Abstract
Discipline and cohesiveness of political parties are essential for legislatures to engage in policy-making. Parties in Brazil have historically been considered ideologically weak and uninvolved in policy issues of national importance. Analyses of roll-call votes, however, have shown that parties can be disciplined government supporters. This paper tests the claim that Brazilian parties have also become programmatic actors in their own right. The paper uses statutory delegation content to test whether voting discipline translates into greater influence on the substance of legislation. The data analysis shows that party unity among parties of the government coalition does not affect statutory content. Opposition parties, by contrast, are more likely to reduce the executive's discretion when they are more unified. Overall, the support for the hypothesis of programmatic parties is weak, given that executive authorship is the strongest determinant of statutory content.
Introduction
One of the recurring questions in the research on presidential democracies in Latin America has been whether legislatures and legislative parties are up to the task of governing. The perceived weakness of legislatures is often contrasted with the extensive powers of the president and the dominance of the executive's agenda in the government's policy output. Political parties are thought to be partly to blame. In the first decade since the return to democracy, observers characterized Brazilian parties by “their fragility, their ephemeral character, [and] their weak roots in society” (Mainwaring 1995: 354). Congressional representatives eschewed national policy in favor of local demands, and “drove hard bargains” when it came time to vote on presidential policy initiatives (Ames 2001: 206). This assessment of legislators and political parties offered a bleak picture of the system's capacity to represent voters and be held accountable.
Scholars of Brazilian politics continue to investigate the behavior of political parties more than twenty years after the transition to democracy. Some scholars have claimed that Brazil's legislative parties have transcended the parochialism that historically characterized its legislative politics in response to the evolution of voters’ preferences (Lyne 2008) and the crystallization of ideological cleavages with regard to the macroeconomic reform agenda of the 1990s (Hagopian, Gervasoni, and Moraes 2009). The work based on the survey of federal legislators conducted by Timothy Power, César Zucco Jr., et al., on the other hand, has not shown strong ideological foundations on which programmatic behavior could be built (Power and Zucco Jr. 2009; Zucco Jr. and Lauderdale 2011). Research on the location of parties along an ideological spectrum has revealed that voters have a tenuous understanding of the meaning of ideological labels, with the implication that political elites have a limited ability to appeal to voters in ideologically consistent ways (Ames and Smith 2010), with the Worker's Party (Partido dos Trabalhadores – PT) being the exception (Lucas and Samuels 2010).
This article joins the debate on the nature of Brazil's political parties by testing the claim of increased programmatic behavior among Brazilian legislative parties through the lens of statutory content. The content of legislation – and more specifically, how much discretion legislation extends to the executive and how much direct policy instruction it contains – provides a measure of the relationship between the branches and the ability of legislative parties to act cohesively and deliberately in the policy-making process. Theories of delegation predict that as conflict increases between government and opposition, the willingness of the legislature to delegate policy decisions to the executive declines. Instead of delegating, the legislature passes statutes with high policy content, leaving few decisions to the discretion of the executive (Huber and Shipan 2002; Epstein and O'Halloran 1999). This behavior, however, requires that parties be capable of agreeing on policy details.
This research highlights the importance of party unity not only to whether and how legislation gets passed, but also to its content. Agreeing on policy requires reaching consensus and accommodating competing preferences. If, as Ames claims, the electoral system in Brazil yields an obstructionist and chaotic legislature over which the party leadership has imperfect and uneven control (Ames 2001: 275), we would not expect the legislature to be able to agree on policy details. Under these conditions, it would still be possible for levels of voting discipline to be high as a result of the party leaders’ ability to coordinate legislative action in favor of the executive's agenda. If, on the other hand, voting records also reflect programmatic unity, we would expect statutory delegation to vary with inter-branch conflict: as conflict increases, the opposition closes ranks and produces more detailed legislation that reduces executive discretion.
Statutory delegation provides a complementary measure of inter-branch relations and party behavior to floor voting. Roll-call votes tell us to what degree legislative parties “act in harmony” at the time of voting (Morgenstern 2004: 4), but explain little about their ability to effectively influence the substance of policy. The existence of party discipline – meaning, the ability of the leadership to rally the vote – has been documented by many scholars of Brazilian politics; we know less about the effect of party characteristics on the substance of legislative work. Using delegation content in addition to roll-call votes to analyze the relationship between the branches allows us also to assess the extent to which the executive relents in terms of policy control. Ames raises the point that the voting record masks the concessions made by the executive in the content of policy (Ames 2002: 189–190). If legislation emerges watered down by legislators’ demands, then obtaining approval from congress is less of a presidential victory than the voting record would lead us to believe. Statutory delegation provides a window into the division of labor and the policy bargains reached between the branches independently of the vote. Ultimately, this article seeks to understand whether the reported increase in party discipline and programmatic behavior contribute not just to making the executive's job easier but also to a stronger legislature, namely, one more intently engaged in the task of policy-making on behalf of those it represents.
The data analyzed in this paper does not support the claim that programmatic parties are increasingly involved in policy-making in Brazil. In particular within the government coalition, voting discipline does not translate into a collective contribution to the substance of policy. The evidence of policy cohesion among parties in the opposition is somewhat stronger, which is consistent with the view that the parties in the government coalition pursue office resources while those in the opposition must build a policy platform to attract votes. Furthermore, this research confirms that the executive is the central actor in defining policy, with the legislature assuming an ancillary role.
The article proceeds as follows: The next two sections explain the logic of statutory delegation and review the literature on legislative parties and executive dominance in Brazil. These sections explain how the content of statutes can serve as proxies for politicans’ interests in policy outcomes and their efforts to control them (Huber and Shipan 2002: 45), and the importance of party cohesiveness for the logic of delegation to apply to systems with strong executives and weak parties, such as that of Brazil. Sections 4 and 5 present and test, respectively, the hypotheses connecting party unity and inter-branch conflict to statutory delegation. The concluding section discusses the results and the implications of executive authorship for our understanding of legislative party behavior.
The Choice to Delegate
Despite the apprehension caused by the notable asymmetry between the authority of the executive and legislative branches in Brazil, researchers agree that the legislature has made a choice to delegate authority to the executive (Shugart and Carey 1992; Reich 2002; Pereira and Rennó 2003; Figueiredo and Limongi 1997; Amorim Neto and Tafner 2002). The proof is that in several instances some of this authority has been rolled back and congress has reasserted itself relative to the executive. The most cited example is the change in the rules passed by congress in 2001 regarding the use of presidential decrees barring the executive from continually renewing decrees instead of seeking the approval of congress. Congress has also taken action to restructure its internal organization – for example, by reorganizing the process of amending the budget proposal and streamlining requests for discretionary spending. In other words, the congress is not helpless and has on several occasions taken collective action to change the rules by which it operates.
While the use of decrees and the participation of congress in the process of governing by decree have received a lot of attention, the content of legislation and the authority delegated therein has scarcely been studied. The study of statutory delegation is revealing in two ways: First, it gives a sense of the division of labor between the branches and the trade-offs implied in crafting detailed legislation versus leaving the details of policy undecided. (The remainder of the next section elaborates on this further.) Second, it reflects the partisan dynamics in the legislature. The following section elaborates on partisan dynamics and their potential effect on the content of legislation.
Trade-offs of Delegation
When faced with writing and passing legislation, legislators have two options: to write vague statutes that give the executive broad authority over policy content, or to write detailed statutes that contain policy details and procedural instructions that limit the bureaucracy's scope of action (Huber and Shipan 2002). By introducing specific instructions, legislators seek to remove control over policy decisions from the hands of the executive bureaucracy and to secure specific outcomes. By introducing policy details, for example, legislators are able to target benefits to their constituents. If left unspecified, the agency in charge of implementation may choose to spend the resources elsewhere. Legislators can also seek to control policy outcomes with procedural instructions. Legislators may create a new agency to implement policy and give it greater autonomy, or require various agencies to share jurisdiction over a program to increase the opportunities for access, or to diffuse the influence of a particular agency.
Legislating directly, however, can be costly. Specifying every policy preference in the text of legislation would entail that the legislature build policy expertise, and debate and agree on details, which can incur high opportunity costs for legislators. Policy-making keeps legislators tied up in Brasilia and competes with fundraising activities and constituency service. Delegating, on the other hand, allows legislators to benefit from bureaucratic expertise and has the additional advantage of passing off responsibility to the executive for difficult and sometimes unpopular decisions (Epstein and O'Halloran 1999: 23). Given the relatively low levels of incumbency and the frequent rotation of legislators across committees (Pereira and Mueller 2000: 57), expertise levels in the Brazilian Congress tend to be low and legislators are prone to rely on the executive to fill in the details of policy. The downside of relying on the executive, naturally, is the loss of control over policy outcomes. Legislators have some instruments of oversight and often seek to personally influence the implementation of policy by lobbying the bureaucracy, but oversight and lobbying are costly and imperfect avenues of policy-making. The examples that follow examine two statutes in detail to illustrate how (and when) politicians use statutory language in an effort to control policy outcomes.
The IT Law of 2001
Law 10176 of 2001 provided tax discounts to manufacturers of IT goods. The program had its origins in the market reserve for computers created in the 1970s, which was in place until 1991 when it was replaced by a policy that provided tax incentives to IT companies in order to promote domestic production and to help local firms adjust to more competitive market conditions. In 1999, when the program was set to expire, the computer industry had grown considerably, due to a great extent to the arrival of major foreign manufacturers (Tigre and Botelho 2001; Nassif 2002), who joined state governments in lobbying the federal government for the renewal of the law and the continuation of the tax incentives.
The more salient issues to both the executive and legislators were the duration of the incentives and the definition of the goods eligible for the program. The executive's project, led by the Ministry of Science and Technology, was to renew the program but reduce its impact on tax revenues by phasing out the tax discounts over time and ending the program in 2013, in line with the preferences of the Ministry of Finance to minimize the fiscal impact. Legislators from São Paulo and other southern states held similar preferences, and were particularly interested in ending the program in 2013, when the tax benefits enjoyed by the Manaus Free Trade Zone (ZFM), its main competitor in the production of IT goods, were also set to expire (Gazeta Mercantil 1999b, 1999a). The state of Amazonas, where the ZFM is located, opposed the renewal of the benefits. Since Brazil embarked on trade liberalization in the early 1990s the ZFM had been losing its comparative advantage relative to producers in the rest of the country (Lyra 1995). The details of the IT law were crucial for the relative advantages of producers in the ZFM and other parts of the country. Legislators from the North, Northeast and Central-West regions – who form a regional block in congress -weighed in as well, seeking to introduce language that would assure them a share of benefits (Jornal do Brasil 1999; O Estado de São Paulo 1999a).
The final text of the law settled some of these issues directly, through policy and procedural details, and delegated others. Out of a total of 115 paragraphs, the law contains 7 delegating paragraphs and 74 paragraphs with policy and procedural details that constrain executive discretion. The basic definition of the categories of goods that would be eligible for this program was delegated to the executive (Art. 1). The law, however, also included a list of goods that were not eligible (Art. 16A, §1), and gave the executive the authority to include manufacturers of cell phones and monitors on a case-by-case basis. The Cardoso administration, in power at the time, fought to keep this decision to itself, since excluding these products outright could cause a migration of manufacturers to the ZFM and an even greater loss in tax revenue for the federal government (O Estado de São Paulo 1999b). The states located in the less-developed regions of the country also managed to obtain higher tax benefits (Art. 11).
The inclusion of these policy details, as well as a specific schedule of benefits over the duration of the program, protected legislators from bureaucratic drift and allowed benefits that are sectorial in nature to become geographically targeted in ways that would have been difficult to achieve if these decisions had been left to the bureaucracy. The law also included procedural requirements to be followed by the bureaucracy in an effort to prevent the latter's decisions from straying too far from politicians’ preferences. The final list of eligible goods, for example, was to be drawn jointly by the Ministries of Finance, Science and Technology, and Development and Commerce (Art. 4 §1). By including multiple agencies, no single agency could impose its policy preferences, and legislators had multiple avenues to influence decisions during the implementation stage. The final text of the law differed in important ways from the project submitted by the executive, mostly through the addition of policy details that settled the distribution of benefits among competing states. Interestingly, the preference of the executive was not necessarily to maximize discretion. From the point of view of the Ministry of Finance, for example, it was preferable to limit the scope of the program in its inception, rather than to seek to control the choices made by the implementing agencies during implementation.
Another example is Law 10420 of 2002, which created a harvest insurance program for small-farmer victims of natural floods (Fundo e Benefício Seguro-Safra). The program was first implemented via temporary decree in November of 2001 (MPV 111/01), and then submitted for congressional review. The original decree, written by the executive, specified the areas to be covered by the insurance, the maximum amount that could be paid out to a family, conditions for participation, and a cap on the maximum number of beneficiaries. Further details were delegated to the Ministry of Agrarian Development, though the scope of decision-making was rather limited. Approval from congress to transform the decree into ordinary legislation required broadening the geographic area eligible for the program and eliminating the limit on the number of beneficiaries. While funding for discretionary programs such as this one is controlled by the executive via impoundment measures anyway, legislators often oppose setting limits in the text of the law and prefer to maximize potential access during implementation. Other amendments proposed in congress – for example, to expand the program to the entire country – did not muster the necessary support.
Party Discipline and Delegation
Delegation can be understood as a trade-off between minimizing the costs and uncertainties of policy-making and maximizing control over policy (Epstein and O'Halloran 1999). If we are to apply this interpretation of statutory delegation to the case of Brazil, one prior condition must be met: sufficient party discipline and cohesion. Writing detailed statutes and agreeing on policy details to limit the bureaucracy's scope of action and obtain desired policy outcomes cannot be achieved if parties are unable to act collectively and coherently in the process of policy-making.
The research carried out by Huber and Shipan on delegation in parliamentary systems offers the following prediction regarding delegation in the absence of a single-party majority: majority coalition and minority governments are less likely to delegate policy authority to the executive (2002: 208– 209). The reasoning behind this prediction is that at least one of the parties in the legislative majority holds different preferences from the party or parties in government and is not likely to trust the executive with policy discretion. For example, if the Conservative and Liberal Parties formed a coalition government in Britain, and the Ministry for Pensions belonged to the Liberal Party, the members of the Conservative Party in parliament would be inclined to delegate less and use statutory detail to limit the discretion of the liberal minister over pension policy.
The connection that Huber and Shipan establish between coalition governments and delegation is based on an important assumption the authors made regarding party behavior: parties are “reasonably homogenous and disciplined” (185). In the example above, the fact that the Conservative Party is cohesive enables it to agree internally on its policy position and to deny the Liberal Party discretion over the content of legislation. The incongruence between the Conservative Party in parliament and the liberal minister in the cabinet leads to the choice to delegate fewer policy decisions, while the programmatic cohesion within the Conservative Party enables it to choose statutory policy-making over delegation. While the incongruence between policymakers and policy implementers (inter-branch conflict) is what determines the choice to delegate, cohesiveness is necessary for statutory policy-making to be a cost-effective alternative to delegation.
In Brazil, concerns over excessive delegation and executive dominance are closely related to party weakness. Three versions of executive-legislative politics have been put forth in the literature. In the first version, parties are “loose patronage machines” (Hagopian, Gervasoni, and Moraes 2009: 361) composed of legislators that routinely switch parties, are indifferent to national policy issues, and focus mainly on the interests of their local districts (Mainwaring 1991; Mainwaring and Scully 1995). Never enjoying a majority in the legislature, presidents must expend precious resources to form ad hoc coalitions to pass a national policy agenda (Ames 2002, 2001). In the second version, party discipline is at the president's disposal, delivered by a party leadership in congress in exchange for pork and patronage. Party discipline, however, is not synonymous with programmatic cohesion, but rather the result of a strategic game between party leaders, backbenchers, and the executive (Figueiredo and Limongi 1999; Figueiredo and Limongi 2000; Pereira and Mueller 2004). In these two versions, discipline exists, though in varying measures, as a function of executive resources and cunning.
The third and most recent interpretation speaks of a transformation in partisan behavior in the last twenty years in the electoral and legislative arenas. Hagopian, Gervasoni, and Moraes (2009) report declines in electoral volatility as well as increases in unity in party voting. According to the authors, legislators across the party spectrum are giving up parochial campaigns and accepting leadership directives in congress. In this new scenario of legislative conduct, discipline is not driven by a strategic calculus to obtain benefits controlled by the leadership, but by the increasing value of party labels and programs to legislators’ careers (2009: 361–363). This transformation, they argue, was motivated by the economic reforms of the 1990s that polarized the party system and generated a distinctive programmatic cleavage with respect to state/market policy questions. The financial reforms of the 1990s also reduced the availability of discretionary resources that used to fuel the vote-by-vote construction of legislative majorities, and increased legislators’ reliance on the party label to win elections.
Lyne (2008) also reports programmatic behavior among legislative parties in Brazil in the contemporary democratic period. Her theory is one of voter-driven change. According to Lyne, voters in developing countries fall into collective-action traps where the equilibrium strategy is to vote for politicians who make clientelistic appeals, even when the choice is clearly suboptimal. Lyne argues that since 1989 Brazil has escaped the clientelistic trap, and voter demand for collective goods policy-making has been met by centralized parties that discourage excessive “personalist politicking” (Lyne 2008: 47). According to Lyne, legislative politics in the last two decades have taken the form of well-defined government and opposition coalitions with distinct policy positions. Since the Collor administration, and with few exceptions, government coalitions have been smaller and have projected a clear ideological identity. Membership in either a government or an opposition coalition has become a source of information regarding a party's policy position, enabling voters to use this information to hold politicians accountable and demand better policy outcomes (Lyne 2008: 134–136).
Research on the ideological foundations of Brazilian parties, however, provides only weak support for the claims that parties in Brazil have transcended parochialism and adopted behaviors that provide clear, policy-centered cues to voters. Based on survey responses regarding the ideological placement of parties provided by legislators from 1990 to 2005, Power and Zucco Jr. conclude that while parties in Brazil “can be clearly arrayed on a classic left-right scale and the overall ordering has been relatively stable across time” (2009: 219), ideological differentiation among parties is declining (2009: 228). Using the same survey data, Zucco Jr. and Lauderdale also report a decline in the influence of ideology in legislative voting over time, and the blurring of ideology and the pursuit of political resources among the drivers of legislative behavior, in particular during the Cardoso years (Zucco Jr. and Lauderdale 2011: 389). Lucas and Samuels also find that the Brazilian party system possesses low levels of ideological coherence (Lucas and Samuels 2010: 44).
While this research casts doubt on the pursuit of policy as the driver of legislative party behavior in Brazil, Zucco Jr. and Lauderdale also conclude that the cleavage between government and opposition is a consistent and growing dimension in legislative voting (Zucco Jr. and Lauderdale 2011: 389). It is thus possible that, despite both weak ideological beliefs within parties and differentiation across parties, there is a programmatic basis for the behavior of the government and opposition coalitions. This is also supported by Lyne's findings that the opposition parties vote consistently and in a unified way against the government's agenda (Lyne 2008: 129), indicating that parties can behave cohesively independently of access to government resources. The idea of programmatic cohesion along a government– opposition cleavage has also been put forth by Figueiredo and Limongi in the form of a single government agenda crafted jointly by the executive and its coalition (2009).
This paper contributes to the debate regarding the nature of party behavior in Brazil by putting the claim of programmatic cohesion to the test. If, as some of the authors claim, party discipline and programmatic cohesion exist independently of executive prerogatives, then legislative parties should be able to produce both high and low delegation outcomes, depending on the degree of cohesiveness within coalitions and the amount of conflict between the branches, as predicted by theories developed for democracies with consolidated parties. If party discipline exists only as a function of executive leadership, as previously thought, then delegation will always be high, either because the president has assembled a strong coalition that can deliver, or as a consequence of congress’ inability to agree on policy details.
Party Unity and the Content of Legislation: Hypotheses
The main finding from the studies of inter-branch conflict and delegation in presidential and parliamentary democracies in the U.S. and Europe is that “as conflict over policy increases, delegation to the executive becomes less attractive” (Epstein and O'Halloran 1999: 77). For Brazil, divisiveness is measured by an index of likeness (Lyne 2008) based on the degree of agreement between the government and opposition coalitions on roll-call votes (see explanatory notes in Appendix B). Other independent variables include unity scores calculated from roll-call votes, used to test the relationship between voting unity and statutory content, and the size of the government and opposition coalitions, since it is reasonable to expect larger coalitions to have greater influence over the legislative process. As described in more detail in the following section, two separate and complementary measures of delegation were used: delegation language and policy detail. Delegation language involves direct grants of authority while policy detail entails policy instructions introduced directly into the text of legislation. The former increases discretion while the latter reduces it.
The first set of hypotheses to be tested concerns the conditions under which the legislature is inclined to extend policy discretion to the executive via delegation language:
The second set of hypotheses concerns the introduction of policy details as a means of limiting executive discretion. The ability of the opposition to deny the executive policy discretion and instead set policy via legislative means depends not only on the size of the opposition but on the opposition's ability to reach agreements on policy details. We could then expect the following to be true:
Testing the Hypotheses
The hypotheses set out in the previous section were tested using a random sample of ordinary legislation drawn from the period from 1991 to 2007. The total production of ordinary legislation for this period was 3,218 laws, of which 1,456 were statutes mandating changes to the budget. These were excluded as they are not subject to legislative amendment and do not have a substantive policy content. Purely symbolic legislation, such as laws naming airports or declaring special holidays, was also excluded. The final population from which the sample was drawn contained 1,655 statutes. A random sample of 178 laws was drawn maintaining proportionality to the length of the 16 cabinets that governed during this period.
Variables
Dependent Variables: Measures of Discretion
Two types of language were coded in the sample legislation: The first is delegating language granting authority to the executive. Delegating language is defined as language “that give[s] another governmental body the authority to move policy away from the status quo” (Epstein and O'Halloran 1999: 274). This includes the creation of new programs; definition of decision-making criteria; explicit grants of decision-making authority; and creation of new agencies or committees with the authority to regulate an activity of economic or public interest (Epstein and O'Halloran 1999: 4–5). An example drawn from the cases presented earlier is to give the president the authority to include a certain type of economic activity as the beneficiary of a tax incentive program (Law 10176/01). The variable DELRAT is the ratio of paragraphs with delegating language to the total number of paragraphs in the statute.
The second type of language comprises policy details that give explicit policy instructions to the bureaucracy. This type of language reduces executive discretion by limiting the decisions left open for the bureaucracy to make during implementation. Policy details are distinct from general, descriptive, and non-constraining language (Huber and Shipan 2002: 2, 48). For example, the same law mentioned above lists specific goods, with their corresponding Harmonized Commodity number to identify them, as not being eligible for benefits. As a result, the agencies involved in approving applications for the program do not have the discretion to award benefits to the producers of these goods. Policy details include instructions on how an agency should allocate expenditures; limits on overhead spending and the diversion of funds to other projects; and the identification of groups or states as being eligible for special treatment. Policy details also include conditions to be met by presidential appointees to an agency in charge of implementation, and instructions on how decisions are to be made. For example, several agencies or ministries may be required to consult each other and make joint decisions, or a decision may require final vetting by the legislature or the president. The variable POLRAT is the ratio of paragraphs with policy details to the total number of paragraphs in the statute. See Appendix A for descriptive statistics on delegation and policy detail ratios.
Independent Variables: Inter- and Intra-coalition Behavior
The independent variables used in testing the hypotheses are: government unity, opposition unity, coalition size, and index of likeness. The period from 1991 through 2007 includes the presidencies of Fernando Collor de Mello, Itamar Franco, Fernando Henrique Cardoso, and Luis Inácio “Lula” da Silva, and comprises 16 different cabinets and opposition coalitions (Figueiredo 2007; Lyne 2008). Monthly averages of government unity, opposition unity, likeness, and coalition size were calculated from 1,741 roll-call votes using Lyne's cabinet periodization and coalition composition. Appendix B provides summary statistics and explanatory notes for the independent variables. Appendix C shows a correlation matrix of all the independent variables.
Results
Separate models were run for the two dependent variables: delegation language and policy detail. The first set of hypotheses to be tested concerns the relationship between delegation language and coalition characteristics/behavior. Following existing research, we expect that the government coalition prefers to increase delegation content in order to expand the discretion allotted to the executive, and that the opposition coalition, in turn, prefers to reduce the executive's discretion. All other things being equal, discipline should enable the government and the opposition to realize their respective preferences, and conflict should reduce delegation.
The dependent variable, DELRAT, which is the proportion of delegation paragraphs in a statute, has a very right-skewed distribution, as a large proportion of the statutes do not have explicit grants of authority (see Figure 1).
Distribution of Zero-Delegation Statutes
Law 8494/92 regarding the elimination of the salary index for the adjustment of residential rental agreements is an example of a zero-delegation statute. This statute contains detailed instructions on how residential rents should be adjusted, with a correspondingly high ratio of policy details (DELRAT = 0; POLRAT = 0.43), and does not grant any entity authority to make discretionary decisions on the matter. Another example is Law 11350/2006, which regulates the activities of health agents in the health system. This statute contains no explicit delegation language and only few policy details relative to the extensive descriptions of agents’ qualifications and responsibilities (DELRAT = 0; POLRAT = 0.07).
Because of the skew in the distribution of the delegation variable, we opted to create a delegation dummy as the dependent variable (DELDUM) to fit a logistic model. The objective is to predict whether intra- and intercoalition characteristics (size, unity and divisiveness) predict the occurrence, rather than the quantity, of delegation language in the text of legislation. The results appear in Table 1. None of the independent variables are significant in Model 1. The size of the coalitions, the unity of the government coalition, and the divisiveness in voting between government and opposition coalitions do not appear to affect the decision to extend discretionary authority to the executive. These results are contrary to existing findings on inter-branch conflict and delegation. In the case of the U.S., Epstein and O'Halloran find that as the seat share of the opposition increases, the amount of discretionary authority delegated to the president declines (1999: 135). Huber and Shipan also find that as conflict increases in coalition and minority governments, the discretion awarded to the government declines (2002: 200–201).
Logistic Regression of Coalition Characteristics on Delegation (odds ratios)
Note: SE in parentheses;
denotes significance at the 10-percent level
Since Government Size and Opposition Size have a correlation coefficient r=-0.72, separate models using only one of these variables were run and did not produce substantively different results.
Source: Author's own calculation.
To further explore possible explanations for the delegation data, the sample was split according to authorship, as it is reasonable to expect government and opposition coalitions to behave differently depending on the source of the policy initiative and the opportunities it offers to shape the content of statutes. Three types of bills that result in ordinary legislation were included in the sample: legislative bills, executive bills and executive decrees (Medidas Provisórias, MPV). Legislative and executive bills follow the same procedural path through the legislature. All bills enter the Chamber of Deputies (lower house) via a steering committee (Mesa Diretora da Câmara dos Deputados) that refers bills to the standing and special committees. Once in committee, bills can be amended by committee members; if amendments are accepted, the leader (relator) of a bill in committee produces an amended bill (sustitutivo). Following committee review, bills are sent to the floor for a vote. Bills can also die in committee or be discharged via urgency procedures and submitted to an immediate vote on the floor. Most executive bills are introduced by the president, although other agencies of the executive branch, such as the Supreme Court (Supremo Tribunal Federal) and the Public Ministry (Ministério Público), also propose legislation.
Executive decrees follow a different legislative path. As established by the Constitution of Brazil, executive decrees are reviewed by a joint committee of the House and Senate under an open amendment rule allowing all legislators to propose changes. In actuality, such committees are rarely formed, but a rapporteur is assigned to collect amendments and manage the process of amending, rejecting or converting the decree with no changes into ordinary legislation (Reich 2002). Of the executive decrees in the sample, 48 percent were converted into law without changes.
If we look at the distribution of zero-delegation statutes according to authorship, we notice that zero delegation is much more likely in statutes that originate in the legislature: 86 percent of the bills authored by the legislature have no delegation language, compared to 53 percent of statutes originating in the executive. This is consistent with the idea that the executive wishes to have more discretion over policy, rather than less, and is more likely to introduce direct grants of discretionary authority in the bills it authors. Table 2 shows the relationship between the type of bill and the presence of delegating language. The association between authorship and delegation is significant at the 1-percent level and is clearly driven by the contrast in the distribution of statutes with zero delegation between legislative bills and executive decrees.
Occurrence of Delegation Language (by bill type)
Notes: First row shows frequency and row percentages in parentheses; second row shows cell contribution to χ2.
Source: Author's own calculation.
Given the importance of authorship, we repeated the regression on the delegation dummy (Table 1) for a sample of statutes that excludes executive decrees – that is, it only includes statutes that follow a regular legislative policy-making process (Model 2), since it is possible that party and coalition dynamics do not have an opportunity to play out in the process of amending executive decrees. In Model 2, the variable “likeness” is significant at the 10-percent level, and of the expected sign, but the model has no overall significance. Similarly, in Model 3, which only includes bills of legislative origin, opposition unity is weakly significant, and of the expected sign (more unity, less delegation), but the model overall lacks significance. The tests do not provide support for the first set of hypotheses and show that the size of the government and opposition coalitions, the voting discipline of the coalitions, and the level of disagreement between them are not important determinants of delegation language.
The second outcome to be tested is the use of policy details to legislate directly and reduce executive authority over policy. We would expect the opposition to be interested in specifying policy details during the legislative process to reduce executive discretion during the implementation stage. We would also expect unity within the opposition coalition to enable this to happen and be positively related to the frequency of policy detail in the text of legislation. Opposition size and conflict should be positively related to policy detail.
Table 3 shows the results of multivariate regressions using POLRAT as the dependent variable. Results show that characteristics of the opposition are significantly related to the frequency of policy details. Models 1 and 3, both for the entire sample and for legislative bills only, show that opposition size and unity are significant. The coefficient for opposition size is negative, indicating that as the coalition opposition grows, the frequency of policy details shrinks. This result contradicts Hypothesis 2a, which states that a larger opposition will be more effective in the legislative process, but is consistent with the basic insight that larger groups have greater difficulty acting collectively. The coefficient for opposition unity is positive, supporting the hypothesis (2b) that parties in the opposition are not only able to vote together, but also cohesive enough to agree on policy content. The coefficients increase slightly in size in Model 3, indicating a greater interest among legislators in influencing the content of bills authored within their ranks. The overall explanatory power of Model 3 (Adj. R2 = 0.17) is also substantially higher.
Regression Estimates of Coalition Unity and Inter-branch Conflict on Policy Details
Note: p-values in parentheses;
denotes significance at the 5-percent level.
Source: Author's own calculation and compilation.
The results presented in Table 3 lend further support to the idea that the origin of bills matters to their content and to the behavior of legislative coalitions. A final look at the average levels of delegation and policy detail by author and type of bill also reveals not only that authorship is an important determinant of legislative content but that executive authorship determines more content overall, with both types of language present with greater frequency in executive bills and decrees (see Table 4).
Average Delegation and Policy Detail (by author and bill type)
Note: Difference of means test with unequal variances;
denotes significance at the 1-percent level.
Source: Author's own compilation.
Discussion and Conclusions
This article tests claims of increased programmatic behavior in the Brazilian legislature through the lens of statutory content. The measures of statutory content are delegation language, extending authority to the executive to make policy decisions, and policy details, which specify policy decisions and decision-making processes, effectively reducing the policy discretion available to the executive. Based on research carried out in the U.S. and parliamentary democracies in Europe, we would expect a large and cohesive government coalition to be better able to secure policy discretion on behalf of the government, and large and cohesive opposition coalitions to seek legislation with fewer grants of authority and more direct policy instructions. We would also expect less delegation as conflict between government and opposition increases, and more delegation when legislative and executive branches are more closely aligned.
The results can be summarized as follows: Executive authorship is a dominant factor and constitutes the backdrop for the interpretation of the results for the unity and conflict variables. Bills authored by the executive (executive bills and decrees) have more content than bills authored by the legislature – that is, they are much more detailed in terms of deliberate grants of discretion and policy specifics. While only one-third of the statutes have delegation language, the majority (87 percent) were authored by the executive. Executive decrees in particular have the highest frequency of delegating language (57 percent). This is consistent with the idea that the executive prefers to grant itself discretion to make decisions, but it also indicates that the executive relies heavily on its legislative powers, rather than its coalition in congress, to influence the content of legislation. It is interesting also that the executive uses its legislative initiatives to introduce specific policy content, possibly in an effort to limit the discretion of the bureaucracy as decisions move away from the presidency after legislation is approved.
Perhaps most surprising is the finding that the characteristics of the government coalition, its size and unity, have no incidence on the content of legislation. The Brazilian political system of the contemporary period, in particular beginning with President Cardoso, has been characterized in the literature as being driven by a close partnership between the president and the party leadership heading the government coalition in congress. This party leadership, seated in the powerful College of Leaders, is responsible for mustering the vote and passing the executive's agenda by appointing rapporteurs who steer bills through committees and by discharging bills from committee before legislators can modify their content. This latter practice, in particular, has been touted as a key component of the executive's strategy in congress (Pereira and Mueller 2000: 48). In the sample used in this article, 60 percent of legislative and executive bills were subjected to discharge petitions, and two-thirds of these originated in the executive. 1 The high frequency of executive bills among bills subjected to discharging is typical in the Brazilian Congress and confirms that the executive uses the government coalition to shepherd bills of its own creation through the legislative process but not to influence its content. On the contrary, the purpose of discharging a bill is to prevent the committee members, whom the party leaders have appointed, from acting on a piece of legislation. From this we can conclude that unity in voting is a good measure of discipline, in the sense of the leadership's ability to mobilize the vote and control legislative procedure, but provides little information regarding policy cohesiveness, since legislators in the government coalition have little opportunity (or inclination) to participate in the policy process.
Executive decrees not included, since they do not follow the regular legislative process during which discharge petitions can be used. Additionally, information on the legislative process was not available for six legislative bills and six executive bills, bringing the total number of bills included in the urgency count to 108.
It is possible, however, that the parties that compose the government coalition have a programmatic influence on the content of legislation during the bill-drafting phase in the cabinet. In recent research, Figueiredo and Limongi (2009) have suggested that politics in Brazil should be understood in terms of a government–opposition cleavage in which the government espouses a majority agenda that fuses the preferences of the president and the members of the coalition (79). If this is the case, and the government coalition in Brazil functions like a parliamentary coalition government where the cabinet collectively crafts policy, then the apparent lack of influence on the part of the coalition parties in congress on the content of legislation is actually an acknowledgment that these parties already made their contribution via their representatives in the cabinet. In the absence of evidence that the executive functions as a collective policy-making body in Brazil, it is not possible to interpret the results of our analysis as anything other than an absence of programmatic behavior.
The opposition coalition, on the other hand, has an observable effect on the content of legislation: the opposition is able not only to organize itself for voting purposes, but also to introduce policy details into the text of legislation as internal unity increases. This is consistent with Zucco Jr. and Lauderdale's findings that “government opposition rather than ideology is now the dominant dimension of conflict” (2011: 391) as well as earlier research on legislative production in Brazil that finds that legislators who belong to the opposition are more likely to engage in behaviors that create a policy platform from which they can campaign, in contrast with legislators who belong to the government coalition, who prefer to leave legislating to the executive and concentrate on constituency service (Amorim Neto and Santos 2001).
Finally, it is worth noting the absence of significance of the conflict variable (likeness), since inter-branch conflict is considered a key driver of delegation in the literature. In Brazil, the cleavage that exists between government and opposition coalitions in the voting records does not translate into visible changes in the content of legislation. This is consistent with the behaviors of the coalitions described above: the opposition, removed from the spoils system controlled by the executive, focuses on policy content independently of a government coalition focused on procedural effectiveness.
This research does not support the idea of programmatic parties taking charge of national policy-making in Brazil, nor does it reveal a chaotic legislature unable to act coherently. The discipline observed in the voting records is present within the government coalition that enables the executive to move its agenda through the legislature, and within the opposition. Programmatic cohesion, however, is evident only in the behavior of the opposition. Overall, the research presented here suggests that the programmatic element is weak in terms of the work legislators do within the congress relative to the influence of the executive. Legislative policy-making continues to be strongly influenced by the executive's control of the legislative agenda and the substance of legislation. This research reinforces the image of an executive-dominated political system in Brazil: party discipline exists, but operates in the shadow of the executive.
Footnotes
Descriptive Statistics: Dependent Variables
| Cabinet | Length of statutes (in no. of paragraphs) | Freq. | Discretion | |||
|---|---|---|---|---|---|---|
| Average | Maximum | Minimum | DELRAT | POLRAT | ||
| Collor II | 24 | 141 | 7 | 18 | 0.07 | 0.16 |
| Collor III | 11 | 11 | 11 | 1 | 0 | 0.27 |
| Collor IV | 33 | 59 | 14 | 4 | 0 | 0.44 |
| Franco I | 20 | 23 | 14 | 3 | 0.10 | 0.24 |
| Franco II | 32 | 62 | 10 | 3 | 0.12 | 0.10 |
| Franco III-IV | 37 | 76 | 10 | 7 | 0.04 | 0.32 |
| Franco V | 69 | 374 | 8 | 11 | 0.06 | 0.24 |
| FHC I | 67 | 610 | 8 | 15 | 0.06 | 0.33 |
| FHC II | 48 | 331 | 7 | 27 | 0.06 | 0.24 |
| FHC III | 52 | 547 | 8 | 28 | 0.02 | 0.30 |
| FHC IV | 33 | 87 | 8 | 9 | 0.04 | 0.26 |
| Lula I | 28 | 148 | 7 | 11 | 0.03 | 0.18 |
| Lula II-V | 73 | 844 | 6 | 24 | 0.03 | 0.18 |
| Lula VI | 37 | 128 | 8 | 9 | 0.04 | 0.14 |
| Lula VII | 11 | 11 | 11 | 1 | 0 | 0.18 |
| Lula VIII | 33 | 88 | 10 | 8 | 0.02 | 0.05 |
| Avg. Std. Dev. | 38 19 | N=17 8 | 0.04 0.09 | 0.23 0.20 | ||
Note: Cabinet periodization from Lyne (2008).
Source: Author's own compilation.
Descriptive Statistics: Independent Variables
| Cabinet | Coalition unity
a
|
Coalition size
b
|
Index of likeness c | Roll-call votes d | ||
|---|---|---|---|---|---|---|
| Gov | Opp | Gov | Opp | |||
| Collor II | 0.78 | 0.81 | 173 | 215 | 0.68 | 69 |
| Collor III | 0.85 | 0.84 | 139 | 210 | 0.61 | 18 |
| Collor IV | 0.88 | 0.86 | 230 | 85 | 0.59 | 13 |
| Franco I | 0.90 | 1.00 | 314 | 36 | 0.44 | 3 |
| Franco II | 0.67 | 0.54 | 355 | 42 | 0.80 | 40 |
| Franco III-IV | 0.64 | 0.89 | 274 | 82 | 0.50 | 22 |
| Franco V | 0.74 | 0.85 | 243 | 85 | 0.57 | 8 |
| FHC I | 0.75 | 0.89 | 296 | 79 | 0.45 | 142 |
| FHC II | 0.78 | 0.95 | 372 | 69 | 0.32 | 392 |
| FHC III | 0.81 | 0.97 | 359 | 79 | 0.52 | 358 |
| FHC IV | 0.81 | 0.97 | 235 | 74 | 0.39 | 31 |
| Lula I | 0.92 | 0.61 | 249 | 124 | 0.64 | 147 |
| Lula II-V | 0.83 | 0.76 | 293 | 112 | 0.57 | 253 |
| Lula VI | 0.78 | 0.92 | 261 | 122 | 0.74 | 37 |
| Lula VII | 0.94 | 0.81 | 289 | 123 | 0.61 | 55 |
| Lula VIII | 0.86 | 0.79 | 357 | 115 | 0.55 | 153 |
| Avg. Std. Dev. | 0.78 0.14 | 0.85 0.15 | 295 64 | 101 43 | 0.52 0.22 | N=1,741 |
Note:
Coalition Unity was measured by the weighted Rice Index for each vote. For each vote, the Party Rice Index (|percent Aye - percent Nay|) was weighed by the party's participation in the respective coalition. The Rice Index for government and opposition coalitions for each vote was then weighed using Carey's measure of closeness: 1 - 2*(|0.5 - percent Aye|) (Carey 2000: fn. 11). This prevents consensual votes from inflating the unity index. All votes that reached quorum were included in the calculations.
Coalition size was measured using election results as reported by Nicolau (n.y.) adjusted to take into account party switches. Data on party switches for the 49 Legislature (1991–1995) was obtained from the House Journal (Diário da Câmara dos Deputados); party switches during the 50th Legislature (1995-1999) and for the years 1999 to 2000 were obtained from the Library (see Secretaria-Geral da Mesa); and party switches during the period from 2000 to 2007 were calculated directly from the roll-call records. The compositions of the coalitions for each cabinet were taken from Lyne (2008).
We use Mona Lyne's index of likeness to measure “the degree to which members of two groups vote the same way on a bill.” The higher the values, the less difference between the voting records of government and opposition coalitions (Lyne 2008: 105).
Roll-call records from which unity scores were calculated were obtained from the official website of the Chamber of Deputies (Câmara dos Deputados, www.camara.gov.br/Internet/plenario/votacao.asp;).
Source: Author's own compilation.
Correlation Matrix of Independent Variables (N=178)
| Government Rice | Opposition Rice | Government Size | Opposition Size | Likeness | |
|---|---|---|---|---|---|
|
|
1.0 | ||||
|
|
-0.0558 | 1.0 | |||
|
|
0.0096 | 0.2733 | 1.0 | ||
|
|
0.1644 | -0.2696 | -0.7209 | 1.0 | |
|
|
-0.1843 | -0.3442 | -0.2672 | 0.2919 | 1.0 |
Source: Author's own calculation.