Partial Presidential Vetoes and Executive–Legislative Bargaining: Chile, 1990–2018

Abstract

Defined as a credible threat that strengthens the bargaining position of the executive, presidential vetoes, widely understudied, carry a stigma of confrontation between state powers. But under some institutional setups, partial vetoes can be an additional step in the executive–legislative bargaining process. After a discussion of whether partial vetoes are a proactive legislative tool or a bargaining tool to induce executive–legislative cooperation, we test four hypotheses using the 2,346 bills introduced in Chile between 1990 and 2018 that reached a vetoable stage. We identified 97 partial vetoes (4.2 percent) and one total veto. Presidents are more likely to veto bills with more complex legislative processes and when they have stronger support in at least one chamber, but more popular presidents do not veto more bills. As most presidential vetoes in Chile are partial, they are an additional executive–legislative bargaining step in the lawmaking process rather than evidence of hyperpresidentialism.

The presidential veto is a dissuasive power. The threat of using it gives presidents a powerful advantage over the legislature, but it can potentially increase tensions between the executive and the legislature. When presidents have additional legislative tools—like agenda-setting, gatekeeping, and amendment powers, and the means to speed up debate on their preferred bills—a veto power might seem redundant. In fact, the existence of a veto might feed the perception that presidents enjoy too much power. Often, the stigma of excessive presidentialism in Latin American presidential democracies feeds calls to strengthen the legislature.

But the veto itself might play a more nuanced role in executive–legislative relations. When presidents can choose between issuing a total or a partial veto—with the latter allowing the president to delete, modify, or add additional text to a bill—partial vetoes can be an additional step in executive–legislative bargaining (Alemán and Tsebelis 2005; Tsebelis and Rizova 2007). A partial veto is not a zero-sum game. It is an opportunity for additional negotiations as it affords the president the option to partially modify a bill, granting the legislature the option to accept the modification with a simple majority or insist on the original bill with a qualified majority. Thus, presidents can use the veto as a negative and a positive power (Tsebelis and Rizova 2007, 1174). While total vetoes might reflect excessive presidentialism, partial vetoes are tools that can foster executive–legislative bargaining.

To explore how partial vetoes work in presidential systems and determine under what conditions presidents will use them, we rely on the case of Chile, where presidents can choose between full and partial vetoes. Between 1990 and 2018, there were 2,346 bills that reached a vetoable stage. Presidents issued partial vetoes on 97 bills (4.2 percent) and a total veto only once. Anecdotal evidence points to partial vetoes being an additional step in executive–legislative bargaining. For example, in September 2005, when the ruling left-wing coalition and the right-wing opposition in Congress agreed on a comprehensive constitutional reform—the largest since democracy was restored in 1990—President Ricardo Lagos (2000–6) issued a veto with 29 observations that modified, replaced, and added provisions to the compromise bill. Congress approved 28 of those observations, and the government and opposition celebrated the enactment of that major constitutional reform.

After discussing the determinants of the use of presidential vetoes and distinguishing between total and partial vetoes, we postulate four hypotheses on the issuance of a partial veto. We expect that vetoes are more likely to be issued when there is more bargaining between both powers. Thus, the hypotheses are associated with the electoral calendar, presidential approval, seat-share support for the president in the legislature, and the complexity of the lawmaking process for each bill. We then present the methodology, the case, and the results. We finish by discussing the implications of our findings for the understanding of partial presidential vetoes as an additional bargaining tool in executive–legislative relations.

Types of Presidential Vetoes

The origin of the presidential veto dates to the Roman Empire, when magistrates used the intercessio—interposition—to block legislation introduced by patricians (Spitzer 1988). In the modern era, Alexander Hamilton ([1788] 2018) introduced it into the United States Constitution, justifying it in The Federalist No. 51 and No. 73 as a protection for the executive against the legislature. In Latin America, after the veto was introduced by Simón Bolivar in the Bolivian Constitution of 1826, it was eventually adopted by all presidential systems, though the forms and applications of the veto power have varied over time and across countries (Alemán and Tsebelis 2005; Martínez 2018). In the United States and Latin America, the veto has been a tool for a proper system of checks and balances between the executive and the legislature. Precisely because “men are not angels” (as Madison put it in The Federalist No. 51), each branch of government would be given powerful weapons “to resist the potential attacks and encroachments of the other branches” (Gargarella 2013, 58). But rather than the hoping for the actors to use those weapons, the incentive is that the actors, seeing that the other is also armed, will be more inclined to bargain.

Presidential vetoes are rare events. A veto-power actor is someone whose agreement is required to change the status quo (Tsebelis 2002). The presidential veto is a reactive power to block a bill passed by the legislature (Alemán and Schwartz 2006). As a last instance, it is seldomly used (Rohde and Simon 1985), as its use normally carries high political costs (Londregan 2000). In presidential systems where the executive has limited legislative attributions, the veto is a powerful tool to influence the legislative process (Shugart and Carey 1992; Shugart and Mainwaring 1997). According to the US Senate veto count page, out of 1,518 presidential total vetoes issued between 1776 and 2021, only 112 have been overridden by Congress. And, out of the 34 vetoes issued since 2001, only six have been overridden.

The threat of a total veto can be understood as a sequential process, whereby the congress modifies the content of the bill before passage to avoid a presidential veto on the entire bill (Alemán and Tsebelis 2005; Cameron 2000; Cameron 2009; Palanza and Sin 2013). For that reason, the threat of a total veto is sometimes seen as conducive to cooperation between the executive and the legislature (Alemán and Schwartz 2006, 117). However, once issued, the total veto makes it impossible for the executive and legislature to move beyond its zero-sum-game nature and continue bargaining over the content of a bill (Guenther and Kernell 2021). Thus, while the total veto gives the president a negative power (Alemán and Tsebelis 2005), the partial veto gives the executive positive power to bargain with the legislature (Tsebelis and Rizova 2007).

When issued to amend or modify legislation, a partial veto can be a proactive legislative tool that presidents use to make the change to the status quo closer to their ideal point (Alemán and Tsebelis 2005). In Argentina, for example, partial vetoes are more commonly used than total vetoes, and the latter have higher chances of being rejected (Palanza and Sin 2013). The lawmaking rules and the specificities of how partial vetoes work under different institutional settings impact both the chances of presidents resorting to partial vetoes and the response by the legislature to the issuance of partial vetoes. When a bill receives a set of partial vetoes, the bargaining process between the executive and legislature is not reduced to a zero-sum game. Negotiations can occur between the president and legislature over which partial vetoes will be approved and which will be overridden.

The type of partial presidential veto used triggers different reactions from congress. For example, when the US Congress passes a bill and then adjourns before the 10-day period for the president to issue a veto, the executive can issue a so-called pocket veto, which cannot be overridden because Congress is not in session (Sartori 1997; Stuessy 2019). Though widely used in the nineteenth and twentieth centuries, the pocket veto was last used by President Bill Clinton (1993–2001). The line-item veto—present in several states of the union, often restricted to budget bills (Alemán and Tsebelis 2005), and ruled unconstitutional by the Supreme Court for the federal government (Indridason 2011; Spitzer 1997)—is common in several Latin American countries, where it is known as a partial or suppressive veto.

The features of partial vetoes vary across presidential democracies in Latin America (Alemán and Schwartz 2006; Alemán and Tsebelis 2005) and elsewhere. In Kenya, the constitution gives the president negative veto powers but also allows the executive “to make positive legislative proposals” to bills passed by the legislature, which has resulted in strengthening the bargaining power of the executive (Ochieng 2023, 80). In the Philippines, presidents have total and partial veto powers, but in South Korea, the executive can only use total vetoes (Croissant 2003, 76).

Partial or constructive veto powers allow presidents to add, amend, or replace parts of a bill approved by the legislature. In Argentina, Brazil, Uruguay, and Ecuador, presidents can use a constructive partial veto that can be approved—rather than overridden—by the legislature. In Bolivia, Chile, Costa Rica, and Mexico, legislatures first vote to approve a presidential veto (that is, to approve the modifications or deletions that the president has introduced) and, in case the veto has not been approved, vote to override it (Aninat et al. 2006). If the legislature fails to approve or override, the bill is enacted without the content vetoed or modified by the executive. In general, a qualified majority is required to override a veto that ranges from three-fifths to two-thirds, but the vote required to approve a veto is often a simple majority (Alemán and Schwartz 2006; Alemán and Tsebelis 2005). That makes the partial veto a useful bargaining tool in executive–legislative negotiations.

The wide variance in vetoes has generated growing interest in understanding the implications of different types of veto in Latin America (Basabe-Serrano 2017, 17–22). Presidents who have bill initiation power or can amend legislation introduced by others are more likely to have congress approve their presidential amendments to avoid the threat of a veto (Alemán and Schwartz 2006; Alemán and Tsebelis 2005). In Argentina, presidents are less likely to veto presidential bills than legislator-initiated bills (Palanza and Sin 2013).

The factors that influence the issuance of a veto differ by country (Palanza and Sin 2013). The electoral cycle impacts the frequency of conflicts and, thus, the chances that a president will issue a veto (Rohde and Simon 1985; Wolley 1991). In the United States, presidents have incentives to issue total vetoes early in their term to establish a reputation (McCarty 1997). Because presidents also tend to be more popular in their honeymoons, Congress is more likely to pass bills closer to the president’s ideal point at the start of the term (McCarty 1997; McCarty and Poole 1995). As time goes on, Congress adopts stronger negotiating positions (Rohde and Simon 1985). Presidents lose power toward the end of the term and thus are less likely to risk issuing a veto that can be overridden.

The Determinants of the Issuance of Partial Vetoes

The types of partial vetoes include positive and negative provisions and features that allow the executive to delete, add, or replace one or several sections of bills. Yet, as partial vetoes allow the president and the legislature to further engage in the bargaining process over the content of a bill passed by the legislature, the determinants of the issuance of a partial veto should be similar to those that influence the rest of the legislative bargaining process, which include the electoral calendar, the number of legislative steps for a bill, presidential approval, and the seat-share support for the executive in the legislature (Alemán and Navia 2009; Saiegh 2011).

In some presidential democracies with multiparty systems and broader veto power tools, like Argentina, presidents also issue more vetoes early in their terms to build reputation (Palanza and Sin 2013). Precisely because there will be bargaining over the content of the partial veto, the president might want to establish a reputation to improve their bargaining position. Although Congress should also want to establish a reputation, since previous findings on Argentina report an association between the electoral calendar and the issuance of vetoes, we include a hypothesis that links the issuance of vetoes to the electoral calendar. As presidents seek to establish a reputation when they first take office, they should be more likely to issue vetoes earlier in their term. Likewise, as they become lame ducks, they would be less likely to issue partial vetoes in the last year of their term. Thus, our first hypothesis is as follows:

H1 Presidents are more likely to issue a partial veto in the first year of their term and least likely to do so in their last year in office.

Particularities of the legislative process for each bill also impact the issuance of a veto. In Argentina, landmark legislation is more likely to get vetoed (Palanza and Sin 2014). As presidents prefer some articles of a bill to be worded differently (Rohde and Simon 1985), the divergence in preferences might induce them to issue vetoes to build reputation and signal differences with the median legislator (McCarty 1997). Since presidents can influence the content of the bill before it reaches their desk via amendments (Cameron 2000), the more legislative steps it takes for a bill to pass, the more likely that differences between the executive and the median legislator will become apparent. When more steps are required, those who oppose the bill will have more opportunities to modify it or erase content from the bill. That will create incentives for presidents to issue partial vetoes—like additive or substitutive vetoes—that change the wording of the bill to bring its content closer to the president’s ideal point. It is true that presidents can bargain throughout the legislative process, but a more complex bargaining process might also induce the president to use the partial veto to fix rough edges that were left after an agreement was brokered. A more complex lawmaking process generates more opportunities for bargaining. Thus, we postulate a second hypothesis:

H2 The more legislative steps a bill goes through, the more likely a president is to issue a partial veto.

In the United States, presidential approval is inversely correlated with the ability of Congress to override a presidential veto (Rohde and Simon 1985; Woolley 1991). More popular presidents are more likely to impose their views on Congress. In turn, when presidents are less popular, Congress will be more inclined to pass legislation that the president might oppose, knowing that the president will be more reluctant to veto it. But if this is true, by backward induction, since everyone knows how popular presidents are, a higher presidential approval will not result in more presidential vetoes, as the legislature will refrain from passing bills that can be easily vetoed by a popular president. For that reason, it makes sense that other studies find no effect of presidential approval on the issuance of vetoes, as is the case in Argentina (Palanza and Sin 2013; 2014).

Still, since the literature has widely reported that presidential approval has consequences and more popular presidents are more effective at passing bills and influencing the lawmaking process, we expect that a higher presidential approval will increase the chances of the president vetoing a bill. Thus, our third hypothesis postulates the following:

H3 Higher presidential approval increases the chances of presidents issuing a partial veto.

The frequency of vetoes and the ability of Congress to override a veto depend on which party controls the legislature (Lee 1975). When presidents enjoy more legislative support, they will be more likely to issue vetoes (Woolley 1991) and avoid an override (Colomer and Negretto 2003; Rohde and Simon 1985). Elsewhere, veto overrides are more likely in landmark legislation and when the executive’s party lacks control in at least one chamber (Palanza and Sin 2013), but the issuance of vetoes is not impacted by the seat share for the president in Congress or by presidential approval (Palanza and Sin 2014). Even when they know their vetoes will be overridden, minority presidents might choose to issue a veto for other purposes, like position taking or to polarize the electorate over certain issues. For example, a president can issue a veto when a bicameral legislature has different ideal points on changing the status quo and the president might want to pit one chamber against the other. Thus, fear of being overridden is not the only concern for the president when deciding whether to veto a bill. Still, when presidents have stronger seat-share support in Congress, they should be more comfortable issuing vetoes. We propose a fourth hypothesis:

H4 A higher seat-share support in Congress for the president makes it more likely that the president will issue a partial veto.

The Presidential Veto in Chile

Partial vetoes are common in presidential democracies—with the US presidential system being a notable exception—but studies of the issuance of partial vetoes are scant and mostly “anecdotal” (Tsebelis and Rizova 2007, 1175). We fill that gap by conducting a comprehensive study on the issuance of partial vetoes in Chile’s stable presidential democracy over a 28-year period and seven presidential terms.

Chile’s strong presidential system has been labeled as exaggerated presidentialism (Siavelis 2002), hyperpresidentialism (Nohlen 2011), superpresidentialism (Shugart and Carey 1992), and a strong presidentialism with moderate presidents (Siavelis 2000). But others have highlighted the institutional design that promotes executive–legislative cooperation (Alemán and Navia 2009; Aninat et al. 2006). Still, Chile is classified as the seventh-most strongly presidential democracy among 18 Latin American countries (Basabe-Serrano 2017). As Chile is normally mentioned as a case of strong presidentialism, analyzing the issuance of partial vetoes allows us to test the claim of hyperpresidentialism in that dimension and to enlighten the debate over the use of partial vetoes in presidential democracies in general.

Previous works have underlined the institutional design features conducive to executive–legislative bargaining and cooperation (Aninat et al. 2006). In advancing their agendas, presidents have opted for cooperation and consensus building with the legislature more than confrontation (Siavelis 2002), especially on bills that require high voting thresholds (Aninat et al. 2006), due to the informal structures of power in the lawmaking process (Arana Araya 2013; 2015; Soto 2015b, 207).

The presidential veto power was first introduced in the Constitution in 1818. The supreme director (the executive) had an eight-day window to object to a law passed by the Senate. The Senate could insist on the bill to force its enactment. In the 1822 Constitution, the director of the state could object to a bill within 15 days of its passage and send an amended bill back to the respective committee for an up-or-down simple majority vote. If the bill was also voted favorably in the other chamber, it had to be enacted. In the 1823 Constitution, the executive had bill initiation power and the Senate had veto power. If the Senate vetoed a bill, the supreme director could insist on it. If the Senate vetoed it a second time, the National Chamber was convened to settle the issue.

The presidential veto was introduced in the 1828 Constitution, the first that formally established the office of president (Davis 1985). Since then, all constitutions have included a presidential veto. Article 53 of the 1828 Constitution granted the president the power to return a bill to the chamber of origin with objections within 10 days of receiving it. The objections needed to be voted favorably by both chambers. Article 56 established that if the objections were not approved, the bill would die and could not be reintroduced in the same year. The 1833 Constitution extended the deadline for a presidential veto to 15 days after the president received the bill and gave the president the power to issue a full veto on a bill, banning Congress from overriding it. Article 46 gave the president partial veto power, requiring Congress to approve those changes. Subsequently, that total veto power was removed from the 1833 Constitution, but the president retained a suspensive veto (Martínez 2018). In the 1925 Constitution, the president had 30 days to issue an amendatory veto. If the amendments were not approved, each chamber could insist on the original bill by a two-thirds majority and the president was mandated to enact the bill.

The daily logs of the Ortúzar Commission, which drafted the 1980 Constitution, summarize the debates on the presidential veto. In session 350 on April 12, 1978, the commission discussed a constitutional reform proposed by former president Jorge Alessandri (1958–64) that established a partial presidential veto (Comisión de estudio de la nueva Constitución 1978, 562–74). The commission declined that option because a partial veto could render the entire bill irrelevant. In session 353 on April 19, 1978, the commission decided against the creation of a bicameral conference committee charged exclusively with revising presidential vetoes (623–52). In the final Ortúzar Commission report, the veto kept the same wording as in the 1925 Constitution. Article 73 (Article 70 before the 2005 reforms) discusses presidential veto power in the 1980 Constitution.

How the Presidential Veto Works in Chile

The legislative process in Chile induces executive–legislative bargaining. Bills can be introduced by the president or by legislators. After approval in the chamber of origin, the bill goes to the revising chamber. That chamber can approve, reject, or modify it. If approved, the bill goes to the president, who can enact it or veto it. If the revising chamber modifies it, the bill goes back to the chamber of origin for approval or rejection. If the revising chamber approves, the bill goes to the president. If a bill is rejected by either chamber, a conference committee comprised of five members from each chamber drafts a compromise version. If that compromise version is approved by both chambers, it goes to the president’s desk. Debate in each chamber is conducted mainly at the committee level, where any member can introduce amendments. If a committee votes against the bill, the floor of each chamber can reverse the decision of the committee (Mimica et al. 2022).

A bill approved by both chambers can be vetoed by the president. The president has total and partial (or amendatory) veto power. There are three types of partial/amendatory vetoes: suppressive, substitutive, and additive (Soto 2015a). The suppressive veto allows the president to strike out an article, or part of an article, of a bill passed by Congress. If Congress fails to override the veto with a two-thirds majority and the suppressive veto is not approved by a majority in the legislature, the sections vetoed by the president cannot be included in the law to be enacted (Soto 2015a, 174). This happened when President Patricio Aylwin (1990–94) issued a suppressive veto to Law 19,180 that modified the Constitutional Organic Law of municipal governments.

The substitutive veto replaces an article or a section of an article in a bill passed by Congress. This type of veto is also subject to the same voting rule conditions as the suppressive veto. President Lagos issued a substitutive veto to Law 19,948 that modified the rules for replacements for lost or stolen national identification cards.

The additive veto gives the president the power to add a phrase or other content to a bill passed by Congress. For example, President Michelle Bachelet (2006–10) added an article to a bill that banned the practice of requesting a signed blank check before a patient could be admitted to a private clinic. Substitutive and additive vetoes can only address issues directly related to the content of the bill.

Article 128 of the Constitution empowers the president to issue a total veto of a constitutional reform. If Congress overrides it by a two-thirds vote, the president can convene a national referendum to decide whether the president’s or the congress’s position will prevail. As discussed above, there has only been one occasion where a president issued a total veto and Congress failed to overrule that veto. If the president issues a partial veto of a constitutional reform, a three-fifths or two-thirds majority is required to approve the president’s veto, otherwise the sections included in the presidential veto will not be a part of the reform enacted. Congress can always override a partial veto with a two-thirds majority.

Presidents must issue a veto within 30 days of receiving a bill. Once vetoes are issued, presidents normally do not withdraw them (Soto 2015a, 204). The Organic Constitutional Law of Congress (Law 18,918) regulates the procedure for each chamber to respond to a presidential veto. In case the president of a chamber chooses not to accept the veto, the chamber can reverse that decision by majority rule. If the veto is not accepted by the chamber, the bill will be enacted without the articles vetoed by the president. Once the veto is accepted by both chambers, each chamber votes on each specific article vetoed and on the entire presidential veto. The vote threshold required to approve the veto is the same as that required to pass the bill. For an override—which means that the original content of the bill will be sent back to the president—a two-thirds majority is required in each chamber. If only one chamber approves the changes made by the executive or rejects the veto, the bill will be enacted without the content vetoed or modified by the president (Soto 2015a, 175).

Sometimes, a formal executive–legislative communication takes place in the final stage of the lawmaking process (Soto 2015b, 207). For some bills that require high voting thresholds, after taking a final vote, the legislature is required to issue a formal inquiry to the president asking about the president’s intention to issue a veto. The president formally responds, indicating what sections of the bill will be subject to a veto. Two cases where the legislature formally inquired about the president’s intention and the president responded indicating that a veto would be issued were on Law 19,130—which modified the Organic Constitutional Law of Municipalities, vetoed by President Aylwin—and the 2005 constitutional reform (Law 20,050), vetoed by President Lagos. In turn, cases where the legislature inquired about the president’s intention to veto a law and the president responded signaling no such intention were Law 20,840 in 2015, which established a new electoral system for legislative elections, and Law 20,960 of 2016, which established voting rights for Chileans living abroad. In both cases, presidents honored their word and did not issue a veto.

The partial presidential veto can be a tool that helps executive–legislative bargaining. As a presidential veto can only be rejected by a two-thirds majority in both chambers, but the content of an amendatory veto only needs to be approved by the same majority that approved the original bill, the partial veto helps presidents to improve their bargaining position. If the legislature knows that it cannot override a presidential veto, it will bargain with the president to reach a compromise before sending a bill to the president’s desk or, alternatively, when the bill has already been sent and the president is considering issuing a partial or amendatory veto.

In fact, some authors consider the presidential veto as a “procedure that has had a marginal relevance in executive–legislative relations” (Huneeus and Berríos 2005, 359). The veto “works and has a dissuasive power that is not used frequently” (PNUD 2014, 348), and “induces legislators to obtain the president’s support to pass their bills” (Soto 2015b, 206). Reflecting on the lawmaking dynamics of the 1990–2006 period, former cabinet minister and senator Edgardo Boeninger (2007, 65), one of the key actors in Chilean politics between 1990 and 2006, argued that “the structure of the legislative power is a combination of instances of cooperation, bargaining and veto.”

The partial veto has also been used after the legislature formally requests the president to use it to correct problems in an advanced bill (Huneeus and Berríos 2005). As Huneeus (2009, 260) argues, “in some occasions, the presidential vetoes were requested by the legislators when they found errors in the bills passed by congress.” Huneeus backs this claim, citing information conveyed to him by Carlos Carmona, a lawyer and legislative expert for several center-left Concertación governments. Carmona also later served on the Constitutional Tribunal. Based on that information, Huneeus claims that “between 1990 and 2002—a situation that has remained unchanged—the president used the veto only 45 times. In 32 of those occasions, Congress approved the veto. In 7 occasions, the veto was partially approved and in 4 cases, the veto was rejected. In those 11 latter cases, the congress insisted with the original bill in 3 occasions and in the other 8 cases, Congress chose not to insist. The content of those vetoes shows that they cover different issues and not necessarily important ones” (259–60). ¹ The information reported by Huneeus (2009) has been reproduced in latter studies (Berríos and Gamboa 2006; Martínez 2018). One of those studies argues that “the presidential veto is used not only to resolve conflicts between the executive and the legislature, but it is often used after a request by the legislators to correct errors in a bill” (Berríos and Gamboa 2006, 114).

Yet, prior studies also report that presidential vetoes have occasionally been contentious and have been used when the bargaining process fails (Soto 2015b, 207). Discussing a presidential veto issued in 2006 under President Bachelet, Berrios and Gamboa (2006, 114) write that “in the law-making process of a bill that regulates scientific inquiry on human subjects (Law 20,120), the executive issued an additive veto to include a norm, but the veto was declared inadmissible by the President of the Senate, Eduardo Frei. The justification was that the norm did not address the basic content (idea matriz) of the bill … it was the first time since 1990 that a presidential veto was declared inadmissible.” Berríos and Gamboa conclude that “the use of this mechanism shows that Congress can strengthen its bargaining position with the executive in the law-making process” (114).

In her study of different types of presidential vetoes between 2006 and 2016, Palanza (forthcoming) reports that bills that receive a presidential urgency motion and those that cover more important issues—according to the coding of her research team—are more likely to be vetoed. As Martínez (2018, 90) argues, when discussing the veto issued by President Sebastián Piñera (2010–14) to the budget law in late 2012, the fact that presidents seldomly make use of this tool does not mean that the tool does not induce the legislature to compromise with the executive.

In sum, previous studies report that the presidential veto is part of an institutional design that fosters executive–legislative cooperation. Since presidents can issue partial vetoes, the issuance of vetoes becomes another step in the process of executive–legislative bargaining rather than an expression of excessive presidential powers and prerogatives.

Methodology

To assess whether partial presidential vetoes are more likely to be issued when there is an active bargaining process between the executive and the legislature, we use the case of Chile, where presidents can issue partial vetoes. From Alemán et al. (2022), we obtained a dataset with 2,948 bills that passed the first legislative step (approval by the chamber of origin) between 1990 and 2018. We identified 2,346 bills that received a final vote in the revising chamber. Notice that the revising chamber can pass a bill and send it to the president, modify it and send it back to the chamber of origin, or reject it and trigger the formation of a conference committee. But since there was the option of passing the bill, the president could announce a credible veto threat. For that reason, we use the final vote on the revising chamber to determine whether bills were susceptible to a veto threat. From the Library of the National Congress and press reports, we identified 98 bills that received a presidential veto.

Our dependent variable is the issuance of a presidential veto. In the 98 vetoes issued, the president used a total veto only once, in 1991—shortly after the restoration of democracy. The bill passed by the legislature on a free-trade zone in the northern town of Iquique became redundant as a larger and more comprehensive bill on the same subject had been recently approved. President Aylwin vetoed the entire bill and Congress failed to override the veto. In the other 97 cases, the president used a partial veto, sending a revised version of the bill with text added, deleted, or modified.

The independent variable for H1 is the year in the electoral cycle when the bill reached the last legislative step. The legislative year goes from March 11 to March 10 in the subsequent year. We recorded the year the bill arrived at the president’s desk, with dichotomous indicators for the first year (521 bills) and the last year (551 bills). We expect a positive impact on the issuance of vetoes for bills that reached the president’s desk in the first year and a negative impact for those bills that did so in the last year.

The independent variable for H2 is the complexity of the legislative process. As a bill needs to pass in one chamber before it can be reviewed by the other chamber, there must be at least two legislative steps (a bill passes in both chambers) before the bill can go to the president’s desk. Many bills go through three steps (approved by one chamber, modified by the other, and approved by the original chamber). Some bills go through four or more steps (conference committee, constitutional tribunal review, or presidential veto after being approved by both chambers). We code bills that went through two steps as 0, those that went through three steps as 1, and those that went through more than three steps as 2. Among the 2,346 bills used in the study, 812 (34.6 percent), 701 (29.9 percent), and 833 (35.5 percent) went through two, three, and four legislative steps, respectively. Among the 98 bills vetoed, three (3.1 percent) bills went through three steps, and the other 95 (96.9 percent) went through four steps.

The independent variable for H3 is presidential approval. We averaged the presidential approval reported by the polls conducted by the Center for Public Studies (CEP, in the Spanish acronym)—two or three times per year—for each year of a presidential term. We coded the presidential approval for the most recent year a bill was reviewed by the legislature.

The independent variable for H4 is the seat-share support for the president in the legislature. As indicators for this variable, we used a dummy if the president had a majority in both chambers, the seat share for the president in the Chamber of Deputies, and the seat share for the president in the Senate.

Following the literature that associates the issuance of vetoes with tensions between the executive and the legislature, we control for whether the bill required a special voting threshold, if the bill received amendments (indicaciones), and if the bill received a presidential urgency motion. Prior work on the legislative process in Chile (Alemán and Navia 2009; Mimica et al. 2022) has highlighted how those three variables signal bargaining between the executive and legislature. We also control for bills introduced by the executive, as prior works on the legislative process in Chile underline the predominance of presidential over legislator-introduced bills in the lawmaking process.

We estimated probit models on all bills that reached a point where a presidential veto was a real option, with the dependent variable being a dummy for the bills that received a veto. Others have studied the incidence of vetoes issued every year using Poisson regression models (Shields and Huang 1995). We study, instead, the probability that every vetoable bill is vetoed by the president. Table 1 shows the descriptives statistics for the dependent and independent variables.

Table 1. Descriptive Statistics for the Dependent and Independent Variables

Source: authors, with data from the Library of the National Congress, CEP polls, and the Senate website (www.senado.cl).

The Issuance of Presidential Vetoes in Chile, 1990–2018

Table 2 shows the number of vetoed bills by each possible decision of the revising chamber. Only 1.1 percent of bills passed by the chamber of origin and approved by the revising chamber were vetoed. Vetoes were more likely on bills that were either rejected or modified by the revising chamber, and thus, went to a conference committee or went back to the chamber of origin, respectively. This is consistent with the expectation that bills that go through more legislative steps are more likely to be vetoed. Since most of the laws enacted in the period originated in presidential bills, most of the partial vetoes (60 out of 97) were also issued on bills initially introduced by the president.

Table 2. Bills Vetoed According to the Decision of the Revising Chamber, Chile, 1990–2018

Source: authors, with data from the Library of the National Congress and the Senate website (www.senado.cl).

Table 3 shows the 320 cases where there was formal communication between Congress and the president on the latter’s veto intentions. This communication took place, as required by the bylaws of the legislature, after the legislature approved some qualified majority bills. In 53 bills, the president announced the intention to veto. In 267 cases, the president responded indicating no intention to veto. When Congress had not previously inquired about the president’s intention to veto, presidents issued vetoes on only 46 bills. But since the congressional inquiry was initiated after the congress had taken a final vote on a bill, the president’s response could no longer affect the content of the bill.

Table 3. Formal Legislature–Executive Communication on Presidential Veto Intentions in Chile, 1990–2018

Source: authors, with data from the Library of the National Congress and the Senate website (www.senado.cl).

Table 4 shows the vetoable bills by presidential term, the number of bills introduced in every term that were vetoed, and the number of vetoes issued in every term. Presidents can issue vetoes to bills introduced in previous terms. For example, the bill on digital television (Bill 6190-19) introduced by President Bachelet in 2008 was vetoed by President Piñera in November of 2013, and the veto was ratified by Congress in March 2014, days before Piñera left office. The bill was enacted in May of 2014, under the second Bachelet administration (2014–18) (Verdugo 2014). Out of the 98 vetoes issued since 1990, 50 (51 percent of all vetoes) were issued in the first two governments (1990–2000). In the last two administrations (2010–18), only 18 vetoes were issued. Still, since the veto is a theoretically relevant presidential power and its importance does not depend on presidents making regular use of vetoes, it is worth assessing under what conditions vetoes are issued.

Table 4. Bills that Reached a Point Where a Presidential Veto Was an Option, Chile, 1990–2018

Source: authors, with data from the Library of the National Congress and the Senate website (www.senado.cl).

Results

We estimated probit models on the decision by the president to veto a bill. Table 5 shows the results. Notice that predicting that vetoes never happen would be correct in 95.8 percent of the cases, as only 4.2 percent of the bills were vetoed. Thus, we must analyze the results of the models with caution as presidential vetoes are rare occurrences. In turn, the low incidence of vetoes suggests that the variables with statistical significance have robust explanatory power.

Table 5. Probit Models on the Determinants of the Issuance of Presidential Vetoes in Chile, 1990–2018

Source: authors, with data from the Library of the National Congress, CEP polls, and the Senate website (www.senado.cl).

Notes: standard errors in parentheses. *** Significant at 0.001; ** significant at 0.01; * significant at 0.05. The dependent variable is dichotomous: 1 if the president issued a veto, 0 otherwise.

Hypothesis 1 postulates that presidents issue more vetoes in their honeymoon period and the fewest vetoes in their last year in office. The models show results inconsistent with that hypothesis. An explanation for this result, which goes against evidence provided at least in the case of Argentina, is that, since presidents are more popular in their honeymoons, the legislature might be more likely to modify a bill to make it acceptable to the president before taking a final vote on it. In turn, in their last year, as their term is ending, knowing that they are less likely to impact the behavior of legislators, presidents are less likely to issue vetoes, fearing that a veto override might turn into an embarrassing defeat. Between 1990 and 2018, the electoral calendar had no instances of presidential veto issuance. This goes against what the literature has identified for other presidential systems with partial vetoes, like Argentina, and for systems with full vetoes, like the United States. Our findings suggest that the electoral calendar is not always a good predictor of the issuance of partial vetoes.

Hypothesis 2 suggests that bills that have more legislative steps are more likely to receive a veto. We use three indicators for the complexity of the legislative process: the number of legislative steps, an indicator for bills that had three steps, and an indicator for bills that had more than three steps. The results are consistent with the expectations of H2. The more complex the legislative process, the more likely a bill will be vetoed. The case of the 2005 constitutional reform bill, discussed above, illustrates how bills that go back and forth between chambers and also go through a conference committee are more likely to receive a presidential veto. The impact of this variable is significant, regardless of the indicator we use to assess the complexity of the legislative process. In the period, vetoes in Chile were associated with the complexity of the lawmaking process for each bill.

An alternative explanation for the use of presidential vetoes could be that the congress and the president use the veto as a shortcut to speed up the passage of a bill. Thus, if a bill is approved in one chamber, the other chamber could refrain from modifying the bill and send it directly to the president for the executive to issue a partial veto that would make the bill more acceptable to the executive and to the second chamber. But since there were no cases where the executive vetoed a bill passed by the first chamber and then passed by the second chamber without modifications, we can safely discard that alternative use of partial vetoes.

Moreover, as the chances of presidential vetoes being issued increases with the number of legislative steps, the evidence helps to discard the notion that vetoes are used as a shortcut at any stage of the legislative process. The longer it takes for a bill to pass both chambers and make it to the president’s desk, the more likely it is that the president will issue a partial veto.

Hypothesis 3 suggests that higher presidential approval increases the chances of the president vetoing a bill. However, the models show that presidential approval does not impact the issuance of a veto. The reason behind this might be methodological. Since the only comparable poll on presidential approval for the entire period is that conducted by CEP, twice or thrice per year, there is limited variance in presidential approval in the dataset. Monthly polls for presidential approval in Chile are only available since 2006—but since there are so few vetoes in the entire period, using only part of the period would make it even less likely to identify the determinants of presidential vetoes. Thus, we take this result with caution.

Hypothesis 4 postulates that higher seat-share support for the president in Congress will make it more likely for presidents to issue a veto. The models offer contradictory evidence. When we use a dichotomous indicator, presidents are more likely to veto when they have majorities in both chambers. In two of the four models, presidents are also more likely to veto when they have higher support in the Chamber of Deputies, but in all four models, they are less likely to veto when their support in the Senate is higher. This might be because when presidents have majorities in both chambers, presidents are more confident that their vetoes will not be overridden. When the chambers have different majorities, presidents might be less likely to veto precisely because the bill approved by Congress reflects a bicameral compromise.

The control variables show results that are consistent with expectations. The higher the vote threshold, the less likely a veto is to be issued. This makes sense, as bills that require a higher vote threshold in the legislature are closer to the margin required to override a presidential veto. The positive impact of the approved-amendments variable is also consistent with expectations. In turn, bills that receive a presidential urgency motion are more likely to be vetoed. Interestingly, presidents issue fewer vetoes to bills introduced by themselves than to bills introduced by legislators. The results are consistent with the claim that the issuance of partial vetoes has more to do with the type of bill and the lawmaking process than with the electoral calendar or presidential approval. Thus, presidential vetoes should be associated more with the attributes of the lawmaking process than with the ups and downs of the political cycle. This lends support to the argument that presidential vetoes are part of executive–legislative bargaining rather than evidence of political crises, confrontation between the executive and legislature, or the manifestation of hyperpresidentialism.

The fixed-effects variables show that bills introduced by the first president after the restoration of democracy, Patricio Aylwin, were more likely to receive a partial veto than the bills introduced in subsequent presidential terms. But, even including Aylwin, presidents do not regularly veto bills, but the existence of their veto power induces the legislature to bargain with the president.

Figure 1 shows the predicted probabilities of the president issuing a veto given the number of legislative steps. Though the overall chances of a veto are low, the chances of a veto increase substantially when bills go through a fourth legislative step. The 97 partial vetoes issued in the period went through either three legislative steps (three bills) or four legislative steps (94 bills). There were no cases of partial presidential vetoes being issued on bills that swiftly passed both chambers. This challenges the claim that a presidential veto can be used as a shortcut to speed up the passage of a bill. The higher likelihood of a veto being issued on bills with more legislative steps confirms that partial presidential vetoes in Chile can be treated as an additional step in the lawmaking process.

Figure 1. Predicted Probability of the Issuance of a Presidential Veto by the Number of Legislative Steps of a Bill in Chile, 1990–2018

Source: authors, with data from the Library of the National Congress, CEP polls, and the Senate website (www.senado.cl), based on model 1, table 5.

Conclusion

With data from legislative bills in Chile between 1990 and 2018 that reached a point where a presidential veto was an option, we analyzed the determinants of the issuance of partial presidential vetoes. Only 4.2 percent of the 2,346 vetoable bills received a presidential veto. Presidents are more likely to veto bills that go through more steps in the lawmaking process and when they have majorities in both chambers.

Veto power gives the president a stronger bargaining position, but when the president enjoys partial veto power, the executive can use the veto to more effectively influence the content of bills approved by Congress. When presidents have other proactive legislative powers, the veto power might be just another tool in the executive’s toolbox, albeit an extremely powerful one. Still, in the case of Chile, presidents seldomly use that power. In 28 years, only one total veto was issued. Presidents prefer to use partial vetoes, a move that clearly indicates that they want to influence the content of the bill rather than block its enactment. An analysis of the determinants of the 98 vetoes issued in that period underlines a pattern consistent with the claim that partial vetoes can be understood as an additional step in the lawmaking bargaining process between the bicameral legislature and the executive.

Further studies should explore what leads presidents to use specific types of partial vetoes—suppressive, substitutive, and additive—and whether any of those partial vetoes is more likely to be approved or rejected by the legislature. As each bill includes different types of partial vetoes, those studies would need to look at the combination of partial vetoes in each bill. Since we have shown that the dynamics of partial vetoes present different incentives and opportunities in executive–legislative bargaining, it is safe to expect that specific types of partial vetoes will also offer unique incentives and opportunities for the executive and the legislature to bargain over the content of a bill.

The way in which veto power was used in Chile during the period also calls into question claims about exacerbated presidentialism. With one exception, total vetoes were never used in the 28 years between 1990 and 2018. Moreover, the rare use of partial vetoes points to its use as a tool in a cooperative bargaining relation between the executive and the legislature, not as an expression of hyperpresidentialism. Presidents in Latin American democracies have often been defined as having too many proactive legislative powers and attributions. Adjectives like excessive, hyper, and strong are often used to describe presidentialism in Latin American democracies. By analyzing the use of one of the powers more often associated with the alleged excessive powers of presidents, we find that, in Chile, vetoes are seldomly used and, when used, are normally issued in an extended bargaining process between the executive and the legislature. Presidents do have a stronger bargaining position, but presidential vetoes are used not as a zero-sum game in executive–legislative relations but as an additional step in an often extended lawmaking process.