Too Legit to Quit? Analyzing the Effect of No-Confidence Motions on Cabinet Members’ Instability in Presidential Systems: The Cases of Colombia and Peru

ABSTRACT How does the execution of horizontal accountability mechanisms affect cabinet members’ instability? This article analyzes distinct features of no-confidence motions (NCMs) in presidential systems, using a mixed-method research design that identifies elements of legislative control mechanisms in Peruvian and Colombian polities. Although the congress in presidential systems rarely approves NCMs, high salience policy shocks trigger their proposal, resulting in the dismissal or resignation of the cabinet member in question. Those results are subtle opportunities for opposition legislators to indicate the incompetence of the incumbent government in particular policy areas. This study contributes to understanding how contextual factors affect the effectiveness of the check and balance principle in presidential systems.

NCMs, the Colombian Congress has not passed any since their adoption in the respective constitutions.
However, many failed attempts at NCMs have resulted in resignations in both countries. Examples are the recent cases of the Ministry of Information and Communication Technologies in Colombia for corruption charges (El Colombiano 2021) and the Ministry of Health in Peru for wrongdoings concerning the Covid-19 vaccine management (RPP Noticias 2021). Therefore, this study seeks to answer the question, How does the execution of horizontal accountability mechanisms affect cabinet members' instability? Considering that Colombia and Peru share some institutional features and that the approval of an NCM is a rare event in both polities, the constant use of this horizontal accountability mechanism by opposition legislators in Colombia and Peru is puzzling. Therefore, I theorize that policy shocks incentivize opposition legislators to propose NCMs to highlight the incompetence and the need for a change of cabinet ministers appointed by the president, whether by the NCM approval or the voluntary resignation of the cabinet minister.
NCMs are a horizontal accountability mechanism used to control inappropriate behavior and abuses of power by branches of government within democratic systems. 1 They grant state agents oversight actions, such as demanding information, imposing sanctions, and challenging the improper behavior of political leaders. NCMs are powerful instruments of control to punish political leaders for wrongdoings (Kenney 2003). Still, considering that NCM proposals vary depending on the targeted political leader, one could ask, What is the effect of NCMs despite often being unsuccessful?
To answer this question, we should consider the conditions that create favorable outcomes for opposition parties proposing NCMs. For example, in parliamentary systems, opposition parties use NCMs to highlight the incompetence of cabinet members or prime ministers. These are used strategically before constitutionally mandated elections to gain the favor of the electorate, enhancing the possibility of the opposition party taking power (Williams 2016). Alternatively, NCMs in presidential systems signal a salient policy's failure or incompetence and can be proposed only against cabinet members-not presidents. NCM procedures contribute to the public debate, illuminating the performance of cabinet members in the ruling government administration (Duque Daza 2015a). Therefore, I argue that in presidential systems, opposition legislators propose NCMs to raise public awareness about the ruling government's wrongdoings or incompetence when a critical event emerges.
A critical distinction between presidential and parliamentary regimes is the use of NCMs versus impeachment procedures. Both regimes use NCMs against cabinet members. However, in parliamentary systems, NCMs can be used against prime ministers, whereas some presidential systems can use impeachments only against presidents. Therefore, it is relevant to study the instability of cabinet members to test novel theories concerning the effects of NCMs in presidential regimes. So far, empirical explanations concerning cabinet instability have ignored the effects of NCMs when policy shocks emerge in presidential systems (Martínez-Gallardo 2014). In that sense, the dismissal or resignation of cabinet members in presidential systems through NCMs is relevant to analyze because those political leaders are responsible for implementing policies on behalf of the elected president.
To implement policymaking strategies, the president provides different cabinet portfolios to political parties to build a broad interparty coalition in Congress (Mainwaring and Shugart 1997;Amorim Neto 2006). However, this partisan portfolio allocation strategy is risky, due to the agency loss issue-when appointed ministers have alternative political incentives and use the executive authority to pursue goals contrary to the president's aims. To avoid this risk, the president appoints nonpartisan ministers to cabinet positions who are potentially more loyal to his policy vision (Martínez-Gallardo and Schleiter 2015). This appointment mechanism is a cooperative strategy in which elected officials use such appointments to reward political parties, interest groups, and social groups that help them win and maintain office (Camerlo and Martínez-Gallardo 2017). In presidential systems, patronage politics becomes essential to stay in power (Thiébault 2017).
Consequently, legislators of the party coalition prefer to work with partisan cabinet members in hopes of being rewarded with bureaucratic positions, contracts, or other forms of support. The empirical evidence of patronage politics explains the president's power over ministerial instability thus far explored in the literature. However, other factors concerning policymaking strategies are also relevant to consider, such as the electoral calendar and emergent scandals (Martínez-Gallardo 2014;Camerlo and Pérez-Liñán 2015b).
This article proceeds to highlight the literature gap about using NCMs against cabinet members in presidential systems. Then it develops a formal model and, considering the current strands of the literature on presidential systems, derives theoretical expectations emphasizing the relevance of policy shocks in the NCM setting. The analysis uses process tracing to test the logic of the proposed hypotheses in the selected countries, Colombia and Peru. As a way of triangulation, the research provides evidence using primary and secondary sources to demonstrate that factors concerning timing, cabinet member partisanship, and party coalition governance in Congress affect the strategic interactions of political actors when policy shocks trigger the execution of NCMs. The concluding section highlights the motivations of opposition legislators who use this horizontal accountability mechanism in both polities.

LITERATURE REVIEW
Although horizontal accountability mechanisms proposed by opposition parties have been tested in parliamentary regimes (Lijphart 2012;Williams 2016), academic literature on this topic in presidential regimes is limited. Research on those mechanisms in presidential systems focuses on impeachment procedures and policy monitoring. Impeachments are proposed by the legislative power to expose presidential mistakes and express discontent after policy shocks (Llanos and Pérez-Liñán 2021), especially when exposing presidential involvement in corruption scandals (Kim 2014), and partisan support in Congress for the president reduces the hazard of impeachment execution (Martínez 2017). Conversely, cabinets created by ideologically distant party coalitions provide stronger policy supervision among coalition members, resulting in lower ministerial replacement in the government (Silva and Medina 2022). In sum, the current literature on this topic in presidential systems has left out a critical tool of legislative oversight: NCMs toward cabinet members.
Indeed, certain features of presidential systems affect the use of NCMs against cabinet members. Gerring et al. (2009) claim that presidential systems politics predominantly center on the performance of individual politicians rather than established organizational entities. The lower and upper houses of a bicameral legislature in presidential systems have dual legitimacy 2 , which affects NCMs (Mainwaring and Shugart 1997). In addition, the fixed terms in presidential systems could impede the efficacy of NCMs if they were proposed before legislative and executive elections (Linz 1990). Moreover, the legislative and executive powers have incentives for cooperation, such as joining and maintaining government coalitions in highly fractionalized presidential party systems (Cheibub and Limongi 2002). These incentives, plus the use of particularistic norms, such as corruption, clientelism, and patrimonialism, could produce a negative vote on NCMs (Helmke and Levitsky 2004).
Explanations for the instability of cabinet members examine topics of coalition building as a policymaking strategy (Martínez-Gallardo 2012). Also, fragmentation in presidential cabinets, the strength of cabinet parties in Congress, and their economic performance significantly impact the tenure of cabinet ministers (Shin 2013). In sum, institutional and contextual factors characterizing legislative-executive relations are relevant for understanding the effects of NCMs in presidential systems.
In Latin America, the 12 countries that can use the NCM mechanism diverge in their treatment of an NCM vote. These countries are Argentina, Bolivia, Colombia, Costa Rica, Ecuador, El Salvador, Paraguay, Peru, Guatemala, Panama, Uruguay, and Venezuela (Duque Daza 2015a). Scholars can track polity variations and their impact on NCM binding decisions that lead to member resignation or removal from office. 3 The scope of NCMs varies between countries and depends on whether they apply to the chief of cabinet of ministers or all cabinet ministers. 4 Research also suggests that in presidential systems, crucial factors affect NCM execution, such as inadequate personnel training and institutional capture by hegemonic political projects affecting enforcement (Alcántara 2017, Page 15). In sum, NCMs in Latin American polities are rare events, and opposition legislators face tremendous challenges in successfully removing the targeted cabinet minister.
This inquiry recognizes the relevance of the contextual factors surrounding NCMs in presidential systems in Colombia and Peru. Both countries share a common historical heritage, the potential to elect outsiders as presidents, and state weakness reflected in security threats, among other issues. Conversely, these countries provide substantial variation to understand NCM proposals within different polities. For example, their constitutions differ in using NCMs, and their presidents adopt different cabinet appointment strategies, and their party systems contain many variations. These elements are key assumptions to properly analyze the strategic interactions between political actors who intend to override or sustain a cabinet member in their portfolio. Several studies of party coalition governance in presidential systems emphasize the influence of presidential powers on shaping the timing and agenda setting of the legislative power (Figueiredo andLimong 1999, 2000;Shugart and Carey 1992;Shugart and Haggard 2001;Colomer and Negretto 2003;Freitas 2016). Here, I argue that critical events also affect the relationship between NCMs and cabinet instability.

THEORETICAL EXPECTATIONS
I theorize that in presidential systems, policy shocks incentivize opposition legislators to propose NCMs to highlight the incompetence, wrongdoing, and need for turnover in the leadership of cabinet portfolios appointed by the president. Let us consider what prompts opposition legislators to propose an NCM when a public scandal emerges. One could argue that those opposition legislators propose NCMs to gain public recognition through legislative oversight actions toward the ruling government in this context. Still, those legislators should execute those proposals strategically because using them is not usually effective, but it could raise public awareness, leading to cabinet instability.
Policy shocks create uncertain political environments, affecting cabinet member instability in presidential systems. Lupia and Strøm (1995) argue that potentially critical events "are meaningful only if they affect politicians' abilities to achieve their legislative and electoral goals (p. 652)." Policy shocks framed as critical events can be caused by exogenous or endogenous factors. Salient policy shocks affect cabinet members' pursuits because they attract the critical eyes of the public and opposition legislators, who can take the opportunity to highlight the improper management of the cabinet portfolio.
Examples of policy shocks range from corruption scandals to a decline in the public perception of security. These political upsets provide unique opportunities for opposition legislators to utilize NCMs to signal a policy failure to the public, questioning the government's overall performance (Williams 2011). Moreover, policy shocks increase the risk of minister turnover in presidential systems, but their effects are constrained by strategic considerations (Camerlo and Pérez-Liñán 2015a). Thus, when a policy shock emerges, opposition legislators may initiate NCMs against cabinet members to undermine the executive power's range of maneuver.
Modeling NCMs using a simple formal model based on a sequential game serves to frame the strategic interactions between political actors involved in this horizontal accountability process. This model assumes that those actors-opposition legislators, cabinet members, and median legislators-choose from a set of available actions in a sequential game, in which each step provides ranked preferences for producing particular outcomes. The action I consider in this inquiry is the NCM proposal in Congress by opposition legislators against cabinet members. Based on the formal model that explains this mechanism in parliamentary systems (Williams 2016), I update it to show how it works in presidential systems. In this revised model, I include new nodes that express the possibility of the resignation of the targeted cabinet member.
The model assumes that all players are rational actors and make choices to achieve specific outcomes. Moreover, I assume that all players are policy maximizers and that Congress members are policy-and office-seeking agents. This means that opposition legislators want to show themselves as a more reliable ruling government alternative or to change the behavior of the incumbent government; active cabinet members want to finish their terms; and median legislators want to be reelected and to influence policy. Figure 1 shows the payoffs for each outcome in an NCM procedure against cabinet members in presidential systems, as ranked by the players' lexicographic preferences. These preferences involve ranking choices and selecting the most important option. The information in this sequential game can be assumed to be perfect, although the payoffs for each move may be subject to scrutiny. But scholars can use information about previous choices to predict the future behavior of a specific legislature.
In this game, there are three players: an opposition legislator (OPP) who proposes the NCM in Congress due to the emergence of a policy shock, the cabinet member (CM) who faces the NCM in Congress, and the median legislator (ML), who is a key actor to approve this horizontal accountability mechanism.

Figure 1. Formal Model of No-Confidence Motions (NCMs) Against Cabinet Members in Presidential Systems
One could analyze the players' behavior in this sequential game using backward induction. Looking ahead to player CM in the last node, their optimal choice is to resign after surviving an NCM procedure in Congress (CM-R*), so we can convert their decision node to a terminal node with payoffs (3,7,9). In the previous node, the ML compares the CM's payoff from approving the NCM procedure (NCM-A) (7, 5, 8)-that is, a payoff of 8-against a payoff of 9 from NCM-A in the terminal node if they decide to reject the NCM procedure (NCM-R). ML would opt, then, for NCM-R. Therefore, considering that in the previous node, the optimal choice for the CM is to stay (CM-S) when OPP proposes an NCM with payoffs (2, 4, 6)-that is, a payoff of 4-against a payoff of 5. Moreover, the OPP would begin the game by choosing between presenting the NCM proposal to Congress for the first time and ending the game ends in Status Quo (SQ) if it does not happen. Thus, If OPP chooses P, the cabinet member (CM) faces the NCM in Congress-that is, a payoff of 0-against a payoff of 2. Consequently, in this game, there is one strategy that captures the Subgame Perfect Nash Equilibrium: This strategy shows that the best option for a cabinet member facing an NCM proposal would be to resign after surviving the first NCM attempt. 5 It also considers that this would happen if the CM did not have proper support from the median legislator of the ruling party coalition. 6 Additionally, in response to a policy shock, the opposition legislator has the first-mover advantage in determining the direction of the game. 7 In that sense, I derive the following hypothesis: Hypothesis 1. In the presence of policy shocks, a cabinet member who has survived their first NCM in Congress will resign if they face a second NCM proposed by opposition parties.
On the basis of the outcome of this game, one could argue that the resignation of the cabinet member provides subtle incentives for opposition legislators to highlight the incumbent government's ineptitude in specific policy areas. However, considering the regime-type setting in which the NCM is proposed, it is relevant to include some analytical elements of presidential systems that might affect the oversight relationship between Congress and the executive power. These are the time constraints (Linz 1990) reflected in the honeymoon period and the electoral calendar (Chaisty et al. 2014), and the strength of the ruling party coalition in Congress, built on a strategy of the cabinet portfolio location (Shugart and Carey 1992;Mainwaring and Shugart 1997;Amorim Neto 2006;Cheibub 2007;Alemán and Tsebelis 2011;Kellam 2015;Mejía-Guinand et al. 2018;Silva and Medina 2022). In that sense, I derive the following hypotheses: H2. In the presence of policy shocks, the electoral calendar, reflected in the term limits of the legislature and the honeymoon period of the presidential term, will affect the passage of an NCM proposed by opposition parties.
H3. Following the emergence of policy shocks and the call for an NCM by opposition parties, the median legislator is more likely to block an NCM procedure if their political party belongs to the coalition majority in Congress.
Moreover, cabinet members are appointed by the president and represent the executive power. Therefore, if a cabinet member faces an NCM, they must gather majority support in Congress or resign. Presidents often appoint nonpartisan cabinet members to implement policy in order to avoid the risk of agency loss (Martínez-Gallardo and Schleiter 2015). However, their lack of partisanship makes them easy targets for opposition parties. Thus, one would expect nonpartisan cabinet members to be challenged politically more than partisan members. Therefore, I propose the following hypothesis: H4. In the presence of policy shocks, nonpartisan cabinet ministers are more prone to removal through NCMs proposed by opposition parties.
Although scholars have analyzed the effects of NCMs in parliamentary systems, which are primarily related to policy shocks (Popping and Wittek 2015;Williams 2016), the critical assumptions asserted here also demonstrate that policy shocks and contextual factors in presidential systems limit the effect of horizontal accountability mechanisms over ministerial turnover. The substantive contribution of this inquiry is to shed light on how policy shocks and factors concerning the timing of NCM implementation in Congress, cabinet member partisanship, and party coalition cohesion in Congress, bolstered by a strategy of cabinet portfolio allocation, matter in NCM execution, which directly affects cabinet member instability in presidential systems.

RESEARCH DESIGN
The analysis uses the cases of Colombia and Peru to test the effects of NCMs on cabinet instability in presidential systems, based on the formal model and the scope conditions of presidential regimes that include NCMs as a horizontal accountability mechanism for the executive branch. This inquiry adopts a mostsimilar research design. Colombia and Peru were selected because these countries share some scope conditions concerning historical linkages and state weakness (Mauceri 2001). Still, these countries differ primarily in their institutional polity structure that shapes executive-legislative relationships.
To be clear, although the Colombian polity is considered a presidential system, Peru is ruled through a presidential system that counts parliamentary elements in its polity. On the one hand, Colombia's political system is a moderate presidential regime with a demanding bicameral legislature. Presidents are elected for four-year terms and were allowed two terms from 2005 to 2015. 8 Congress is a bicameral legislature comprising a senate and a house of representatives. 9 Congress members are elected for four-year terms and allowed unlimited reelection.
On the other hand, some scholars classify the Peruvian political regime as semipresidential (Elgie 2011). However, the Peruvian institutional makeup does not comply with all the requirements to be categorized as a semipresidential system. According to Duverger (1980), three main features are required for a political regime to be considered a semipresidential system: the president is elected by universal suffrage, the president possesses considerable power, and the president has opposite along with them a prime minister and ministers who possess executive and governmental powers and can stay in office only if the parliament does not demonstrate opposition to them. The Peruvian polity adheres to the first two requirements but not the last one-also known as cohabitation-in which the president and the prime minister may belong to opposing parties or party coalitions (Hellwig and Samuels 2008). Therefore, I argue that the Peruvian polity is a special case of presidentialism with parliamentary features. Let us analyze those elements in turn.
The 1993 Peruvian Constitution included some elements of the parliamentary system, such as the creation of the position of prime minister (also known as premier), the inclusion of horizontal accountability mechanisms (i.e., the proposal of NCMs against cabinet members), the declaration of permanent moral incapacity against the president, and the procedure of interpellation and dissolution of Congress. 10 In this polity, the president chooses the premier instead of the popular vote and is in charge of coordinating cabinet portfolios and executive power policy actions, making the Premier as Super Minister. In that sense, I agree with other scholars that the president's extensive power over policy execution, fixed term of mandate, and cabinet member appointment power indicate a presidential system. Thus, Peru's institutional makeup concerning NCMs is unique among other presidential systems, making it an interesting case to compare and analyze based on the proposed theory.
In that regard, one could ask why NCMs are sometimes successful in Peru but not in Colombia. Using the case study methodology suggested by Van Evera (2016), I defined the temporal space of analysis for each country from the dates their current constitutions were established until 2021 (i.e., 1991-2021 for Colombia, 1993-2021 for Peru). I also organized information for the presidential terms during these periods. Considering that both constitutions incorporate NCMs as horizontal accountability mechanisms, one could argue that the emergence of policy shocks facilitates the strategic use of NCM mechanisms in those polities. Therefore, I evaluate the empirical support of my theoretical expectations based on cabinet member instability in Colombia and Peru.
This inquiry is subject to some limitations. Despite incomplete reporting from the official legislative websites, I used primary and secondary data gathered from them, along with news reports, to track NCM initiatives in Colombia and Peru. However, information concerning legislators' choices given an NCM proposal was unavailable. Still, secondary data resources detailed the policy shocks and triggers for the NCMs.
I also consulted academic literature in Spanish that highlighted constitutional constraints. Of the 29 NCM initiatives from Colombia, zero passed, and 7 led to the cabinet member's resignation. However, in Peru, of the 40 initiatives tracked, 4 NCMs were approved, and 10 led to the cabinet member's resignation (see appendixes 1 and 2 for data details).
This research explores the relationship between the occurrence of NCMs in Congress and their effect on minister turnover. The dependent variable proposed here is Cabinet member instability, which can arise in two scenarios: resignation of the questioned cabinet member or dismissal after NCM approval in Congress. The main explanatory variable is the occurrence of an NCM, as they are horizontal accountability mechanisms, or tools in the oversight process between official institutions-such as legislatures, judicial branches, ombudspersons, and general prosecutors-who control the abuse of power by other branches (O'Donnell 1998).
This inquiry uses process tracing to test the proposed theory. Process tracing helps identify the mechanisms between the independent variable and the outcome of the dependent variable (George and Bennett 2005). Additionally, this qualitative method reliably tests theories generated from other research methods, including formal models-as argued by Lorentzen et al. (2017). Based on the scope conditions and the information gathered on NCMs in Peru and Colombia, I contrast evidence to assess each proposed hypothesis as a means of triangulation. The analysis of process tracing in both country cases helps gain inferential leverage concerning observable implications reflected in the contextual elements affecting the use of NCMs against cabinet members in presidential systems. Let us analyze both cases in turn.

THE LIMITED EFFECT OF NCMS IN COLOMBIA: TRANSACTIONAL PRESIDENCY AND PARTY FRAGMENTATION
NCM proposals by opposition legislators in Colombia are not always effective because some factors limit their execution. These are the strategic use of legislative cooptation actions by the executive power to ensure governance, the continuous tendency toward fragmentation in Colombia's party system, the timing of the legislature, and the strict requirements for NCM approvals. To understand how these elements concerning the use of NCMs interact, I will clarify how contextual dynamics affect the proposal and unfolding of NCMs against cabinet members in Colombia's political system.
Colombian presidents have moderate legislative powers leading to legislative cooptation, adopted in 1991 (Duque Daza 2015b). Nevertheless, the president relies heavily on "collaborative" relationships in Congress to approve policies on their agenda (Carroll and Pachón 2016). Cabinet portfolio allocation is strictly the president's decision, following a "bait and switch" logic (Guinand and Botero 2017). The president allocates portfolios to cabinet members of their party coalition to ensure their power in Congress and limits partisan ministers' abilities to affect policy by modifying their office, mitigating agency loss risk (Mejía-Guinand et al. 2018). This partisan allocation is generally a successful strategy for the executive power to advance its legislative initiatives. 11 Every four-year legislative term is made up of four annual periods. Legislative years are divided into two periods of Congressional activity: July 20-December 16 and March 16-July 20. No new legislative action can be processed outside of these two periods. There is an old saying about how the balance of power between the legislative and the executive party shifts during each legislative period: The first year belongs to the party of the president. The second is a shared year, which means that the executive and legislative power jointly approve bills. The third year belongs to Congress, and the last year belongs to no one. The latter is considered to have no owner since the Congressmen are campaigning for reelection. (La Silla Vacía 2021) These shifts in power throughout legislative periods highly influence NCM timing by opposition legislators.
Due to a fragmented party system, Colombian politics is highly influenced by informal institutions, such as clientelist strategies and pork-barrel politics (Kugler and Rosenthal 2005). Research suggests that the Colombian party system has shown symptoms of electoral volatility and deinstitutionalization over time (Albarracín et al. 2018), including low partisan attachment among citizens, high decentralization of the political parties (Dargent and Muñoz 2011), the formation of cartel parties (Castañeda 2018), and the use of party-switching strategies by congress members to remain in the government coalition (Illera Cajiao and Buchely 2015).
Colombia's highly fragmented party system works because the president negotiates with coalition parties for political support of the policy agenda. The changes that allowed for the modification of presidential reelection rules in 2000 and 2015 are examples of this strategy at work. This quid pro quo strategy is rewarded through different cabinet portfolios, diplomatic assignments, contracts, and local and regional investments, among other formal and informal incentives. Thus, NCM failures are not surprising, due to the informal institutions influencing their approval process. Therefore, I argue that there is always an incentive for opposition legislators to propose NCMs, despite their inefficacy as a direct horizontal accountability mechanism to limit executive power when a policy shock emerges.
Since 1991, Article 135.9 of the Colombian Constitution has stated the requirements for proposing and approving NCMs. NCMs may be proposed in the event of policy shocks related to ministerial cabinet portfolio mismanagement or a member's negligence in responding to a legislative subpoena. After an NCM proposal, congressional leadership must schedule the public hearing with no fewer than five days' notice. The cabinet minister must also answer a written questionnaire sent by the general secretary of the Senate or the House of Representatives before the hearing. Voting is scheduled first in the house in which it was proposed between the third and tenth days following the end of the NCM debate in Congress (Senado de la República de Colombia 2021). In the event of an affirmative vote, the member will be relieved of their functions.
A constitutional amendment in 2007 made the approval of NCMs more attainable by modifying the majority required for approval while removing the ability to propose new motions on matters from previously rejected NCMs. Before 2007, NCM approvals required an absolute majority in both legislative chambers, making them almost impossible to pass (Duque Daza 2015a). If an NCM were rejected, a new motion could be proposed on the same matter if supported by new facts. Since the amendment, NCM approvals now require a simple majority (one-half plus one) of the proposing legislative chamber. However, a new NCM cannot be proposed on the same matter if the motion is rejected, regardless of new evidence. One could argue that NCM proposals in Colombia have a limited effect on the instability of cabinet members, due to the aggregate endogenous limitations provided by their institutional design, as seen in the timing of the legislatures and the internal approval procedure.
As such, it would be easier for a camel to pass through the eye of a needle than for an NCM to be approved in Colombia's Congress if the executive power supports the ruling party coalition in the legislature. There are several obstacles to the successful removal of a cabinet member through NCMs, including institutional design elements, the transactional relationships between the legislative and executive powers, and the fragmentation of Colombia's party system. The need to negotiate with the party coalition to maintain political stability in a highly fragmented party system is inevitable for the executive power. However, in the context of a policy shock, using NCMs is an effective strategy for opposition members because NCMs raise public awareness of policy issues. If those NCMs are implemented strategically, they can result in leadership changes in those cabinet portfolios.

THE EFFECTS OF USING NCMS IN PERU: THE SUM OF A STRONG PRESIDENCY, COOPERATIVE PORTFOLIO ALLOCATION STRATEGY, AND A COLLAPSED PARTY SYSTEM
The inclusion of parliamentary elements disencumbers the efficacy of NCMs in Peru's presidential system. In Peru, the president is the head of the executive branch and shares power with Congress, making a cooperative relationship necessary to achieve policy change. The president has exclusive rights over budgetary and economic policies to initiate ordinary bills through decrees, propose constitutional amendments and referendums, dissolve Congress (if Congress denounces the Council of Ministers twice), declare a state of siege, and demand an expedited prioritization of presidential initiatives in Congress, among other constitutional prerogatives (Ponce 2016).
Some parliamentary system features in Peru give Congress a significant role in the agenda setting of the executive power. Each time the premier and cabinet members are officially appointed, they must go to Congress for a confidence vote on their government plan. NCMs are used as accountability mechanisms to regulate executive power. For that reason, Peruvian presidents must consider executivelegislative relationships when making cabinet appointments.
Peru's unique polity and its collapsed party system change the role that Congress plays in the setting of horizontal accountability mechanisms. The Peruvian Congress is a unicameral body of 130 seats, assigned to each region corresponding to its population. Legislators serve five-year terms and can be reelected indefinitely. Elections for Congress coincide with presidential elections. The new Constitution of 1993 by the Alberto Fujimori administration (1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000) created selfreinforcing mechanisms, in which politicians are elected without party affiliation (Ponce 2016), making them political agents free from party ties and creating a highly volatile electoral system. Thus, personalistic parties or "coalitions of independents," the participation of outsiders in elections, the use of partyswitching strategies by congress members to remain in power, and the hiring of freelance "operators" as a substitute for party organizations, among other practices, have become standard features of this polity (Levitsky 2018).
In this way, a cooperative and nonpartisan portfolio allocation strategy benefits the executive branch of the Peruvian polity. Considering that party coalitions are informal and short-lived, the president bargains for appointing single ministers rather than grouping ministers with other political and economic actors and interest groups to govern and implement policy efficiently. The president appoints nonpartisan ministers to key positions, such as the minister of finance, to avoid agency loss and to obtain feasible policy outcomes (Vera and Carreras 2017).
Many aspects of the hyperpersonalization of Peruvian politics make cabinet members particularly vulnerable during policy shocks. Majoritarian electoral rules and electoral volatility, which lead to the collapse of the party system, incentivize the election of presidents who adopt median positions. Therefore, presidents use strategies to guarantee political stability, such as supporting median legislator bills (Ponce 2016) and using legislative cooptation strategies (i.e., competing informal institutions). Median legislative support decreases and executive-legislative conflict increases when an independent politician or outsider holds the presidency (Carreras 2014), making cabinet members more vulnerable to NCMs. Thus, policy shocks provide several incentives for opposition legislators to propose NCMs.
The Peruvian Constitution established NCMs in 1867. Still, in Articles 132 and 133 of the 1993 Constitution, NCMs are explained as follows: the NCM must be targeted against the premier or any other cabinet minister; it must be supported by at least 25 percent of the legal number of congress members; and it requires three days of study before voting between the fourth and tenth day after its official presentation in Congress. A simple majority approves NCM proposals in the legislative chamber. If an NCM is approved, Congress removes the cabinet member (Romero Herrera 2019). If the targeted minister is the premier, it could lead to a cabinet ministerial crisis-the complete resignation of the entire cabinet (Gobierno del Perú 2021). However, if the targeted cabinet minister resigns before Congress votes on the NCM, the NCM would be taken off the day's legislative agenda.
In this polity, it is important to distinguish between an NCM and a vote of confidence. The latter is a political move in which a cabinet minister stakes their position on Congress's adopting a particular policy decision. Therefore, when a cabinet member uses this mechanism and does not get approval in Congress, they could lose their position permanently (Cairo Roldán 2015). In Peru, if an NCM is approved or a vote of confidence is denied in the same executive administration, the president can dissolve Congress and call for new congressional elections within four months after its dissolution, except, according to the Constitution, in Congress's last year of the 4-year legislative term office (Gobierno del Perú 2021). Hence, Peruvian political dynamics require a functional relationship between the executive and legislative powers.
NCM proposals by opposition legislators in a policy shock context in Peru significantly affect the cabinet's instability. The main effect of NCMs is to draw attention to the government's incompetence in managing policy shocks and to demand changes in the leadership of the cabinet portfolios, including the premier. Still, Congress must execute this horizontal accountability mechanism strategically because the president can use veto power to dissolve Congress. Thus, cabinet member resignations are effective alternative outcomes of NCMs considered in this analysis.

DISCUSSION
Regarding the research question, we have seen that beyond NCM approval, related resignations are subtler successes for opposition legislators in a context of policy shock. However, three factors influence the effect of an NCM on cabinet members' instability in presidential systems. Figure 2 illustrates the relationship between these aforementioned factors. For triangulation purposes, I analyze the evidence gathered to support these claims.
To avoid the possibility of reverse causation, I claim that theoretically, NCM proposals are horizontal accountability mechanisms, defined as institutional tools that oversee the branches of government to each other, rather than just within their own branch (Mechkova et al. 2019). A cabinet member's resignation does not start an NCM procedure; opposition legislators initiate NCM proposals to demand a minister's resignation in the framework of policy shocks. Empirically, through tracking observable implications of the hypothesized factors within a case (Bennett and Checkel 2014), one could test the sequential effect of timing, the partisanship of the cabinet minister, and the support of the ruling coalition in Congress concerning the relationship between NCM proposals and cabinet instability. Concerning timing, the fixed term limits in presidential systems have a high effect on cabinet instability. The observable implications of this factor are reflected in the honeymoon period and the timing of the legislative terms. Until the fourth month of the term, one can track the honeymoon period, which is when public opinion, the legislature, and the media extend goodwill and cordiality to the newly elected president (Feres Junior and Sassara 2018). Although the approval of an NCM against the education minister in 2021 demonstrates the ability of the Peruvian Congress to use this accountability mechanism during the honeymoon period (Perú21 2021) The last year of a legislative term also negatively affects NCM approvals. In Colombia, one does not find recorded NCM proposals in the last year of the legislature, due to congress members' focus on reelection by visiting their constituencies to secure votes. Peru provides four events in which the last year of a term affected an NCM proposal: two at the end of the legislative period (Agencia Peruana de Noticias 2010; Expreso 2016) and another two at the end of the government administration (Congreso del Perú 2000; Expreso 2016). Table 1 provides evidence to illustrate the number of NCM proposals linked to disaggregated timing factors. This table is made with the information provided in appendixes 1 and 2.
Regarding the second factor, NCMs are often proposed and approved against nonpartisan cabinet members in presidential systems. The observable implication of this factor is cabinet partisanship. In the Peruvian collapsed party system, there is a growing tendency to appoint expert (nonpartisan) ministers. According to Vera and Carreras (2017), more than 73 percent of ministers appointed during the 1980-2014 period were nonpartisan. Therefore it is unsurprising that 23 of the 40 NCMs proposed were against nonpartisan cabinet members, including the four approved NCMs.
Conversely, in Colombia, only 5 nonpartisans out of 29 cabinet ministers were targeted by NCMs. The remaining NCMs were brought against members of the ruling party coalition. In that sense, opposition parties can be expected to target both partisan and nonpartisan cabinet members using NCMs. Therefore, a targeted member's lack of support from political parties is a strategic element to consider in the effectiveness of NCM proposals in Congress.  (1990)(1991)(1992)(1993)(1994) 3 (1994)(1995)(1996)(1997)(1998)  Furthermore, evidence indicates that the support of the ruling party coalition in Congress is critical in deterring cabinet instability. The observable implication of this factor is reflected in the approval or rejection of the NCM proposal in Congress. In Colombia, evidence shows that presidents depend on the legislative coalition majority to execute their political agendas, including NCM rejections (Duque Daza 2015a). Hence, the president uses competing informal institutions to gain support from the ruling party coalition in Congress to support the political agenda (Carroll and Pachón 2016). Likewise, in Peru, individual legislative cooptation strategies by the executive are necessary to sustain a ruling majority in Congress and similarly undermine horizontal accountability mechanisms (Levitsky 2018).
In that sense, I argue that the median legislator is incredibly influential in Peru, where party-switching strategies by members of Congress dictate NCM approvals. Three events concerning the approval of NCMs in Peru highlight this claim (Prensa Congreso del Perú 2021; El País 2015; La República 2016). Thus, evidence shows that NCMs are rarely approved when the president has the support of the majority party coalition in Congress, making the median legislator-as a representative of the ruling party coalition-highly influential in the NCM vote.
Aside from NCM approvals, cabinet member resignations are subtler successes for opposition legislators making NCM proposals. Many targeted cabinet members use resignations to avoid damaging their political capital and the presidential administration. This study, found that in the wake of a policy shock, cabinet members analyze their support by the party coalition in Congress and the likelihood of an NCM approval or rejection. In both polities, the data showed that cabinet members resigned before NCM votes and after tight majority votes in their favor. 12 Suppose there is a lack of sufficient political support. In that case, there is an incentive for the targeted cabinet members to resign before the vote takes place or after a marginal win in their favor to avoid tarnishing the executive's reputation, considering that targeted ministers lose political support every time they face an NCM.
Let us analyze the evidence for this factor. In Colombia, there were three events in which cabinet ministers who had survived previous NCMs were targeted with new NCMs following new policy shocks and resigned before the voting procedure (El Tiempo 2000Caracol Radio 2003). In Peru, retargeted cabinet members generally did not survive the second NCM challenge, due to a lack of support from the ruling party coalition in Congress (Perú21 2013). However, some cases in Peru showed that ministers targeted by repeated NCMs did not resign, due to their partisanship ties to the ruling party coalition in Congress (Congreso del Perú 1996, 1997, 2007.
In sum, considering the evidence of the effect of those analytical factors mentioned above, I argue that there is always an incentive for opposition legislators to propose NCMs when a policy shock emerges, even if they are rejected in Congress. NCMs provide unique opportunities for ministerial turnover by raising public awareness until public pressure increases to the point that the cabinet member loses the support of the ruling party coalition and will eventually fall into disfavor when new policy shocks arise.

CONCLUSIONS AND IMPLICATIONS
This article has argued that cabinet ministers are not "too legit to quit" when opposition legislators point out their wrongdoings or incompetency with NCMs in a policy shock context. This project identified three conditions under which opposition parties propose NCMs and how they affect cabinet instability in presidential systems. To understand this relationship, this study developed a rational choice model based on a sequential game that explains the strategic interactions among political actors in the proposal of NCMs to affect cabinet instability. Then it tested the model using case studies with two aims: to comprehend how this horizontal accountability mechanism works in each polity in the emergence of policy shocks, and to understand which factors affect ministerial turnover in Colombia and Peru.
In this inquiry, I have argued that policy shocks in presidential regimes provide unique opportunities for opposition legislators to propose NCMs against the cabinet members in authority, despite some conditions limiting NCMs' direct effects. Partisanship, for example, is highly influential; research suggests that presidents try to reduce the risk of agency loss by appointing nonpartisan cabinet members. Yet nonpartisan cabinet ministers do not have the political support needed by the ruling party coalition, which makes them more vulnerable and likely to be targeted by NCMs. The strength of the party coalition, as seen in support from the median legislator, greatly affects an NCM outcome. The timing around fixed term limits in presidential systems, such as the honeymoon period following presidential elections and the final years of the ruling term, also affects NCM outcomes and strategy. Timing is crucial because it reduces the likelihood of NCM proposals during specific periods of the presidential term. Resignations of targeted cabinet members are an effective indirect result, thereby further incentivizing opposition legislators to use NCMs as a horizontal accountability mechanism in presidential systems.
This study has analyzed the effects of NCMs in Colombia and Peru, which provides an excellent opportunity to consider the political implications of the abovementioned conditions limiting horizontal accountability mechanisms in presidential systems. In Colombia, elements linked to the timing of the legislative period, the cohesion of the party coalition shaped by quid pro quo strategies, and the flawed design of the NCM approval process affect the success of NCMs. Conversely, in Peru, the influence of the legislative period, the term limit of the presidency, the naming of nonpartisan cabinet ministers, and the lack of party coalition support in Congress affect the execution of NCMs. Therefore, I argue that using an NCM proposal is an effective mechanism, despite political challenges, to highlight and maintain accountability for the offenses of cabinet members following policy shocks.
Moreover, I claim that there is always an incentive for opposition legislators to trigger NCMs when a policy shock emerges. Research suggests that those NCM proposals are noisy signals highlighted by opposition parties concerning government performance because they raise public awareness around the targeted ministers' offenses and demonstrate the possibility of being more reliable government leaders for the electorate (Somer-Topcu and Williams 2014). In that sense, those NCMs in presidential systems provide some rewarding incentives to opposition legislators, such as public visibility, which usually rewards their reelection ambitions in terms of electoral competition and party leadership.
Scholars could conduct further research concerning NCMs framed in policy shock contexts in two main ways. A quantitative approach could be used in a cross-national study considering more Latin American countries that use NCMs as a horizontal accountability mechanism. Another avenue of research might analyze this topic from a comparative perspective, considering nations located in other geographical areas, emphasizing cabinet minister resignations as a critical element of cabinet instability. Further analysis of this topic undoubtedly would contribute to understanding the legislative-executive relationships in democratic regimes. vote in both chambers. In Costa Rica, an NCM can be targeted to any of the cabinet members and be approved with a two-thirds vote in both chambers. However, in Uruguay, an NCM can be targeted against any cabinet member and approved with three-fifths of the general assembly's vote.
5 8. President Álvaro Uribe Vélez and his legislative party coalition modified the electoral rules in 2000 to be reelected once. After that, Santos was elected, and he also used the of reelection. Later decided to change the rules, with the support of Congress, and limit the presidential office one term. 9. Actually, there are 108 seats the Senate. In the House of Representatives, there are 188 seats for the state circumscription.
10. This mechanism can be understood as an impeachment process toward the president. 11. Some scholars have noted that the presidential power is fueled by this transactional logic becoming a "legislative steamroller" (Duque Daza 2015b).
12. Appendixes 1 and 2 provide a register of those ministerial resignations.