, 2009).Thus, in order to make the forthcoming conclusions transparent and comprehensive, I computed the codification process in Table 05, Conversely, the expected theoretical path conducive to coalition breakdown can intuitively be summarized by the combination: In actuality, the objective of QCA is to find out some 'minimal' paths that produce the outcome -that is pointing out if some reduced configurations of conditions are sufficient or necessary for the result (either 1 or 0).The next session will then try to find out if there is any minimized configuration, among those we are testing here, that is able to explain the survival or breakdown of coalition pacts.

Analyzing the results
I carried out the testing of the hypotheses using the software Tosmana 1.52, developed by Cronqvist (2016).In order to determine which causal configurations should be classified as leading to RESULT, I followed Schneider and Wagemann's (2016) good practice guidelines for QCA, explaining first the paths driving the survival of coalition cabinets, and then those leading to coalition breakdown.Moreover, considering that the first three Chilean cabinets had to deal with the particular constitutional context of Pinochet's designated senators, I expose the results without them11 .
Before going to the results, I computed, first, the 'truth table', which is the expression of every configurational path together with its empirical expression.I have included into the truth table the first three Chilean cases for illustrative reasons.The first finding is that the truth table exposes 17 different paths12 , but among them some contradictions appear (coded with 'C' in the table ).Table 06 shows that paths 08 and 13 present some inconsistencies.Thus, these configuration paths seem not to be decisive (neither sufficient nor necessary) in producing the results.Moreover, this consideration is particularly true when considering that both contradictory paths include cases from the same country.Path 08 is composed only of Bolivian presidents (Paz Estensorro, Banzer and Sánchez de Losada II) and path 13 is composed of two Uruguayan presidents (Lacalle and Sanguinetti II).Thus, the reasons for these contradictions tend to be more circumstantial and confined to local/idiosyncratic aspects13 .
In fact, contradictory configurations (i.e., a configuration of conditions that produce an inconsistent outcome) constitute one of the major challenges of QCA (RIHOUX and DE MEUR, 2009).However, there are several ways to deal with contradictions.The more frequent options are: 01. the inclusion of a new condition deriving a new hypothesis; 02. the recodification of the operationalization (from a theoretical and/or rational basis); or 03. the inclusion of new cases.
However, considering that this study tests the relevance of bicameral majorities on coalition survival in presidential regimes based upon hypotheses picked from the literature, the first option is not adequate as it is not in line with the objective of this work.Secondly, the codification of the conditions already fits existing literature, which means that a recodification would, again, deviate from the objective of theory testing.
Finally, the case selection followed the conceptual framework presented in the first part _____________________________________________________________________________________________________ Note: in italic appear the first three conflictive Chilean cases.In the contradictory paths, the number after the case i.d, consists in the outcome value.Now, when interpreting the data, the first conclusion which can be pointed to through the analysis of the Raw Data Matrix (Table 05) and the truth table is that coalition survival is far from being an uncommon feature of Latin American countries.
Indeed, among the 25 cases, 16 (64%) survived until the last day of the presidential mandates.This score falls a little, but not significantly (59.1%), when the three Chilean cases mentioned above are removed.Thus, we can conclude that coalition pacts in Latin American presidential regimes are more likely to survive than to breakdown, except for Brazilian coalitions (here, the coalition survival rate is 28.5%).Now, I set out the conditions that lead to the survival or breakdown of the coalitions.
When proceeding to the truth table minimization the findings are quite surprising.As a matter of fact, when running the necessity test I found no explanatory solution.That is, there is not any condition, single or combined to any other, that is 'necessary' to produce stable coalition cabinets in the tested model.
Then, I ran a sufficiency test and found an interesting result.The minimized formula for the survival of coalition cabinets in bicameral presidential regimes (Result = 1), with a coverage of 71.42% (10/14) for all the 'survived' coalitions, is sets out as follows: MAJ In other words, 'if a coalition holds a majority in only one chamber (MAJ(1)) or is constituted by a moderate number of parties (PART(1)), then it would survive'.In other words, holding a majority in only one chamber or a moderate number of coalition partners constitutes sufficient conditions to produce enduring coalitions.
These results are not, nevertheless, freed from further explanation or the inclusion of new conditions.In fact, the first path states that having a majority in only one chamber (MAJ = 1) would be, itself, a sufficient condition for the survival of coalition pacts.This result is quite surprising and counter-intuitive as the control of a bicameral majority, which is, logically, a stronger condition, is not sufficient by itself to ensure the survival of coalition cabinets.Thus, this finding may suggest an illogical statement as having a majority in only one chamber is a stronger condition than controlling a bicameral majority.This path should, thus, be taken with great caution, and the cases considered (Bachelet, Piñera and Pastrana) deserve deeper study in order to understand the survival of their coalitions.
Furthermore, the second sufficient condition (PART( 01)) is also partially surprising.Indeed, it states that a moderate number of partners (03 to 05) would behave better (or more efficiently) than the possible minimal number of coalition partners (02).This situation has already been pointed to in previous work (ALBALA, 2016) and can be explained by the fact that a two-party coalition tends 01. to be formed by parties with similar political strengths that compete for the electoral leadership of their political pole, or 02.tends to be unbalanced, with one asymmetrically dominant party and another being a circumstantial partner (RENIU and ALBALA, 2012).Both cases, indeed, favour the generation of potential tensions or disagreement.
However, the most relevant finding of this minimization is that the controlling of a bicameral majority seems not to be a determinant for coalition survival.Indeed, the minimization process did not point to a relevant (either sufficient or necessary) role for bicameral majorities in producing the positive outcome.This finding is very surprising as it is counter-intuitive.As exposed in H1, the control of a bicameral majority was supposed to ensure governability, weakening the shadow of unexpected contexts.
However it seems that this condition is not a determinant of stable coalition cabinets.As a matter of fact, the second mandates of Cardoso, and more recently Rousseff, in Brazil constitute a perfect illustration of the absence of a direct causal relationship between a bicameral majority and coalition survival.Hence, the empirical test disqualifies H1 and

Findings for the neglected outcome (RESULT =0)
The minimization process to explain coalition breakdown was not as linear as for coalition survival.Again, I could not find any 'necessary' condition to explain the negative outcome.I also could not isolate one formula only or the 'sufficiency' test.As reported below, I found four sufficient formulae leading to the neglected outcome: Although moderately confusing, these four formulae are in fact quite instructive and confirm some of the stated hypotheses above.Indeed, the four formulae can be read as follows: Formula 01: If a coalition holds no majority at all (MAJ(0)) or is formed by six or more partners (PART(0)), or holds a bicameral majority composed of only two parties (MAJ(2)*PART( 2)), then it would breakdown.Nonetheless, these formulae need to be carefully considered for several reasons.
First of all, it appears that some paths correspond to the same case.For instance, Batlle is explained either by the path MAJ(2)*PART(2) (formula 01 and 03) or by the path PART(2)*INTERM(0)*CTXT(0) (formula 02 and 04).These equifinalities are quite annoying, as they do not provide much information to help understand the outcome.
Nevertheless, when focusing on the logical expression of these paths it is quite easy to remedy these issues.Indeed, taking the example of Batlle, the path MAJ(2)*PART(2) (formula 01 and 03) appears as an explanation for coalition dissolution.However, this path (a two-party coalition holding a bicameral majority) is, logically, unsatisfactory.It should, rather, read like Batlle's coalition breakdown 'despite' being a two-party coalition holding a bicameral majority.The same goes for the paths corresponding to De la Rúa's coalition breakdown, in which re-election combined with having the two-party coalition ((REELECT(1)*PART(2)) appears as a dubious cause for dissolution 14 .For these cases, therefore, the explanation seems instead to be present in the alternative path, and would require a case study in order to be better explained 15 .
_____________________________________________________________________________________________________ 14 Especially when considering that the coalition broke down less than a year after De la Rúa's election. 15For Batlle, the fact that his coalition was formed by the two traditional Uruguayan parties that usually compete for the presidency, as well as the approximation of the 2002 economic crisis, makes the conjunction of these two conditions more plausible, including the absence of midterm elections (PART(2)*INTERM(0)*CTXT(0)).For De la Rúa, the total lack of a parliamentary majority, the occurrence of internal scandals and an open conflict between the president and his vice-president made the coalition untenable.For these paths a clear finding appears.Indeed, the condition related to an extreme number of partners (above six members, as detailed in the codification process) appears in every formula, revealing a perfect consistency, which thus confirms H3.In other words, the presence of this condition ((PART(0)) is sufficient to produce the neglected outcome -that is the breakdown of a coalition.
Moreover, a second finding is that the lack of an absolute legislative majority (i.e., no majority in any house) also appears as a consistent condition -present in two formulae -confirming the negation of H1.In other words, this condition (MAJ(0)) appears, also, as sufficient for outcome 0, even if in a lower proportion than the former condition (explaining only two cases).

Conclusions
This article addressed the original question of how the holding of a bicameral majority affects -or not -coalition survival in presidential regimes.Despite a recent boom in works on this subject, coalition theories applied to presidential polities have almost never tested how presidential coalitions behave and form under bicameral legislatures.As a matter of fact, it seemed quite evident that holding a majority in both chambers clearly benefits coalition survival.In other words, the general feeling was that for coalitions that exist in a bicameral context, the controlling of a majority in both chambers would be a 'sufficient' condition for its success and survival.And this assertion does not need to be tested.
This article, thus, tested this assertion.The findings are very surprising although they should be taken with some cautious confidence.First of all, coalition survival, contrary to common belief, is far from being an uncommon feature of Latin American polities.Indeed, the majority of coalitions survived until the last day of the presidential mandates.Coalition pacts in Latin American presidential regimes are more likely to survive than to breakdown, except for Brazilian coalitions.
However, I have shown that, again contrary to common belief, holding a bicameral majority does not guarantee coalition maintenance.Furthermore, the negation of this relationship, that is the controlling of no majority at all, seems to produce coalition breakdown.
This article encourages a broader, more systematized consideration of bicameralism when studying coalition cabinets in presidential systems.It constitutes an  2014), as it has contributed to the inclusion of the feature of bicameralism into the agenda of coalition theories in presidential regimes.As a matter of fact, there are still several topics on bicameralism that are important to the thorough analysis of coalition cabinets.For instance, the study of the electoral systems of upper and lower houses, and their impacts, appears as a relevant topic to begin with.In the same vein, as pointed out by Bidegain (2017), studying the vice-president, who is, in some cases, also the president of the upper chamber, is also an understudied field that deserves more attention, particularly with regards to executive-legislative relations under presidential regimes.
Revised by Thomas Rickard Submitted on September 05, 2016 Accepted on May 05, 2017

First
, it is necessary to define the outcome.With the survival/termination of a coalition agreement, I focus on the composition of the coalition cabinets and their original formations.In other words, I consider whether the partisan members of the coalition cabinets are the same from the ID of the mandate to the last day of the presidential mandate.For this reason, I shall not consider cabinet reshuffles.A coalition cycle is considered 'terminated' when any alteration in the original composition occurs.This could be either 01. the defection of a member (or party), whatever the motivation for walking away, or 02. the inclusion of a new member in the coalition 7 .Both these scenarios imply a change in the balance of power and influence within the original coalition.I shall consider a 'survived' coalition as a coalition that could keep all the founding members together from ID to the last day of the mandate.On the contrary, any change in coalition composition would lead it to be considered 'terminated'.Similar to coalition formation processes, the rupture of a coalition agreement and the defection/inclusion of a coalition member may have several external or internal origins.Nevertheless, coalition survival, in presidential regimes, is not an entirely unexplored theoretical field.Some scholars have already discussed theoretical frameworks that relate to the fostering of coalition maintenance/dissolution 8 .Therefore, the objective is to determine whether the controlling of a bicameral majority by a presidential coalition has a relevant impact on the survival of coalition pacts.With the broader objective of identifying 'why' coalition cabinets in multiparty presidential regimes last, I added three other variables and a control variable, all chosen from existing literature on coalition politics under presidentialism.I looked to map and establish some inferential configurations on coalition pact survival/termination in presidential regimes.Below, I present the four most common arguments from coalition theory literature as complements to the bicameral hypothesis (H1).The re-election condition (H2) was included as a main external constraint.Most scholars who have worked on government coalitions in presidential systems have established that the possibility to run for re-election can operate as a strong constraint against the temptation to walk away(ALTMAN, 2000;CHASQUETTI, 2008;GARRIDO,

'
The midterm condition (INTERM)' corresponds to H4 and considers the occurrence of relevant midterm elections (for the legislative or subnational level) during the mandate as a possible source of tension within a coalition.Hence, the dichotomous operationalization relates to: in which the values of every condition for each case can be easily observed.I highlighted the MAJ condition in order to facilitate the understanding of its potential relationship with the outcome.Hence, following the formulation of the five hypotheses and remembering the operationalization process presented above, the expected results regarding the survival of a coalition pact should logically be summarized by the intuitive combination: MAJ(2)* REELEC (1)*PART(1)*INTERM(0)*CTXT(1) => 1 10 .This combination reads as follow: when a coalition holds a bicameral majority (MAJ = 2); faces an institutional configuration that allows a president to run for reelection (REELECT = 1); regroups few partners (PART =1); without midterm elections (INTERM= 0); and enjoys a 'favourable' context (CTXT =1), then this coalition would survive (RESULT = 1)._____________________________________________________________________________________________________ 10 In QCA notation, the sign '*' supposes a relation 'AND' (i.e., the combination of two or more conditions); the sign '+' supposes a relation 'OR' (i.e., an equifinality to produce the result).
Thus, I shall deal with the contradictions and include them in the minimization test, following the standard guidelines for crisp-set QCA and mvQCA(RIHOUX and RAGIN, 2009), with the inclusion of logical remainders in the minimization process.I focus, first, on the analysis of the present outcome (RESULT = 1), then I proceed to the analysis of the neglected outcome (RESULT = 0).I operate through the truth table minimization process, which relates to finding relevant conditions and combinations of paths present in the truth table.For instance, let us consider three conditions (A, B and C) for explaining an outcome 'Y'.The minimization process eliminates the conditions that are revealed as redundant or inconsistent.Hence the process operates as follows: to Diermeier et al.'s (2007)  previous findings on parliamentary regimes.

Formula 02 :
If a coalition holds no majority at all (MAJ(0)) or is formed by six or more partners (PART(0)), or is composed by two parties that did not face a midterm election but suffered an unfavourable context (PART(2)*INTERM(0)*CTXT(0)), then it would breakdown.Formula 03: If a coalition is formed by six or more partners (PART(0)) or is formed by two parties that hold a bicameral majority (MAJ(2)*PART(2)), or is formed by two parties that faced a midterm election (REELECT(1)*PART(2)), then it would breakdown.Finally, formula 04: If a coalition is formed by six or more partners (PART(0)) or is formed by two parties that faced a midterm election (REELECT(1)*PART(2)), or is composed of two parties that did not faced a midterm election but suffered an unfavourable context (PART(2)*INTERM(0)*CTXT(0)), then it would breakdown.

Table 01 .
Bicameralism in Latin America

Table 02 .
Vice-Presidents in Executive-Legislative Intermediation

Table 03 .
Coalition cabinets in bicameral polities in Latin America (concluded cabinets)

Table 04 .
Operationalization of the conditions ECLACThe phenomenon to be explained, that is the 'outcome', consists of the survival or breakdown of a coalition cabinet.Its operationalization is, therefore, quite easy.Thus, following its definition as exposed above, I coded my outcome (RESULT) through a dichotomization as follows: Coalition Presidentialism in Bicameral Congresses: How does the Control of a Bicameral Majority Affect Coalition Survival?(2017) 11 (2) e0002 -14/27

Table 05 .
Raw data matrix of coalition cabinets' survival/termination in Latin America

Table 06 .
Truth Table for the Context Conditions for Coalition Survival/ Dissolution Coalition Presidentialism in Bicameral Congresses: How does the Control of a Bicameral Majority Affect Coalition Survival?(2017) 11 (2) e0002 -18/27 of this work, in which I clearly defined coalition cabinet and bicameral states.In other words, following these definitions, no other case could enter into the sample.