Measuring Presidential Dominance over Cabinets in Presidential Systems : Constitutional Design and Power Sharing

This study focuses on the degree of political dominance exercised on cabinets by the executive chief in presidential systems. According to a debate that began in the 1990s, presidential systems are characterized by a non-collegial decision-making process, led by and personified in the figure of the president, in contrast to parliamentary systems where a joint decision-making process is prevalent. The key argument of this research note is that, although the majority of presidents have the constitutional power to remove cabinet ministers, the executive decision-making process in presidential systems is not necessarily vertical or based on a noncollegial process. By building a new index, we reveal a significant variation in the executive power exerted by presidents over their cabinets. To classify the degree of political dominance of presidents over their cabinets, we analyzed the rules of cabinet decision-making processes as defined in 18 Latin American constitutions.

he debate on constitutional designs and decision-making processes of presidential systems re-emerged in modern political science due to the third wave of democracy (HUNTINGTON, 1993).According to Carey (2005), the third wave of democracy exposed certain important trends, such as the establishment of democratic regimes in countries with no previous democratic experience, re-establishment of democracy in countries that had experienced periods of authoritarian regimes, and the expansion of independent states after the collapse of Soviet and European communism.An important consequence of these has been the attention given to the constitutional rules regulating the competition and the exercise of political authority in a democracy.
Traditionally, presidential systems have been characterized as a noncollegial decision-making process, led by and personified in the figure of the president (LIJPHART, 1992;SARTORI, 1997).By considering recent empirical comparative studies (AMORIM NETO, 2006;PAYNE, 2007), we argue against the assumption that the president's power to remove ministers will lead to a verticalized or non-collegial decision-making process, and argue that this assumption underrates the variations that may exist within the degrees of presidential dominance over the cabinets.By building a new index that provides a greater understanding of the degree of presidential dominance over their cabinets in the decision-making process, our results reveal a significant variation in executive power exerted by presidents over their cabinets.Moreover, our analysis of the variation of presidential dominance over cabinets in 18 Latin American constitutions allows us to classify the decision-making processes of the executive branch as decentralized to the benefit of ministers' powers, shared between ministers and the president, or centralized in the figure of the president.
If the institutional theory is correct that it is possible to predict political phenomena by considering the effects of rules on the behavior of individuals, it is reasonable to expect that different cabinets' rules can lead to different decisionmaking processes in presidential governments.The costs of a presidential decision to remove ministers, for example, may vary depending on other institutional factors such as rules that empower ministers.It may also vary according to other political factors, such as whether a minister's party is the pivotal actor in maintaining the coalition in the legislature (FORTUNATO and STEVENSON, 2013).Furthermore, the T Victor Araújo, Thiago Silva & Marcelo Vieira (2016) 10 (2) e0007 -3/23 distribution of resources for policy-making in the cabinet is directly related to the degree of influence the parties can exert on the composition of the executive agenda (ARAÚJO, 2016).Thus, the variation of how powerful the president is, vis-à-vis the cabinet, is of fundamental importance to enhancing our understanding of presidents' powers over their cabinets.
In the next section we briefly review the literature on presidential systems; in the third section we explore the puzzle that guides this study, and we present our argument; in the fourth section we develop our index and present the data and method used in our analysis; in the fifth section we present and discuss our preliminary findings, and; in the concluding section we present our final comments.

Literature Review
According to Elgie (2005), it is possible to identify at least three stages of the debate on presidential systems in the comparative politics literature.In the first stage, the classical Linzian argument prevails, in which Linz (1978) argues that parliamentary systems are superior to presidential systems, due to parliamentary systems offering greater political stability and decision-making capacity.Linz (1990Linz ( , 1994) ) argues that the independent election of the president and the legislators in a presidential system would lead to recurring conflicts between these actors, and would lead the presidents to claiming a greater democratic legitimacy for themselves, as the nation's representative.Moreover, argue Shugart and Carey (1992), presidentialism also presents the perilous fact that presidents represent both the state and the government and have a fixed term that is independent of parliamentary confidence.
The arguments of Mainwaring (1993) and Mainwaring and Shugart (1997) greatly influence the second stage of the debate on presidential systems.According to these authors, the problem does not reside in presidentialism itself, but in the combination of a presidential system with a highly fragmented legislature.In other words, the problem would result from the combination of presidentialism with a multiparty system in the institutions adopted in the newborn democracies in Latin America.According to Mainwaring (1993), in a system of government that is guided by non-cooperative incentives between the executive and legislative powers-a system which consequently offers no incentive for the formation of coalitions- presidents under a legislative minority would have no choice but to implement their policy agenda unilaterally.
Finally, in the third stage of this debate, studies were developed based on the criticisms to the inferences suggested by the scholars in the two previous stages.
To show that presidential democracies could produce governance conditions that are as stable as parliamentary democracies, many analysts in this stage of the debate revealed empirical evidence showing that in presidential systems, it is possible to formulate and implement policies without the interruption of the democratic order, albeit in contexts of low party system institutionalization (ALTMAN, 2000;ALTMAN and CASTIGLIONI, 2010;AMORIM NETO, 2006;CHASQUETTI, 2001;CHEIBUB, 2002;CHEIBUB, PRZEWORSKI and SAIEGH, 2004;DEHEZA, 1997;MEJÍA ACOSTA, 2009;NEGRETTO, 2006;ZELAZNIK, 2001).
From the above literature, we learned that: i.Although there are differences regarding the conditions of the governmental structure, the decision-making process in presidential and parliamentary systems of government may share similar characteristics (CHEIBUB, 2007;CHEIBUB, ELKINS and GINSBURG, 2014;TSEBELIS, 1995), and; ii.Potential analytical gains can be achieved when analyses overlook the differences between regimes, and consider instead intra-government scheme variations (EATON, 2000).
The empirical evidence that allows us to infer that the variable "system of government" is unable to adequately predict policy outcomes in presidential democracies is the same evidence that causes us to rethink the rules that structure the decision-making process in these democracies.
Models that predict how cabinets work and how their policy-making processes are defined should be evaluated empirically (ELGIE, 1997), instead of inferring results from assumed expectations.Moreover, in contrast to the idea that the decision-making processes in parliamentary and presidential cabinets are dichotomous, we should evaluate the possibility that the power-sharing among government members varies on a continuous scale (ANDEWEG, 1985(ANDEWEG, , 1988(ANDEWEG, , 1997;;BARBIERI, 2003;KEMAN, 2006;VERCESI, 2012).
From this perspective, it is possible to capture through empirical comparative analysis intra-system variations based on the president's degree of dominance over the cabinet.Instead of accepting the argument that the decision- making process in the cabinet within presidential systems has an inherently vertical character, with the presidents at the top and the ministers at the bottom (AMORIM NETO, 2006;LIJPHART, 1992;SARTORI, 1997), in this study we argue that there is a wide variation between cases and within cases with regard to presidential powers over cabinets in Latin America.
Thus, even when we consider that the president has the prerogative to dismiss ministers in the majority of presidential contexts, it does not automatically follow that presidents are plenipotentiary in relation to the decision-making process of the government cabinet.In other words, we cannot automatically imply that: i.All presidential constitutions grant the president vertical control over the cabinets, or; ii.That the power to select and remove ministers implies that the presidential power over ministers is unrestricted or unbounded.

Presidents' executive powers over cabinets: building the puzzle
With the publication of Presidential or Parliamentary Democracy: Does It Make a Difference?(LINZ, 1994), comparative scholars initiated a debate on what system of government would be best for the consolidation of new democracies, and what would be the best way to define and distinguish presidential and parliamentary systems (LIJPHART, 1992;LINZ, 1994;SARTORI, 1997;SHUGART and CAREY, 1992;VERNEY, 1992).
According to Lijphart (1992), the distinctiveness of government systems is defined in terms of mutually exclusive factors.In presidentialism, the head of government has a constitutionally-established fixed term, and cannot be, under normal conditions, removed from power by the legislature.Also, presidential systems have one person, a non-collegial executive, and the head of government is elected directly (or in some cases indirectly, such as in the United States) by the voters.In parliamentary systems, the head of government-usually called the prime minister-is selected by the legislature and can be removed by a legislative vote of no confidence or by censure.In this system, the prime minister and the cabinet form a collective executive body which has a collegial decision-making process.
One of the main contributions of Lijphart (1992) to the definition of government systems is his emphasis on the nature of the decision-making process within the executive branch.However, according to Amorim Neto ( 2006), Lijphart's  (2016) 10 (2)  e0007 -6/23 definition (1992) presents two problems.The first problem is the mistaken proposition that the concentration of executive power in the president does not favor the formation of coalition governments, which necessarily involves the sharing of power.Amorim Neto (2006) reveals that there are coalition governments in several presidential democracies.Thus the "sole executive" is not a distinctive feature of presidentialism.
The second problem of the Lijphartian definition (1992), and the focus of our study, is the argument that presidents, unlike prime ministers in a parliamentary system, exert a vertical control over the executive branch.Lijphart's argument (1992) assumes that there is a hierarchical relationship between the chief executive and the cabinets in every presidential system.This argument is supported by Sartori (1997) when he states that "the line of authority [in presidential systems] is neatly streamlined from the president down" (Sartori,p. 84).
Granted, several presidential constitutions define cabinet members as heads of ministries or public agencies, and "mere" presidential advisers.However, as highlighted by Amorim Neto (2006), some constitutions require the countersignature of a member of the cabinet for all or some acts of the president, while other constitutions establish some form of collegial decision-making within the executive branch.This is the case, for example, of the former Venezuelan According to Amorim Neto (2006), the aforementioned constitutional articles show that from a formal point of view Lijphart (1992) would be incorrect in assuming that the absence of collegial decision-making rules is a defining and universal feature of presidentialism.However, from a substantive point of view, the fact that the president has power to dismiss his ministers at any time leads Amorim Neto (2006) to agree with Lijphart (1992) and Sartori (1997), arguing that the president's power to dismiss her ministers verticalizes the president's relationship In this research note, we argue that the president powers to select and remove ministers do not automatically imply a non-collegial or a vertical decisionmaking process in presidential systems.By building a new index, we reveal evidence of intrasystem variations in presidential influence over cabinets.

An Index on Chief Executive Power over Cabinets
In this section, we present the research design and the data we used, and we develop an index to help scholars analyze how powerful the president is vis-à-vis the cabinet, according to specific constitutional rules that verticalize or horizontalize the decision-making processes of the executive power in Latin American democracies.
Although the relationship between political actors is more complex than the analysis of formal rules within constitutions, there are at least two advantages in considering the formal rules of the decision-making process within the executive branch.First, the analysis of the constitutions enables the comparison of cases in time and space, allowing a greater number of democracies to be investigated.
Second, given that the main inferences about presidentialism are based on typologies formulated by the analysis of formal rules, analyzing these constitutions allows an evaluation of these arguments using evidence from the current literature.
Moreover, as stated by Cheibub, Elkins and Ginsburg (2014), by focusing on constitutions we tie "our analysis of institutional design to an identifiable activity and mode of behavior-constitution making-and thereby keep the process in sight.In order to deal with the problem of comparability and reliability of the constitutional texts, in our analysis we used the data collected and classified by the Comparative Constitutions Project (CCP)2 (ELKINS, MELTON and GINSBURG, 2015).This project was developed with the goal of improving-temporally and spatially-the efficiency and systematization of the information contained in different constitutions (ELKINS, 2013;ELKINS et al., 2014).
In an attempt to measure the degree of presidential governmental dominance, we have taken into account five decision rules: 01.Cabinet removal; 02. Cabinets' selection procedures; 03.The restrictions of eligibility for the office of minister; 04.The minister's countersignature for executive decisions, and; 05.
The legislative powers of ministers.
These decision rules come from a conceptual framework developed and refined by the CCP investigators that includes over 600 attributes of constitutions.This framework was then converted to a guided survey instrument, which CCP coders used to read, interpret and compare the constitutional texts (ELKINS et al., 2014, p. 12).
The rule of cabinet removal was interpreted based on three questions: Are the cabinet/ministers collectively responsible for their actions, or can they be Therefore, regarding each of these rules as they appear in our analysis, the procedure for cabinet removal is the set of rules that limits or increases the presidential autonomy to dismiss cabinet ministers.If there are no limits to presidential autonomy, the cost of removing cabinet ministers is lower for the president, and consequently, presidential dominance over the cabinets will be greater.This variable is measured as a binary variable, where a value of 01 is assigned to the rule of the president to freely remove the ministers, and a value of 0 is assigned in those cases where the power of the president to remove ministers is limited (for example, when the president needs the support of other political actors to remove ministers).
Cabinet selection indicates whether the president can freely appoint her cabinet ministers or if there is any institutional rule that limits this power (for example, whether the president needs the ratification by other political actors regarding the ministers appointed).If the presidential nomination is free, the dominance exercised by the chief executive is greater.This binary variable is assigned a value of 1 when the president is free to appoint her ministers, or it is assigned a value of 0 if there is a rule conditioning the presidential appointment.
Restrictions on eligibility for the cabinet offices indicate that ministers need to meet certain requirements to be appointed by the president.The greater the number of restrictions on eligibility, the cost of presidential ministerial replacement is greater, and consequently, the influence of the president over the cabinets will be smaller.This binary variable receives a value of 0 in the presence of restrictions (beyond minimum age and nationality), and a value of 1 in the absence of this rule.
The requirement of a ministerial counter-signature for executive decisions indicates a collegial character between the cabinet and the president, whereas the absence of this rule implies a higher dominance of the president.
This measurement is also a binary variable, wherein the need for countersignature by ministers is assigned a value of 0, and the absence of a countersignature rule is assigned a value of 1.
Lastly, the distribution of legislative powers within the cabinets indicates whether ministers have some legislative prerogative or not.If ministers have power to legislate, the presidential dominance over the cabinet will be smaller.This binary variable receives a value of 0 when ministers have legislative powers, and a value of 1 when ministers have no power to initiate legislative proposals.
By having constitutions as units of observation (cases) and countries as the level of analysis, we have adopted a cross-section type of research design (KEMAN, 2014).A database was built that details the rules that verticalize or horizontalize the decision-making processes in 18 Latin American constitutions: As Fortin (2013) argues, there are at least two strategies to measure presidential powers in comparative politics.The first, originally formulated by Shugart and Carey (1992), then adapted by Metcalf (2000) and Frye, Tucker and Hellman (2001), presidential legislative prerogatives and presidential nonlegislative prerogatives are considered, based on 10 categories, resulting in an ordinal scale ranging from 0 to 4. Among the legislative powers of the presidents are: the decree power; the veto power; the power to initiate laws; budgetary competence, and the power to initiate popular referenda.The non-legislative powers of the president include: the power to select and remove ministers, and the power to censor and to dissolve the parliament.We can also include in this group the Legislative Institutional Power Index (IPIL) developed by García Montero (2009).García Montero (2009) compares the influence of the president and the legislature on the law-making process by considering five dimensions of the legislative process-legislative initiation, formation and regulation of legislative committees, symmetry of bicameralism, presidential veto powers, and the president's decree power and extraordinary prerogatives.
The second measurement strategy highlighted by Fortin (2013) has its origin in Duverger (1980), and consists of potential formal powers that can be attributed to presidents.From this measurement strategy, scholars have created indices of presidential powers, most commonly comprised of binary variables that are coded on the existence (or nonexistence) of the above powers.Several aggregated indices are based on this strategy, such as the indices suggested by Lucky (1993), Frye (1997), Siaroff (2003), and Armingeon and Careja (2007), in addition to the presidential power index proposed by Doyle and Elgie (2015).
These authors suggest an index that combines 28 categories in a comparative and local perspective, in addition to considering a larger time series compared to the previous cited studies.
The index proposed in our study fits into the second measurement strategy of presidential powers, as described above.This index reflects our perception that the current measurements produced by the literature do not adequately measure the phenomenon addressed in this study.Also, this index is justified by its parsimony-taking into consideration five categories-which makes it easy to understand and replicable.Another justification is its homogeneity, because all the categories considered in this index are related to the power sharing between the president and the members of the executive cabinet.Thus, this index avoids the criticism of Fortin (2013), who argues against existing indices that include legislative powers and non-legislative powers in the same index.

Analysis
The examination of presidential dominance over cabinets based on our index will be explained by an in-depth, descriptive analysis of the 18 selected constitutions.Through our analysis we explore the dispersion and central tendency of the aforementioned rules.Where appropriate and necessary, we will explore the specific information through graphics and tables.In Figure 01, for example, we depict the degree of presidential dominance over cabinets in the current constitutions of our selected 18 Latin American countries.
As we can see in Figure 01, Bolivia presents the most "horizontal" decision-making process concerning the presidential executive power vis-à-vis the cabinet ministers.In other words, according to the Bolivian Constitution, none of the criteria that compose our index of presidential dominance over cabinets is met.In the other extreme, Brazil, Guatemala, Mexico, Panama and Venezuela are the countries where the presidential dominance over cabinets presents a higher value.It is important to note that, according to our index, of the According to the Brazilian Constitution, the only limitation to the dominance of the president over the cabinet is the requirement of a countersignature from the ministers.The president's actions and decrees must be countersigned by the ministers of state, who also have the power to dispatch instructions on the enforcement of laws, decrees and regulations.

Results and Discussion
From our analysis of the variation of presidential dominance over cabinets in the 18 selected Latin American constitutions, we can infer at least three clear results (standards): Latin American presidential democracies can have 01.Decentralized, 02.Shared, or 03.Centralized executive decision-making processes.
Bolivia, Argentina, Dominican Republic, Peru, and Uruguay are the five countries whose constitutions present smaller empowerment of the president vis-à-vis ministers and government officials.The values below the mean (2.8) of our index show that not all presidential constitutions create a hierarchical governmental decision-making process.The Bolivian presidency is the most restricted executive power among the selected Latin American constitutions, with a value of 0 in our index, followed by the Argentine presidency that receives a value equal to 1 in our index, and the Dominican Republic, Peruvian and Uruguayan presidencies, which indices are equal to 2.
Constitutions of countries such as Ecuador, El Salvador, Chile, Colombia, Costa Rica, Honduras, Nicaragua, and Paraguay grant an average empowerment to their presidents, representing an amount equivalent to 3 according to our index-slightly above the mean value of 2.8.The value of 3 in our index for the constitutions of these countries verticalizes the decision-making process of the governments slightly more, granting more dominance to the president over the cabinets.
Finally, Brazil, Guatemala, Mexico, Panama and Venezuela present the strongest dominance of the president over cabinets among the selected Latin American countries, with an index equal to 4. Brazilian, Guatemalan, Mexican, Panamanian and Venezuelan presidents are those with the highest dominance in the executive decision-making process.Therefore, it can be argued that Lijphart's argument (1992) that the vertical character of presidentialism opposes the collegial character of parliamentary systems, which was accepted by Sartori (1997), Amorim Neto (2006), and the consolidated literature on presidential systems, cannot be sustained.The degree of dominance exercised by the chief executive on the cabinets has at least three distinct patterns if we take into account the current constitutions of the 18 selected Latin American countries with presidential systems.The variation in the dominance rate of the presidents in the executive decision-making process ranges from 0 to 4, and corroborates our argument that Latin American presidential constitutions centralize the responsibilities to compose and administrate the government in the presidency, and that we cannot automatically imply that, i.All presidential constitutions are necessarily vertical, or ii.That the power to select and remove the ministers makes the presidential power over ministers unrestricted or unbounded.
A possible counter-argument, in favor of the Lijphartian (1992) argument that has been consolidated by the mainstream comparative analysis on presidential systems, could be that, of the 18 countries analyzed in our study, eight feature indices with the same value, which could explain the low value of the standard deviation.However, the same value for our index in these countries hides important differences within these cases.The fact that Chile, Colombia, Dominican Republic and Ecuador, for example, were assigned an index equal to 3, does not mean that Chilean, Colombian, Dominican and Ecuadorian presidents have the same restrictions on the selection and removal of ministers, on their options to select their ministers, on the requirement of ministerial countersignature, or on the prerogative to initiate legislative proposals.There is variability in the rules that compose our index among the countries assigned with the same values.
As can be seen in Table 01 below, the distribution of countries according to each verticalization (+) or horizontalization (-) factor that composes our index varies significantly among the constitutions analyzed in this study.
members.Therefore, althoughAmorim Neto (2006) has been one of the first authors to indicate the possibility of intrasystem variation in the presidential influence over cabinets, he does not develop an argument on this variation nor evaluate it empirically 1 .
Importantly, we know who did what and when" (CHEIBUB, ELKINS and GINSBURG, Whereas our sources for the classification of formal rules are the constitutions themselves, validity does not appear to be a problem in our measurement and comparison of formal rules.However, reliability issues-that is, issues of conceptual and measurement equivalence-emerge when comparing the constitutions of different countries(ELKINS, 2010).The problem in comparing different constitutions is the lack of systematic classification of the constitutional texts.
dismissed individually?Who has the authority to dismiss the cabinet/ministers?What are the limitations on the ability to dismiss the cabinet/ministers, if any?The interpretation for the rule of cabinet selection was based on two questions: Who nominates/appoints the cabinet/ministers?Who approves the cabinet/ministers?Three questions guided the interpretation of the rules of eligibility for cabinet: ______________________________________________________________________ (2016) 10 (2) e0007 -9/23 What is the minimum age limit for eligibility to serve as a member of the cabinet?What restrictions does the constitution place on members of the cabinet/ministers ability to serve in the legislature?What additional restrictions does the constitution place on the eligibility to serve as a member of the cabinet?The question Does the constitution mention the executive cabinet/ministers? was used to interpret the rule of the establishment of cabinet, and the powers of cabinet were interpreted based on the question What specific powers are granted to the cabinet?
Argentina (1994),Bolivia (2009), Brazil (2014),Chile (2014),Colombia (2013),   Costa Rica (2011), Dominican Republic (2010),Ecuador (2011), El Salvador   (2003),Guatemala (1993), Honduras (2013), Mexico (2007), Nicaragua (2005),   Panama (2004), Paraguay (2011), Peru (2009),Uruguay (2004), and Venezuela   (2009).The information was initially gathered from the "Constitute Project"(2015), and later revised with information gathered directly from the original constitutions in order to increase the reliability of the data.In order to measure the presidential dominance over cabinets, we created a summation index of the codified rules regarding the executive powers of the president vis-à-vis cabinet ministers, ranging from 0 to 5, with 0 indicating the absence of presidential dominance and the value of 5 indicating the absolute dominance of the president over the cabinets.From the 18 Latin American Constitutions analyzed in this research, the observed values for this variable are distributed between 0 and 4, and no country received the maximum value of 5 in the variable "presidential dominance over cabinets".By proposing a new index, we aim to measure the dominance of the president vis-à-vis the other members of the executive cabinet.Unlike most of the indices proposed by the literature-which measure only the power of the president (or executive) against the actors who are members of the legislative Victor Araújo, Thiago Silva & Marcelo Vieira (2016) 10 (2) e0007 -11/23 branch-our index can capture some of the constraints the chief executive faces within the cabinet.Our index considers the variation of the power sharing within the executive cabinet according to the formal rules established by the constitutions.

Figure 01 .
Figure 01.Degree of presidential dominance over cabinets in Latin American Constitutions