Differences in the behaviour of chambers in bicameral legislatures

Starting with the classification by Arend Lijphart (Reference Lijphart1990) on the levels of congruence and symmetry in bicameral systems – concepts which refer to the rules that govern both their composition and functioning – extensive work has shown that the behaviour of legislators responds to incentives associated with electoral rules and other incentives associated with their time horizon, the size of each chamber, limits on re-election and the law-making rules in each chamber (Fortunato et al. Reference Fortunato, König and Proksch2013; Heller and Branduse Reference Heller and Branduse2014; Llanos and Nolte Reference Llanos and Nolte2003; Tsebelis and Money Reference Tsebelis and Money1997). While most studies on bicameralism focus on the formal powers and compositional differences of each chamber (Heller and Branduse Reference Heller and Branduse2014; Hug Reference Hug2010; Lee and Oppenheimer Reference Lee and Oppenheimer1999), others also include a legitimacy dimension associated with the representativeness of both chambers (Mueller et al. Reference Mueller, Vatter and Dick2023).

Early studies of state legislatures in the United States identified differences in how each chamber functioned in bicameral congresses (Squire Reference Squire1992, Reference Squire2017). The presence of a second chamber does not necessarily have an impact on legislative productivity (Rogers Reference Rogers2003), but it forces the executive to act strategically to advance the government’s agenda (Fisk Reference Fisk2011). The distinctiveness between bicameral state legislatures can be associated with the ratio of seats, bipartisan representation and constituency dissimilarity (Makse Reference Makse2022). Term-limit provisions also vary across states and within bicameral legislatures (Sarbaugh-Thompson Reference Sarbaugh-Thompson2010), as does the power of minority parties (Clark Reference Clark2015). The formal powers of the chambers’ leadership (Mooney Reference Mooney2013) and gate-keeping powers (Anzia and Jackman Reference Anzia and Jackman2013) also vary across states and within state legislatures. The distinctiveness in the composition and functioning of bicameral state chambers has important consequences in the law-making process (Makse Reference Makse2022). In other cases, differences in electoral rules and law-making procedures also account for different outputs between the upper and lower chambers (Patterson and Mughan Reference Patterson and Mughan1999).

In Latin American presidential systems, the executive is often said to have proactive legislative powers and the Congress reactive powers (Cox and Morgenstern Reference Cox and Morgenstern2001; Morgenstern and Nacif Reference Morgenstern and Nacif2002). The executive has tools to influence the behaviour of legislators, particularly those of its own party or the ruling multiparty coalition (Figueiredo and Limongi Reference Figueiredo and Limongi2000). Incentives and rules condition the behaviour of national legislatures. However, in a study of the functioning of the legislatures of seven Latin American countries, Eduardo Alemán and George Tsebelis (Reference Alemán and Tsebelis2016) concluded that the devil is in the details, highlighting how incentives and rules condition the behaviour of national legislatures. Similarly, in a study of roll-call votes in the United States and Latin America, Scott Morgenstern (Reference Morgenstern2004) associates differences with institutional incentives.

To understand the functioning of Congress as a whole – and not just one of its chambers – we must account for how the different institutional designs and law-making rules in both chambers affect the behaviour of legislators. Applying the model of the functioning of Congress as advanced by Thomas Romer and Howard Rosenthal (Reference Romer and Rosenthal1978) therefore needs to incorporate differences in the way each chamber operates (Tsebelis and Money Reference Tsebelis and Money1997). After all, in every vote, the median voter in one chamber may be affiliated to a different party than the median voter in the other (Tsebelis and Money Reference Tsebelis and Money1997).

The determinants of legislative cohesion

The nature of bicameralism is anchored in the idea that the decisive legislator in each chamber is not necessarily from the same party as the other or holds the same ideological position (Cutrone and McCarty Reference Cutrone, McCarty, Weingast and Wittman2006). If the rules governing the election of legislators and the functioning of each chamber are different, the behaviour of their members might also differ (Llanos and Nolte Reference Llanos and Nolte2003; Llanos and Sánchez Reference Llanos and Sánchez2006; Uhr Reference Uhr, Rhodes, Binder and Rockman2006). However, a few comparative studies on the behaviour of legislators in the two chambers have reported more similarities than differences, despite their different composition, electoral rules and time horizons (Desposato Reference Desposato2006).

Since legislators must respond not only to the interests of their principals – normally the voters – but also to their parties of affiliation, divergent interests and loyalties can be understood as reflecting competing principals (Carey Reference Carey2007). The influence of principals depends on the rules governing electoral and legislative processes. If the rules and nature of representation – whether national or at the district level – are different for each chamber, legislators’ behaviour and loyalties will also be different. Similarly, as their time horizons might be different, their relationship with the executive might also respond to different incentives. If lower chamber legislators within the government were elected on the coattails of the presidential candidate in the same election, they might be more loyal to the executive than upper chamber legislators elected in the previous cycle. It would be logical to expect that cohesion in roll-call votes by government and opposition legislators would be different for the upper and lower chambers.

Prior work has shown that legislators from the ruling party will face different incentives than minority legislators or legislators who represent parties that oppose the executive (Saiegh Reference Saiegh2011). Under the model proposed by Romer and Rosenthal (Reference Romer and Rosenthal1978), whoever controls the legislative agenda has more capacity to influence the outcome of the process. Thus, in presidential democracies where the executive controls the agenda, ruling party legislators should have more incentives to line up with the executive’s priorities. In turn, in presidential democracies where Congress has more proactive legislative powers, legislators from the parties that control the majority should have more incentives to be cohesive so as to take advantage of their position (Cox and Morgenstern Reference Cox, Morgenstern, Morgenstern and Nacif2002).

Law-making rules also matter (Cox and McCubbins Reference Cox and McCubbins2005; Krehbiel Reference Krehbiel1992). Majority parties take advantage of these rules to discipline their members (Cox and McCubbins Reference Cox and McCubbins2007). In institutionalised party systems, parties induce legislators to show greater discipline, setting aside their particular interests (Cox and McCubbins Reference Cox and McCubbins2007; Wright and Schaffner Reference Wright and Schaffner2002). Parties punish legislators for undisciplined behaviour by establishing incentives and sanctions that may even include a ban on seeking re-election (Carey Reference Carey2007; Hix Reference Hix2004; Sieberer Reference Sieberer2006). In an analysis of institutional effects on party unity in 19 countries, John Carey showed that ruling parties tend to show less unity in presidential than parliamentary systems (Carey Reference Carey2009). Nonetheless, the executive has some tools to foster greater cohesion among government legislators or, at least, lessen cohesion among opposition legislators (Figueiredo and Limongi Reference Figueiredo and Limongi2000). In Chile, there is also evidence of greater cohesion among government legislators in the Chamber of Deputies (Campos-Parra and Navia Reference Campos-Parra and Navia2017; Figueroa and Navia Reference Figueroa and Navia2021; Toro Maureira Reference Toro Maureira2007) and the Senate (Campos-Parra and Navia Reference Campos-Parra and Navia2022). Ideology also affects how legislators vote (Cox and McCubbins Reference Cox and McCubbins2005; Morgenstern Reference Morgenstern2004). However, since ideology and party membership often coincide, the two concepts are usually treated interchangeably.

The time horizon

Among the rules that govern how legislators are elected and what they can do in office, the length of the term and the moment when the election occurs – concurrently with presidential elections or not – are often mentioned as the most salient factors that impact party discipline (Cheibub Reference Cheibub2007, 170; Shugart and Carey Reference Shugart and Carey1992, 86–87; Shugart Reference Shugart1995, 328, 337) . When legislative terms have the same duration in both chambers and legislative and presidential elections are held concurrently, we would not expect to find differences between the two chambers in the president’s ability to influence government legislators. In turn, when senators have a longer term than deputies and part of the Senate is elected non-concurrently with the president, we expect ruling coalition senators to show less cohesion than ruling coalition deputies, whose term begins and ends with that of the executive. The president has fewer tools to influence government senators who did not get elected on the coattails of a victorious presidential candidate. Therefore, we expect that the executive will be more effective in uniting ruling coalition deputies (elected alongside the president) than ruling coalition senators (half of whom were elected before the president) in the votes that matter most for the executive. This leads to our first hypothesis:

In strong presidential systems with concurrent elections, presidential elections predominate over legislative elections. Popular presidential candidates often have a positive impact on the performance of their party’s legislative candidates. In their honeymoon period, presidents enjoy greater popular approval which, in turn, makes government legislators more inclined to approve their bills – and opposition legislators less inclined to block them (Alemán and Calvo Reference Alemán, Calvo, O’Donnell, Tulchin and Varas2008). Indeed, this honeymoon period is ‘the most propitious time for presidential initiatives’ (Fleisher et al. Reference Fleisher, Bond and Wood2008, 10).

In turn, if immediate re-election is not permitted, the executive’s influence will weaken as the end of the presidential term approaches. In this so-called lame-duck period, legislators show increasing loyalty to the more popular presidential candidates rather than the outgoing president (Barrett and Eshbaugh-Soha Reference Barrett and Eshbaugh-Soha2007): when there are term limits for the president but not legislators, at the end of the term, government legislators will show more interest in their own re-election – or other ways of advancing their political careers – than in loyalty to a lame-duck president. In the case of the chamber that is not re-elected in its entirety concurrently with the president, the electoral calendar should have less effect than in the chamber that ends its term together with the president. Therefore, we postulate two hypotheses that suggest a greater effect of the electoral calendar in the Chamber of Deputies than in the Senate:

The content of the bill

The content of the bill being debated also matters. Naturally, parties care more about voting cohesively in issues they deem more important. The importance of a vote depends not only on valence but also on the issue in question and the vote’s importance in the legislative process (Hagopian et al. Reference Hagopian, Gervasoni and Moraes2009). Valence, in turn, depends on the party’s ideology and the political situation (Hagopian et al. Reference Hagopian, Gervasoni and Moraes2009). For example, for a green party, environmental questions will be more of a valence issue than for a conservative Catholic party. However, the objective elements have to do with how advanced a bill is in its passage and its institutional importance. Therefore, while valence depends on values and priorities and will be different for different parties, a bill’s importance will be assessed equally by all parties.

Similarly, at least in the short term, a vote on a more advanced bill will be more important than one on a bill that is still in the early stages. In turn, a bill that implies constitutional reforms or reforms to more important laws – like constitutional organic laws in Chile – should also be more important for all parties. Also, if we assume that the executive has more tools to influence government legislators compared to opposition legislators, we should expect to see greater cohesion among the former in the chamber, which is more susceptible to the executive’s influence when voting on bills that are more advanced and require a qualified majority for their approval. Hence, our fourth hypothesis is:

Bicameralism in the Chilean legislative proces

In parliamentary democracies, research focuses on the main chamber. In presidential systems with bicameral legislatures, however, upper chambers often have greater powers, yet Latin American research concentrates on lower chambers, leaving the upper houses underexplored. Moreover, there are even fewer comparative studies on the behaviour of both chambers in Latin American presidential democracies.

Presidential democracies with bicameral legislatures present unique challenges when seeking to assess the way in which both chambers behave, because not only do they need to struggle collectively with the executive for influence, but they also fight for influence with each other (Tsebelis and Money Reference Tsebelis and Money1997). In many Latin American presidential democracies, fluid party systems and strong presidents result in weakened legislatures. Hence, unsurprisingly, Latin American legislatures have been referred to as reactive, as opposed to proactive executives (Cox and Morgenstern Reference Cox and Morgenstern2001). However, in democracies with institutionalised party systems, and where the Congress plays an important role in the law-making process, it is important to understand the balance of powers within the legislature.

For this reason, the case of Chile is a useful case to assess differences in the behaviour of legislators in both chambers, as that country has a presidential democracy where the legislature exhibits important attributions and has an increasing influence in the law-making process (Alemán and Navia Reference Alemán, Navia, Alemán and Tsebelis2016; Huneeus and Berríos Reference Huneeus and Berríos2005; Martínez and Dockendorff Reference Martínez, Dockendorff, Llanos and Marsteintredet2023; Mimica and Navia Reference Mimica and Navia2024). Although presidents have strong powers, including the use of presidential urgencies (Huertas-Hernández and Palanza Reference Huertas-Hernández and Palanza2024), the legislature has tools to check the power of the president and exert its oversight prerogatives (Alemán and Micozzi Reference Alemán and Micozzi2022; Alemán et al. Reference Alemán, Micozzi and Ramírez2018; Mimica et al. Reference Mimica, Navia and Cárcamo2024).

The general perception is that, in Latin American bicameral systems, the upper chamber is more important than the lower chamber, given how its members are elected, the length of their term, and the greater requirements for standing as a candidate (Llanos and Nolte Reference Llanos and Nolte2006, 20; Sánchez et al. Reference Sánchez, Nolte and Llanos2005). Following the classification proposed by Lijphart (Reference Lijphart1990), Mariana Llanos and Detlef Nolte (Reference Llanos and Nolte2003) identify the Chilean system as having a high level of inconsistency and a low level of symmetry compared to other Latin American bicameral systems.

In Chile, certain types of bills must start their passage in a particular chamber. Initiatives involving general pardons and amnesty bills must be presented first to the upper chamber (Article 65 of the constitution, second paragraph). Those related to taxes or involving the fiscal budget and recruitment to the lower chamber (Article 65, second paragraph). The legislative process is sequential, requiring that a bill be debated first in one chamber and then reviewed by the other (Soto Velasco Reference Soto Velasco2015). When submitted to a chamber, bills are assigned to a permanent commission for their debate. The passage of bills comprises up to four stages. The First Constitutional Procedure (primer trámite) corresponds to their debate in the chamber to which they were presented, the Chamber of Origin. If it approves the bill, it then goes to the other chamber, the Reviewing Chamber, where it begins its Second Constitutional Procedure. If approved in this stage, it is then sent to the president who can sign it into law or present a veto.

If the Reviewing Chamber rejects a bill, it is submitted to a Conference Committee (Comisión Mixta), formed by an equal number of senators and deputies, to resolve the discrepancies between the two chambers. If the Reviewing Chamber modifies a bill, it is returned to the Chamber of Origin, initiating its Third Constitutional Procedure. If the Chamber of Origin accepts the changes, the bill goes to the president for its signing or veto. If the Chamber of Origin rejects the changes, the bill again goes to a Conference Committee. If the Conference Committee resolves the discrepancies and the bill in its resulting form is approved by both chambers, it is dispatched to the president. Otherwise, unless the executive insists, the bill is filed as rejected.

During the period studied, the electoral rules for both chambers were similar. Senators and deputies were elected under the rules of an open-list proportional representation in two-seat districts. Thus, although there were more districts (60) in the Chamber of Deputies than in the Senate (19), the district magnitude for all districts was the same. The parties formed stable coalitions, and each coalition presented two candidates per district. Senators, who are elected for 8 years, have a longer time horizon than deputies, who have a four-year term. Half the Senate is re-elected in each presidential cycle.

Due to this institutional design, final negotiations also tend to occur in the upper chamber. For example, the annual fiscal budget law – which is introduced by the president and can only be modified by the legislature by making cuts, not additions – must begin its passage in the lower chamber. This gives the Senate an advantage in the negotiations because the lower chamber must make its preferred cuts explicit before the upper chamber does. As the upper chamber has broader attributions and is often where the most important negotiations for a bill’s approval take place, it is particularly important to avoid generalisations based on studies of the lower chamber alone, the Chamber of Deputies.

Before the breakdown of democracy in 1973, the Senate underpinned the stability and survival of the Chilean political system (Agor Reference Agor1971). In the post-1990 presidential era, the Senate has also played an essential role, due both to its composition, which included non-elected members in the early years (Londregan Reference Londregan2000, 82), and because it has more powers than the Chamber of Deputies (Campos-Parra and Navia Reference Campos-Parra and Navia2022). In a study on Senate committees, John Londregan (Reference Londregan2000) found that ideology interacts with the valence that legislators attribute to certain issues. In his research on legislative behaviour in various Latin American countries, Morgenstern (2004: 41) analysed 583 votes in the Chilean Chamber of Deputies between 1997 and 1999 and 94 votes in the Senate between 1992 and 1998, finding more cohesion among senators than deputies. A study of 313 votes and sponsorships of legislation in the Senate between 2004 and 2006 found greater voting cohesion among government legislators but greater cooperation between the coalitions in sponsoring bills (Alemán Reference Alemán2008). However, apart from Eduardo Alemán (Reference Alemán2008), few studies systematically analyse the roll-call voting patterns of Chilean senators (Campos-Parra and Navia Reference Campos-Parra and Navia2022).

By contrast, studies on deputies’ behaviour in the lower chamber abound (Alemán et al. Reference Alemán, Micozzi and Ramírez2018; Campos-Parra and Navia Reference Campos-Parra and Navia2017; Faúndez Caicedo et al. Reference Caicedo, Vicente, Navia and Pérez Aburto2024; Figueroa and Navia Reference Figueroa and Navia2021; Visconti Reference Visconti2011). Its ability to act as a counterweight to the executive is partly explained by the legislative discipline within parties and coalitions (Toro Maureira Reference Toro Maureira2007). The legislative cartel theory applies in Chile only when there is coordination between the executive and the legislators of the majority coalition or when opposition legislators obtain the support of government legislators for their amendments to bills (Toro-Maureira and Hurtado Reference Toro-Maureira and Hurtado2016). Studies of the behaviour of legislators in the lower chamber who introduce and advance various bills (Dockendorff Reference Dockendorff2019, Reference Dockendorff2021; Escobedo Aránguiz and Navia Reference Escobedo Aránguiz and Navia2020) underline the specific incentive structure at work: to serve their constituencies (Dockendorff Reference Dockendorff2020); to represent women’s interests (Dockendorff et al. Reference Dockendorff, Gamboa and Aubry2022); and to form networks (Alemán and Calvo Reference Alemán and Calvo2013; Le Foulon Moran Reference Le Foulon Moran2020).

In general, bills introduced in the upper chamber are more likely to be approved than those introduced in the lower chamber. Indeed, out of the 2,950 bills that, having begun their passage in the lower (76.8%) or the upper (23.2%) chamber, were approved in their first constitutional step between 1990 and 2018, 29.5% of those introduced in the lower chamber were approved by the upper chamber while 37.1% of those introduced in the Senate were approved in the Chamber of Deputies. By contrast, 46.9% of the bills introduced in the lower chamber were modified in the upper chamber compared to 40.0% of bills introduced in the upper chamber that were modified by the lower chamber. In the same period, 2.8% and 3.8% of the bills introduced in the lower and upper chamber, respectively, were rejected by the other chamber and sent to a Conference Committee.

The literature also asserts that legislators elected in smaller districts represent more particular interests, while those elected in larger districts represent more general interests (Heller Reference Heller2007, 259; VanDusky-Allen and Heller Reference VanDusky-Allen and Heller2014, 719). If upper chamber members are elected in a national district, their concerns should be more general than those of lower chamber members elected in local districts. In Chile, before the reform of the electoral system in 2015 (Gamboa and Morales Reference Gamboa and Morales2016), 38 senators were elected under open lists in 19 two-seat districts, while 120 lower chamber members were elected (also under an open list system with proportional representation) in 60 two-seat districts. In other words, lower chamber legislators generally represented significantly smaller electoral units than upper chamber legislators. On average, each upper chamber district comprised 3.2 lower house districts.

In each 4-year legislative term until 2006, electoral majorities in the upper chamber were distorted by the presence of non-elected senators and, between 1998 and 2006, the life seats awarded to former presidents. The Senate’s importance in the institutional system – particularly during the period when the appointed senators altered electoral majorities – is at odds with the scant information that exists about its performance in the legislative process and senators’ roll-call voting behaviour. From 2006 through to 2018, all 38 members of the Senate were democratically elected. Table 1 shows the composition of the Chilean Congress between 2006 and 2018. The Concertación/Nueva Mayoría (NM) coalition obtained a majority in the three elections that took place in this period. However, the resignation of two upper chamber legislators from the governing parties and their subsequent move to the right-wing benches gave the right-wing opposition a majority in the Senate.

Table 1.

Composition of the Chilean Congress, 2006–2018

Source: Compiled by the authors using data from www.servel.cl.

Previous studies have identified several determinants of roll-call voting behaviour that relate to legislators’ party membership, the rules of the legislative process and the effect of the electoral cycle. Using Nominate scores for the period between 1997 and 2000 (which corresponds to the last year of one legislative term and 3 years of the subsequent term) one study showed more cohesion among right-wing parties than among left-wing parties in the Chamber of Deputies (Alemán and Saiegh Reference Alemán and Saiegh2007). We review the association posited by that study with data from three complete legislative terms.

Figure 1 shows the voting cohesion of the Concertación/NM coalition and that of Alianza/Chile Vamos (CV) in both chambers of Congress in the three legislative periods studied, using the Rice and Unity indicators. As expected, the Rice indicators have higher values than the Unity indicators. Both coalitions show more cohesion in the periods when they controlled the executive; namely, 2006–2010 and 2014–2018 in the case of the Concertación/NM coalition; and 2010–2014 in the case of Alianza/CV. The Rice indicator shows no clear pattern of difference concerning coalition cohesion in the two chambers. The reason is probably that when senators do not want to vote with their coalition, they simply skip the vote; an absence that is not picked up by the Rice indicator. The Unity indicator, on the other hand, treats absence as a vote against the majority behaviour of each coalition and shows the coalitions’ voting cohesion as higher in the lower than in the upper chamber. Between 2006 and 2018, party membership influenced voting cohesion in the Chamber of Deputies but appears to have had less effect in the upper chamber. Although greater cohesion is apparent in both coalitions in both chambers when the respective coalition also controls the presidency, the results systematically show that there was less cohesion within the coalitions in the upper than in the lower chamber.

Figure 1.

Legislative Cohesion by Political Coalition in Chile, 2006–2018

Source: Compiled by the authors using data from www.camara.cl and www.senado.cl.