The Determinants of Conference Committee Formation

The conference committee is “that point in the bicameral legislative process where the separate actions of the House and Senate on similar measures are reconciled” (Longley and Oleszek 1989, 2). Since both chambers must accept or reject the conference committee agreement in its entirety (Tsebelis and Money 1997; Alemán and Pachón 2008), conference committees can be viewed as a pared-down form of unicameralism (Levmore Reference Levmore1992; Grossman Reference Grossman2006) or a de facto third chamber (Longley and Oleszek 1989; Grossman Reference Grossman2006). The power of a conference committee increases further if it has discretion as to the conditions in which it can reach its agreement (Shepsle and Weingast 1987a).

Given that conference committees result from bicameral conflicts, the balance of powers between chambers will affect the conference committee’s powers (Longley and Oleszek 1989; Van Beek Reference Van Beek1994; Tsebelis and Money 1997, Ferejohn Reference Ferejohn1975). Lijphart (Reference Lijphart2000) distinguishes between symmetry and congruence between chambers. They are symmetrical when consensus between them is required to modify the status quo, thereby giving them institutional veto powers (Tsebelis Reference Tsebelis2006), and they are congruent when they share the same method of election and basis of representation (Lijphart Reference Lijphart2000). Strong bicameralism is characterized by symmetry between the chambers and incongruence in the method of election, as, for example, in the United States (Lijphart Reference Lijphart2000).

One powerful argument in favor of bicameralism is the stability in policy and policymaking, since the two chambers represent an obstacle to short-lived and accidental majorities (Riker Reference Riker1992). In the classic theory, bicameralism is an institutional design for mixed regimes that are formed to represent different interests. This theoretical framework allows room for dissent between chambers. Each chamber represents different interests and must also justify its existence to avoid a rubber-stamp condition. This leads to conflict and competition between the chambers (Diermeier and Myerson 1999), but it also generates incentives for both chambers to bargain to solve their collective action problem in promulgating bills.

Bicameral conflicts are greater when party majorities differ in the two chambers (Longley and Oleszek 1989; Tsebelis and Money 1997; König et al Reference König, David and Proksch2013; Heller and Branduse 2014). This would imply more conference committees when the chambers have different majorities. Minority governments in the United States avoid conference committees and prefer “amendment trading” (Ryan Reference Ryan2011). In Germany, governments might be willing to risk their bills’ being referred to a conference committee when the chambers have different majorities (König et al Reference König, David and Proksch2013). When the same party controls both chambers, alternative methods of resolving bicameral differences can be used to avoid conference committees (Krehbiel Reference Krehbiel1987; Longley and Oleszek 1989).

The rules that determine the legislative process and the composition of conference committees are crucial to understanding the negotiations and the policy outcomes (Tsebelis and Money 1997). In the United States, the agreements reached by a conference committee can refer only to those parts of a bill on which the two chambers differ (Tsebelis and Money 1997; Alemán and Pachón 2008). Institutional arrangements determine the power of conference committees. Generally, the members of a conference committee also sit on the permanent committee that initially reported on the bill (Tsebelis and Money 1997; Alemán and Pachón 2008). However, in negotiations, conference committee members can promote their own interests, those of their parties (Longley and Oleszek 1989), or those of their chamber (Gailmard and Hammond 2011). Party members are restricted in representing positions in favor of their party’s preferences, as they cannot form interchamber ad hoc coalitions with legislators from the same party (Lazarus and Monroe 2007).

In addition, the president’s legislative powers may cause differences between Congress and the executive (Longley and Oleszek 1989) and lead the two chambers to engage in a noncooperative game (Diermeier and Myerson 1999) or a bicameral conflict (Martin Reference Martin, Martin, Saalfeld and Strom2014). The existence of bicameralism accompanied by the possibility of a presidential veto increases the incentives for each chamber to use available tools to maximize its influence over the lawmaking process (Diermeier and Myerson 1999). In presidential systems with symmetrical bicameralism, such as the United States and Chile, there is a form of de facto tricameral system, since the agreement of the president and both chambers is required to modify the status quo (Riker Reference Riker1992; Tsebelis Reference Tsebelis2006; Uhr Reference Uhr, Rhodes, Binder and Bert2006). This type of bicameralism is conducive to conciliation mechanisms between the two chambers, because when the two chambers are symmetrical in powers and incongruent in representation, discrepancies between them are more likely.

When the president’s party or coalition controls both chambers, the passage of bills should be smoother, and thus there should be fewer conference committees. Since conference committees are used to settle differences between chambers, their frequency ought to increase when chambers are controlled by different coalitions. In presidential systems—especially when the president controls the legislative agenda—a bigger seat share support for the president in each chamber should reduce the formation of conference committees. Also, the bigger the asymmetry in seat share support for the president in both chambers, the more likely a conference committee is to be formed. Assuming that bicameral conflicts are higher when the president’s seat share support differs in both chambers, our first hypothesis thus suggests that as the seat share support for the president increases in each chamber, a bill is less likely to go to a conference committee.

In the normal legislative process, legislators can use amendments strategically to hamper approval of undesired projects (Longley and Oleszek 1989). Amendments serve to increase a chamber’s influence and can be used as bargaining chips in a conference committee. Thus, a large number of amendments may indicate disagreements or represent a bid to force a conference committee as an opportunity to negotiate the terms of the bill (Longley and Oleszek 1989). To be sure, the process of amendments between chambers in the United States, which can be extended indefinitely, hinders bicameral agreement and may lead a chamber to propose a conference committee (McQuillan and Ortega 1992). For Londregan (Reference Londregan2002), the decisionmaking powers of a conference committee are determined by how much a bill is modified by amendments. If, for example, a conference committee debates a substitute amendment, it would be rewriting the entire bill.

In sum, legislators can use amendments to modify a bill’s undesired aspects and hamper its passage and as bargaining chips in conference committee negotiations (Longley and Oleszek 1989). Thus, in the normal legislative process, the number of amendments can be understood as a proxy for the complexity of the bill. The more complex bills are, the more likely each chamber will produce a diversion of the bill. In the case of Chile and its sequential legislative process, amendments serve only as an indicator of a bill’s complexity, not as a bargaining chip. After all, each chamber gets the bill after the other chamber has approved it. Conference committees, in turn, get the bill after approval from one chamber and rejection from the other. Thus, the presence of amendments serves as a proxy for the complexity of the bill. Since we expect that more complex bills are more likely to go a conference committee, our second hypothesis asserts that bills to which amendments have been approved are more likely to go to a conference committee.

The Chilean Legislative Process

In Chile’s strong presidential system, the president has a number of legislative powers, including the power to introduce bills and the exclusive power to introduce bills on the fiscal budget, the creation of public services, public sector salaries, pensions, and social security norms. Legislator-initiated bills must be introduced in their own chamber (the chamber of origin) while executive-initiated bills can be introduced in either chamber; exceptions are bills referring to budget, tax, or recruitment matters, which must go first to the Chamber of Deputies; and amnesties and pardons, which must go first to the Senate. Budget laws are introduced by the executive and are reviewed by a special budget committee.

Executive-initiated bills are more likely to become laws and to have a speedier passage. Presidents have a greater capacity to implement their legislative agenda during their first year in office, and bills related to economic and financial policy have a greater chance of approval (Alemán and Navia 2009). However, high voting thresholds required for many bills force presidents to bargain with the opposition. The legislature has sufficient tools at its disposal to block, obstruct, or slow down the president’s legislative powers (Aninat Reference Aninat2006; Alemán and Navia 2016).

Chile was once classified as having a symmetrical and incongruent bicameralism (Schiavon Reference Schiavon2004). After the elimination of nonelected senators in 2006, the incongruence between the two chambers became less marked (Heller and Branduse 2014). In each chamber, bills are debated both in committees and on the floor (Londregan 2000; Aninat Reference Aninat2006). Bills are referred to one or more permanent committees for review (Londregan Reference Londregan2000). Chamber of Deputies and Senate standing committees have 13 and 5 members, respectively. Committees can propose amendments to a bill and then prepare a report for a floor vote (Londregan Reference Londregan2000).

This institutional design, with many veto players, a bicameral legislature, and a president with broad powers, is behind Chile’s marked political stability (Siavelis Reference Siavelis2002; Aninat et al. Reference Aninat, John, Patricio, Joaquín, Scartascini, Spiller, Stein, Tommasi and Washington2010). Given the institutional framework’s presidential bias, ideological moderation, combined with legislators’ tight networks of political relationships, has had a positive effect on the relevance of Congress in the lawmaking process (Aninat Reference Aninat2006; Alemán Reference Alemán2009).

Given this institutional design, the chamber in which a presidential bill is first introduced could affect the bill’s chances of success. Following Diermeier and Myerson (1999), as chambers compete against each other to influence the legislative process, if the president has a majority only in one chamber, introducing the bill in that chamber would guarantee it a conference committee even if the other chamber rejects the bill. Yet if a bill passes one chamber, a conference committee is guaranteed, so the revising chamber always has incentives to compromise and pass a revised version of the bill—as a counterproposal to the original chamber. Thus, regardless of which chamber the bill is introduced in, the revising chamber always has incentives to modify the bill rather than to reject it outright. Thus, while we control for the chamber of origin of the bill, there is no reason to expect that a bill will be more likely to make it to a conference committee if it is first introduced in the Senate or the Chamber of Deputies.

Following Alemán and Navia (2016), we use the origin of the bill—president-initiated or legislator-initiated—as an explanatory variable. Following a distributional theory approach, since the institutional setup is tilted in favor of the president and thus in favor of president-initiated bills, we would expect that bills initiated by legislators that manage to pass both chambers will be more likely to end up in conference committees than president-initiated bills. After all, president-initiated bills have higher priority in the legislative process, and the president has more tools to negotiate with both chambers in defense of priorities than individual legislators do. In turn, if there is bicameral conflict, a legislator-initiated bill could find more resistance in the revising chamber, since the legislator who sponsored the bill is a member of the other chamber and cannot formally negotiate in the committee debates with legislators of the revising chamber to move the bill forward. Therefore the third hypothesis is that legislator-initiated bills are more likely than executive-initiated bills to go to a conference committee.

Laws with special voting thresholds suggest a greater likelihood of conflict due to the difficulty of aligning broad majorities in the two chambers and overcoming the higher number of veto players (Tsebelis Reference Tsebelis2006). There are four types of vote thresholds in Chile. The higher the threshold, the more important the bill. Chile’s binominal electoral system, in force between 1989 and 2017, made it difficult for the coalitions to obtain a majority in Congress, making negotiation with the opposition crucial, and may, therefore, have accentuated bicameral conflicts.

Following Alemán and Navia (2009), the importance of the bill should affect its chances of making it to a conference committee. Other studies have shown that conference committees tend to review the most important legislation (Shepsle and Weingast 1987b; McQuillan and Ortega 1996; Lazarus and Monroe 2007). The creation of a conference committee depends on the breadth, complexity, and prominence of the legislation being discussed (Van Beek Reference Van Beek1994). Therefore, when the voting threshold reflects the importance of a bill, more important bills should be more likely to generate bicameral discrepancies. Thus, our fourth hypothesis is that bills that require a special voting threshold for their approval are more likely to go to a conference committee.

Bicameral conflicts are also more likely to emerge when the president has a stronger interest in passing the bill quickly—regardless of whether it is a president-initiated or a legislator-initiated bill. When presidents want to see a bill promulgated quickly, they are more willing to accept compromise. When there is bicameral conflict, the president might need to broker a different compromise with each chamber.

When the president controls the legislative agenda, the president signals his or her interest in a bill by giving it priority. That signal can turn the bill into a battlefield for bicameral conflict, especially when one chamber is friendlier to the president or shares the president’s interest in seeing the bill passed. That signaling by the president should lead the chamber least interested in the bill to delay its passage by forcing a conference committee.

In Chile, the president exercises control over the legislative agenda through motions of “urgency” to debate a bill (Siavelis Reference Siavelis, Mainwaring and Shugart1997; Londregan Reference Londregan2000; Siavelis Reference Siavelis2002; Nolte Reference Nolte2003; Aninat Reference Aninat2006). There are three types of urgency: simple, suma, and immediate discussion—which means that, respectively, the corresponding chamber has 30, 15, or 6 days to vote on a bill. Urgency motions are often introduced and withdrawn, however, because they have become instruments to prioritize bills rather than effective deadlines for the chamber to take a final vote on a bill. When the president introduces an urgency motion, there are two immediate consequences: the bill goes onto a fast track, taking precedence over other bills in the legislative process; and the president signals the importance he or she gives to the bill. Thus, the respective chamber can use that signal to attempt to extract concessions from the president to ensure a speedy passage of the bill. When there are bicameral conflicts, the concessions extracted by one chamber will be different from those extracted by the other chamber, and therefore our fifth hypothesis postulates that the executive urgency indicator makes a bill more likely to go to a conference committee.

Other studies look at whether bills are targeted for local or particularistic interests, but in Chile, spending bills can be initiated only by the executive. Therefore, rather than look for particular interests, which the Chilean legislative process significantly limits, we check for the content of the bill using the committee the bill is referred to. The Finance and Constitution Committees are deemed the most important by legislators in Chile, and they review many bills (Carey Reference Carey2002; Nolte Reference Nolte2002).

Methodology

To evaluate the hypotheses, we use binary logistic regression models. We care about the laws enacted that went to a conference committee. However, since some bills that were never promulgated also went through a conference committee, we include those as well to account for the determinants of conference committee formation.

We compiled information on the 2,103 laws promulgated between March 11, 1990, and March 10, 2018—six presidential terms and seven four-year legislative periods—and the additional 80 bills that went through a conference committee and were not promulgated as laws. A conference committee was formed in 482 of the 2,183 bills (22.1 percent). Most conference committees were formed after the original chamber rejected the changes made by the revising chamber (81.7 percent). Only one of these cases happened when the revising chamber rejected the bill passed by the first chamber. Naturally, some bills passed by one chamber before 2018 were rejected, revised, or approved by the revising chamber—and, in some cases, promulgated—after 2018. Some bills introduced before March 2018 have made it to conference committees since—or will, eventually—but like all studies on legislative output, our sample is truncated by the need to establish a starting and ending point.

We collected information from the Senate website on the bill’s chamber of origin, whether it was executive-initiated or legislature-initiated, whether the bill had any urgency motion, the referral committees, the date the bill was introduced, and the date it was promulgated (or the date for last legislative action on bills that were not promulgated). We also collected information on the history of each law (Historia de la Ley). We coded whether a conference committee was formed, whether the bill had amendments, and the required voting thresholds for each bill.

The dependent variable is whether a bill went to a conference committee. To test the first hypothesis, we used the seat share of the president’s multiparty coalition in the Senate and Chamber of Deputies, respectively. Presidential and legislative terms coincided in 1990–94 and since 2006. In 1994–2006, there were presidential elections every six years (at the end of 1993 and 2000) and legislative elections on a four-year calendar (at the end of 1993, 1997, 2001, and 2005). We coded the seat share for presidents in every legislative term. We also coded significant changes in the composition of Congress in off-election years. In Bachelet’s first government (2006–10), the Concertación ruling coalition lost its majority in both chambers when two senators and five deputies switched parties and gave the right-wing opposition a majority in the Senate and half the seats in the Chamber of Deputies for the rest of the period (2008–10). Though there was significant stability in the seat share support for the president in both chambers during the period, as this is one of the strongest variables in explaining the success of legislative initiatives, we included it in our models as well.

The second hypothesis was tested with a dummy variable that indicates if a bill had approved amendments from the executive or legislators. Since amendments— or indicaciones, as they are called in Chile—can differ in their specificity, counting the number of approved amendments is not a good indicator of how much a bill has changed. Moreover, the number of approved amendments is not easily obtainable, as it requires going through the debate on each bill and identifying—and often guessing—how many amendments were approved. In the dataset, 26 percent of the bills were promulgated without amendments. Thus, we use the presence of amendments as an indicator of the complexity of a bill and a sign of a potential bicameral conflict. After all, approved amendments are added to a bill in one chamber and must be accepted by the other for the bill to avoid a conference committee. This variable takes the value of 1 if a bill had approved amendments.

The third hypothesis was tested using a dummy variable that indicates if the bill was executive-initiated (0) or legislator-initiated (1). The fourth hypothesis was tested with a dummy variable for each type of voting threshold. There are four thresholds, ranging from a majority of members present (simple majority) to qualified majority (majority of sitting legislators), constitutional organic laws (a four-sevenths majority of sitting legislators), and constitutional reforms (either three-fifths or two-thirds majority of sitting legislators). Since special thresholds apply to organic constitutional laws or constitutional reforms, the threshold is a good proxy for the importance of the bill. In the inferential analysis, we included only the bills that had qualified majority and organic constitutional law thresholds, as the number of constitutional reform bills was too small. Our reference category is bills that require no special voting threshold. In models not shown here, we also alternatively coded voting thresholds as a dummy variable (special thresholds and no special thresholds), and the results were similar to those shown here.

For the fifth hypothesis, we included a dummy variable for each type of urgency (simple, suma, and immediate discussion). Our reference category is the bills that did not received a presidential urgency (19 percent of all cases, as shown in table 1).

Table 1. Frequencies of the Independent Variables

Source: Authors, using data from the websites of the Senate, the Chamber of Deputies, and the Library of the National Congress (BCN).

Table 1 shows the frequencies of the independent variables. Of all bills that passed in both chambers, 482 were referred to a conference committee (22.1 percent). Only 80 were not promulgated because there was no agreement in the conference committee, one of the chambers rejected the conference committee agreement, or the president vetoed it and Congress did not overrule the veto. There were fewer conference committees when the government had a majority in both chambers. In addition, 25.5 percent of bills with amendments went to a conference committee, as compared to 12.5 percent for those without amendments. Bills initiated by legislators have a slightly higher chance of making it to a conference committee (24.6 percent) than those initiated by the executive (20.8 percent). Only 12.2 percent of bills requiring a simple majority for approval went to a conference committee. For constitutional organic bills and bills requiring a qualified majority, the bills going to a conference committee rose to 36.7 percent and 29.4 percent, respectively. Bills with presidential urgency requests were slightly more likely to end up in conference committees.

We also included control variables that are common in studies of legislative affairs in Chile (Alemán and Navia 2009; Toro Maureira et al. 2010; Visconti Reference Visconti2011; Campos-Parra and Navia 2017). A dummy variable indicates if a bill was reviewed by the Constitution Committee or the Finance Committee. Table 2 shows the committees for the first and second referral of all the bills. The Finance Committee reviewed almost a third of the bills and the Constitution Committee reviewed 12.5 percent.

Table 2. First and Second Committee Referrals in the Chamber of Origin for All Bills, 1990–2018

^a Some bills were concurrently referred to more than one committee. In those cases, a “united committee” (Finance and Labor, for example) was formed to review the bill. We count each committee separately.

We also control for the chamber of origin of the bills: 0 for the Senate and 1 for the Chamber of Deputies. As discussed above, when a chamber passes a bill, a conference committee is guaranteed—and the revising chamber thereby has incentives to modify the bill rather than to reject it outright. For the electoral cycle, we included dummy variables if the bill made it to a conference committee or to the president’s desk in the first or last year of the presidential term. During its first year, the government and Congress are disposed to reach agreements; the acceleration of the legislative agenda during a government’s last year may imply more conflicts in Congress.

We also included a dummy variable for each administration since 1990, using the second Bachelet administration (2014–18) as the reference category. We included the number of days it took for the bill to reach the conference committee or the president’s desk (for those laws that did not go through a conference committee). The longer it takes for the bill to move forward, the more complex or the less support a bill probably has.

As shown in table 3, conference committees are more frequent for bills introduced in the Senate and in the last year of a presidential term. In the Aylwin, Frei, and Lagos administrations, when there was a progovernment majority in the Chamber of Deputies but not in the Senate, the percentage of conference committees was higher than under the three subsequent governments. The percentage of conference committees was lower in both Bachelet administrations (2006–10 and 2014–18).

Table 3. Frequencies of Control Variables

Source: Compiled by the authors using data from the websites of the Senate, the Chamber of Deputies, and the Library of the National Congress (BCN).

The data are ordered according to the government in power when the corresponding law was enacted. Some bills went to a conference committee just before the end of a presidential term and became law under the subsequent administration, producing a lag during the period immediately before a change of power. However, this did not affect the results.

Table 4 shows the composition of Congress for each legislative term. In Chile, political parties have formed stable multiparty coalitions since democracy was restored in 1990. The center-left Concertación coalition (Nueva Mayoría since 2014) governed in all but one term. The center-right Alianza coalition governed between 2010 and 2014. Though the parties have retained their individual identity, their voting record reflects high levels of coalition cohesion (Alemán and Saiegh 2007; Toro Maureira 2007; Campos and Navia 2017; Argote and Navia 2018).

Table 4. Government Coalition Seat Share in Congress by Legislative Period, 1990–2018

^a In 2008–2010, due to the resignation of two senators and five deputies from the PDC, the Concertación lost its majority in both chambers.

Source: Compiled by the authors using data from the Electoral Service (SERVEL).

Inferential Analysis

We used a binary logistic model in which the dependent variable is a dummy that distinguishes those bills that went to a conference committee from the rest of the laws promulgated. To avoid multicollinearity between the independent variables, we used four regression models. For example, the majority in each chamber variable (models 1 and 2) is related to the variable of the government under which a law was enacted (models 3 and 4) because presidential and legislative periods coincide. Similarly, we used the voting threshold variables in only two models (models 1 and 3) because the relevant information was available only for 1,979 bills.

The models in table 5 support three of the five hypotheses and offer partial support for another. Our first hypothesis states that the use of conference committees increases as the president’s coalition seat share in each chamber diminishes. The results show that this is the case for the composition of the Senate but not for the Chamber of Deputies. The probability of a conference committee forming is negatively affected by the seat share the president has in the Senate and positively affected by his or her seat share in the Chamber of Deputies. This finding is not fully surprising. The probability of a conference committee increases when the president has a majority in only one chamber. There might also be an ad hoc explanation for the opposite effects of seat majorities in both chambers. Between 1990 and 2006, the presence of nonelected senators artificially gave right-wing parties a majority in the Senate. Since the presidents in those years belonged to the center-left Concertación coalition, the presence of nonelected senators (most of whom were right of center) blocked that coalition from transforming its electoral majority into a seat share majority in the Senate.

Table 5. Logistic Regression on Determinants of Conference Committee Formation

***p<0.01, **p<0.05, *p<0.1. Standard errors in parentheses.

Not surprisingly, during those terms—as shown in table 3 above—there was a higher proportion of laws that went through a conference committee than in subsequent terms. An alternative explanation might be with the type of bills that were being promulgated. After all, in the early years of democracy, many key bills that set the foundation for the new institutional system were promulgated. Yet because the rate of bills that went through conference committees decreased under the two Bachelet administrations (2006—10 and 2014–18), when she had majority support in both chambers, and increased in the right-wing Piñera administration (2010–14) when he lacked majority in both chambers, we are inclined to believe that the seat share the president has in Congress is behind the occurrence of conference committees more than the content of the bills.

In line with hypothesis 2, a bill that is amended also has a greater chance of referral to a conference committee. Amendments introduced and approved by one chamber may be rejected by the other chamber, giving rise to the bicameral discrepancies that trigger a conference committee. Since the bills can be amended by legislators, by the president, or both, the presence of approved amendments signals more complexities in the passage of a bill. Without specific information on who introduced the amendments that were approved, it is difficult to determine what caused the bicameral discrepancy, but there is a strong impact of the presence of approved amendments on the formation of a conference committee.

Belying hypothesis 3, the type of bill—whether it was presented by the president or by legislators—is not a significant indicator (except in one of the four models). This result, which shows that the probability of a conference committee is the same for both types of bills (president- or legislature-originated), does not support the extensive literature on the powerful legislative role of the executive branch in Chile.

With respect to hypothesis 3, legislator-initiated bills that became laws were not more likely to end up in a conference committee. Since the legislative agenda is controlled by the president, many of legislator-initiated bills that move forward are those to which the president assigns urgency. Thus, a potential bicameral conflict might not be appropriately perceived in looking at the original sponsor of the bill. After all, bills that get the president’s support—regardless of whether the president initially introduced them—are the most likely to move forward in the legislative process.

Confirming hypothesis 4, bills with a special voting threshold—constitutional organic laws or bills requiring a qualified majority—are more likely to go to a conference committee. The larger the number of legislators that must be aligned behind a bill, the more complex its approval. and therefore the larger the number of conference committees.

There is also evidence in support of hypothesis 5. Bills with presidential urgency requests are more likely to end up in conference committees. The results can be interpreted as the legislators taking advantage of the president’s signaling the importance of a bill for the administration. The more urgencies the president introduces to a bill, the clearer the priority the president gives to that bill. Thus, as the president wants the bill to be passed more expeditiously, each chamber uses the conference committee as a negotiating tool with the executive, as well as with the other chamber. Another interpretation would be that Congress, facing presidential pressure for a quick passage when the president has a majority in both chambers, might decide to refer a bill to a conference committee as a means of expediting its passage.

We also estimated a model (not shown here) on the effect of urgencies only for legislator-initiated bills to assess if the president’s use of an urgency affects a bill’s probability of being referred to a conference committee. The effect of urgencies in legislator-initiated bills is similar to that for all bills. This might be explained by the fact that when legislator-initiated bills receive an urgency, it is usually because the content of the bill is also a priority for the president. Therefore the best indicator of the importance of a bill for a president is not whether the president introduced the bill but the president’s decision to assign it an urgency motion.

In the case of the control variables, bills entering Congress through the Senate go to a conference committee more often than those starting their passage in the Chamber of Deputies. The theory of bicameral conflict also offers an insight to explain this pattern. The less powerful Chamber of Deputies might exert influence by modifying a bill that entered through the Senate or by rejecting the amendments to a bill introduced by the Senate. An alternative explanation could be that bills that can enter only through the Senate (pardons or amnesties) are more prone to go to a conference committee—but out of the 2,183 bills that passed in both chambers, only 11 were about amnesties or pardons.

Bills reviewed by the Constitution Committee are more likely to go to a conference committee, and bills reviewed by the Finance Committee are less likely to do so. The results suggest that there is greater consensus in Chile about bills on economic and financial matters and greater discrepancy on political and institutional design issues. The frequency of conference committees is not conditioned either by the type of bill or by the point in the electoral cycle at which the bill could be referred to a conference committee.

Similarly, the longer the processing time for the bill, the more likely the bill is to go to a conference committee. This might reflect either the bill’s complexity or the obstructive power that Congress might use to slow down a bill that reflects the president’s priorities. Additionally, if a bill approved by one chamber in one legislative period reaches the other chamber only in the subsequent period, the differences in party lineup between one legislature and the other may cause friction between the two chambers and complicate the bill’s passage.

In addition, the dummy variables for each government show significant results in the first three presidential terms after the restoration of democracy, when conference committees were more frequent. This can be interpreted as a sign that distrust, the legislators’ lack of experience, or the presence of nonelected senators caused more friction between the chambers in the early years of democracy. Moreover, the Concertación governments of Aylwin, Frei, and Lagos had a majority in the Chamber of Deputies, but because of the nonelected senators, the right-wing opposition had a majority in the Senate. Nonelected seats were introduced so that senators loyal to the military dictatorship could influence lawmaking, and it would therefore not be surprising if they explain the larger number of conference committees during those years—and might help explain why, when the president lacks a majority in the Senate, the chances of a conference committee increase. After nonelected seats in the Senate were eliminated starting in 2006 (by the 2005 constitutional reforms), the number of conference committees dropped significantly. A divided Congress, in which the president always had minority support in the Senate, might be the reason for most of the conference committees seen during the first three administrations.

However, the Piñera administration (2010–14) also lacked a majority in both chambers, but it was not reflected in more conference committees. It is therefore possible to speculate that the Aylwin, Frei, and Lagos administrations, knowing they had a majority in the Chamber of Deputies, risked presenting bills closer to their ideal point, forcing rejection by the opposition-controlled Senate, with the resulting creation of conference committees. The Piñera administration would, on the other hand, have promoted policies further from its ideal point because, given its lack of a majority in both chambers, it could not afford to risk the possibility of a conference committee where the opposition would most likely have a majority.