Research on policy decision-making at the subnational level remains limited in Brazilian political science, especially with regard to the Northern states. Which have attracted far less scholarly attention than those in the South, Southeast, and Northeast regions (
Pioneering studies of Executive–Legislative relations in Brazil’s states generally portrayed legislatures as subordinate to the will of governors. In his now-classic study, Fernando
The significance of this thesis for the study of Executive–Legislative relations in Brazil’s states is undeniable, as
Research in this field has revealed important variations in institutional arrangements across Brazilian states, both in terms of legislative output and of the internal organization of state assemblies. Accordingly, scholarship on legislatures and Executive–Legislative relations must broaden its analytical lens and move beyond the longstanding neglect of newer Brazilian states, which have consequently received far less scholarly attention. The limited representation of these states in the literature constrains our understanding of subnational institutional dynamics and of the insights these experiences may offer. Addressing this gap requires greater research efforts, particularly in legislative assemblies where access to data is hampered by weaker levels of institutional development, poor archival digitization, and limited transparency in legislative procedures.
This study focuses on a specific case: Executive–Legislative relations in the state of Acre between 2007 and 2022. The topic has received little sustained attention in the literature, whether in studies centered on the State Legislative Assembly of Acre (ALEAC) or in analyses covering longer periods. This is understandable, largely because of the limited information and incomplete records on bills available on the institution’s website, which often makes on-site archival research unavoidable. Still, examining four consecutive legislatures allowed us to identify recurring patterns and evaluate the persistence of certain dynamics, which are discussed in greater detail below.
Studying Executive–Legislative relations in Acre is particularly valuable insofar as it broadens the empirical scope of research on coalition presidentialism at the subnational level by bringing into focus a historically understudied state. The political dynamics observed in Acre offer novel empirical contributions and open new avenues for future research, particularly for comparative analyses with other state legislatures drawing on an original, publicly accessible database developed in accordance with best practices in transparency and replicability.
Acre has undergone one of the most significant political and social transformations in the recent history of the Legal Amazon, making it a particularly valuable case for evaluating established theoretical models of interbranch relations. The state also provides a useful basis for comparison with other governments in Northern Brazil, helping to shed light on broader patterns of regional variation. In addition, examining its Legislative Assembly through the use of an original database contributes to a broader understanding of how Brazilian state legislatures operate, especially those located in the Legal Amazon.
The empirical effort to analyze an understudied state creates opportunities to engage with and replicate findings from the literature on subnational politics, opens new avenues for research, and offers fresh perspectives for the study of subnational legislative assemblies.
This research note is guided by the following question: How should we characterize relations between the Executive and Legislative branches in the state of Acre? To address this question, we first draw on the typology proposed by
From a methodological standpoint, our study contributes by applying generalized linear mixed models (GLMMs) with random effects in this context. We rely on two complementary analytical strategies. First, we analyze aggregated party-level data and estimate random party effects in order to capture organizational and strategic differences that cannot be directly observed. Second, we analyze individual bills using random effects for bill sponsors, allowing us to control for heterogeneity among legislators in terms of legislative skill, experience, and political networks.
Although widely employed in other areas of the social sciences, this methodological approach remains relatively uncommon in studies of Brazilian subnational legislatures. Accounting for within-group dependence allows the models to produce estimates more accurate of the factors associated with legislative success while also measuring the relative weight of party-level and individual-level characteristics in the legislative process.
The identification of individual characteristics as determinants of legislative success led us to engage with the literature on institutionalization (
Efforts to collect and organize data on ordinary bills (PLs), complementary bills (PLCs), and proposed constitutional amendments (PECs) between 2007 and 2022 unfolded in three stages.
Collection of data from the ALEAC digital archive (https://sapl.al.ac.leg.br/materia/pesquisar-materia). After the initial data collection, several inconsistencies became apparent, largely due to outdated records and missing information on the status of bills, which required an additional stage of research.
Review of the index of enacted ordinary and complementary laws to verify and, when necessary, correct the information available in the ALEAC digital archive, while also supplementing the material collected during the first stage by identifying enacted laws and confirming the bills from which they originated.
On-site examination of the physical archives containing records on bills and legislative proceedings, maintained by the ALEAC Undersecretariat for Legislative Activities, in order to resolve remaining gaps in the data and verify information collected during earlier stages.
The universe of analysis consists of bills introduced in the ALEAC between 2007 and 2022, corresponding to the four most recently completed legislatures. Earlier periods were excluded because they had already been examined in a 2009 master’s thesis on Executive–Legislative relations in Acre between 1999 and 2006 (
To frame the analysis presented below, we begin by outlining our empirical expectations for each of the patterns of Executive–Legislative relations identified by
Under the autonomous legislature model, the assembly is expected to play a leading role in substantive policymaking, while the Executive is expected to have a comparatively limited capacity to introduce and secure approval for bills. A low incidence of symbolic bills would further reinforce the Legislature’s central role in shaping the policy agenda. Accordingly, we expect that:
Under the politicized model, the Executive’s strength rests on party coalitions rather than on one-on-one bargaining with legislators. As a result, parties differ in both the number of bills introduced by their members and in their success rates in securing approval. Symbolic bills tend to attract broad support regardless of ideology or alignment with the government, whereas the approval of substantive legislation depends on a legislator’s proximity to the government and inclusion in the governing coalition. We therefore expect that:
Finally, an ultra-presidential scenario would be characterized by a governor who exercises strong control over the legislative agenda and encounters little resistance in securing approval for government initiatives, regardless of when bills are introduced – wheter during the final year of the legislative term, which coincides with the electoral cycle, or earlier (
We now turn to the data analysis. Between 2007 and 2022, legislators introduced a total of 3,052 bills in the ALEAC. Of these, 2,767 (90.7%) were ordinary bills (PLs), 248 (8.1%) were complementary bills (PLCs), and 37 (1.2%) were proposed constitutional amendments (PECs).
Although legislators introduced more bills and even secured approval for a larger number of them in absolute terms, the Executive achieved considerably higher rates of legislative success. Of the 925 bills introduced by the Executive, 861 (93.1%) were approved. By comparison, Acrean legislators secured approval for 1,330 (65.0%) of the 2,046 bills they introduced. Executive dominance – measured as the percentage of bills introduced by the governor and approved relative to all bills approved by the Assembly – reached 39.3%. The relative ease with which the Executive secured approval for its agenda does not appear consistent with an autonomous legislature scenario (H1a, H1b).
Examining the legislatures individually, we find that legislators introduced a larger number of bills in every legislature, with average success rates consistently above 50%, especially during the 12th Legislature (2007–2010), when more than 80% of legislators’ bills were approved. As noted above, Executive dominance over lawmaking remained relatively stable across all included periods, with the Executive responsible for roughly 40% of all approved legislation (
Notes: *Indicates the percentage of Executive-sponsored bills approved relative to all bills approved by the Assembly.
Legislature
Executive Success Rate
Legislative Success Rate
Total Bills Approved
Executive Dominance*
2007-2010
97.8%
81.8%
551
40.7%
2011-2014
97.0%
69.5%
579
39.7%
2015-2018
90.3%
59.3%
416
40.4%
2019-2022
87.5%
55.9%
645
37.1%
In addition to the success rates calculated for each legislature as a whole,
Notes: * Indicates the percentage of Executive-sponsored bills approved relative to all bills approved by the Assembly.
Year
Executive success rate
Legislative success rate
Total bills aproved
Executive dominance*
2010
100%
89.1%
149
39.6%
2014
96.7%
75.6%
91
31.9%
2018
90.0%
48.4%
24
37.5%
2022
81.2%
67.4%
181
35.9%
Throughout the period considered, the governor consistently achieved higher success rates than legislators. The pattern observed under Cameli’s administration, however, diverges from the expectations of the ultra-presidentialism thesis, which holds that the Executive’s success rate should remain stable over the course of a term in office (H3d). Instead, the Executive was unable to sustain the same level of legislative performance achieved by previous administrations, with approval rates declining substantially in the administration’s final year.
Particularly noteworthy is the fact that, during Binho Marques’s administration (2007–2010), 100% of the bills introduced by the Executive secured approval in the Assembly. This period also saw the highest legislative success rates for state legislators across the entire timeframe, with 89.79% of their bills approved. Explaining these variations may require closer attention to party dynamics and the formation of governing coalitions in more politicized political environments.
PT legislators – the party that controlled the Executive during the first three analyzed legislatures (2007–2018) – recorded a success rate of 84.7% under PT administrations and 70.0% under the PP administration (2019–2022). They also introduced more bills than any other party during this period, accounting for 13.7% of all bills introduced and approved by legislators. PP legislators, meanwhile, achieved an approval rate of 75.8% under PT administrations and 80.9% once their own party came to power, suggesting that affiliation with the governor’s party may be an important determinant of legislative success. This difference is illustrated in
Beyond distinguishing bills by approval outcome, the data also allow us to analyze them by subject matter. Compared with federal deputies and senators, state legislators typically achieve higher approval rates for their bills, although these rates tend to decline when attention is restricted to substantive bills (ABRUCIO, TEIXEIRA and COSTA, 2001;
Notes: Each bill was classified into a single subject area.
Executive
Legislative
Subject area
N
%
N
%
Government and public administration
680
73.5
263
12.9
Economy
102
11.0
28
1.4
Budget
39
4.2
18
0.9
Environment
17
1.8
25
1.2
Education, Culture, and Sports
14
1.5
98
4.8
Labor, Social Security, and Welfare
12
1.3
14
0.7
Human Rights
11
1.2
125
6.1
Honorary and Symbolic
9
1.0
1018
49.8
Pandemic
9
1.0
101
4.9
Health
7
0.8
114
5.6
Urban Affairs and Transportation
6
0.6
35
1.7
Public Security
5
0.5
27
1.3
Agriculture
5
0.5
18
0.9
Consumer Protection
4
0.4
59
2.9
Science, Technology, and Communications
4
0.4
6
0.3
Public Utility Declarations
1
0.1
90
4.4
Other
0
0.0
7
0.3
In addition to legislative output, we also considered the ideological composition of each legislature and its proximity to the Executive’s party ideology (
Legislature
Assembly Ideology
Executive Ideology
Ideological Distance
2007-2010
3.68
3.00
0.68
2011-2014
4.54
2.62
1.92
2015-2018
4.92
2.47
2.45
2019-2022
4.72
6.00
1.28
The Executive’s ideological orientation shifted more sharply with the transition from PT administrations (2007–2018) to the PP government beginning in 2019. Even so, the ideological distance between the governor and the assembly during this period was the second smallest observed across the period analyzed. Moreover, except for the first legislature examined, the assembly’s ideological average remained relatively stable, already reflecting the broader rightward shift in Brazil’s party system. Thus, it was during Binho Marques’s PT administration (2007–2010) that both the Executive and the Legislature recorded their highest approval rates, precisely in the legislature where the ideological distance between the two branches was smallest and at a time when the PT still controlled a large share of seats in the ALEAC.
Up to this point, the findings point to a politicized pattern of Executive–Legislative relations, while also showing that some expectations associated with other models do not hold in this case. The analysis now turns to how party dynamics help account for the legislative outcomes observed.
To better understand how ideological proximity between the branches affects the approval rates of party-sponsored bills, we first estimated two logistic regression models. In this case, the dependent variable is binary, corresponding to whether or not a sponsor’s bills were approved, grouped by party. The models include each party’s ideological distance from the assembly average, its ideological distance from the Executive party, and the size of its legislative caucus (
Notes: * *** p < 0.001; ** p < 0.01; * p < 0.05.
Model 01
Model 02
Coefficient
𝜷
exp(𝜷)
exp(𝜷)
Intercept
0.195
1.22
0.401
1.49
Distance from the Assembly
0.351***
1.42
0.355***
1.43
Distance from the Executive
-0.279***
0.76
-0.279***
0.76
Caucus Size
0.312***
1.37
0.307***
1.36
Polarization
-0.114
0.89
Model 01 indicates that, for each additional point in the ideological distance between the sponsoring party and the assembly’s average ideological position, the probability of approval increases by 42%. In other words, parties located farther from the assembly’s average ideological position tend to achieve higher success rates. At first glance, this finding is somewhat counterintuitive and runs counter to our initial expectations. One possible explanation is that parties positioned farther from the average legislator tend to introduce fewer bills, or only do so when other factors increase the likelihood of approval. Alternatively, this finding may simply indicate that ideology plays a limited role in structuring legislative outcomes in this context. It is important to note that the ideological data refer to parties at the national level, and ideological differences at the state level may therefore, not be fully captured.
In the same model, we find that the closer a party is to the governor, the greater the likelihood that a sponsor from that party will secure approval for a bill: each additional point of ideological distance from the governor reduces the likelihood of approval by 24%. This finding is consistent with expectations, given the higher approval rates previously observed among members of the governor’s party throughout the analyzed period, which further reinforces the importance of alignment with the Executive. Finally, Model 01 indicates that caucus size is an important predictor of legislative success, as expected. Each additional seat held by a party is associated with a 37% increase in the likelihood that a bill sponsored by one of its members will be approved.
Model 02 introduces legislative polarization as an additional explanatory variable. Polarization is measured through the assembly’s internal ideological dispersion and captures the degree of fragmentation within the legislature. To operationalize this measure, we use the weighted standard deviation of the assembly’s ideological average, with weights based on caucus size. The results show little variation between the two models and indicate that polarization has no significant effect.
One important consideration is that caucus size, much like ideology, does not capture the full range of differences between parties. Even parties with similarly sized caucuses and distinct ideological profiles may differ in their programmatic orientations and political interests, effectively making them different kinds of political ‘animals’. Caucus size and ideological distance capture part of what makes a party more or less successful, but party-specific characteristics – such as organizational resources, voting discipline, historical reputation, and networks of interparty cooperation – are not fully captured by these variables. In addition, bills introduced by members of the same party share unobserved characteristics that create within-group dependence, thereby violating the independence assumptions underlying traditional regression models. In other words, the models estimated thus far assume that, once the included variables are controlled for, all bills have the same baseline probability of approval. Once heterogeneity across parties is taken into account, however, a different model specification becomes necessary.
Accordingly, by incorporating the assumption of heterogeneity, we propose a logistic regression model with random effects (GLMM), in which bills introduced by members of the same party are expected to resemble one another more closely than bills sponsored by members of other parties. Rather than including a dummy variable for each party – which would lead to overparameterization given the large number of parties and the small number of legislators within each one – the model presented in
Notes: * *** p < 0.001; ** p < 0.01; * p < 0.05.
Fixed effects
Estimate
p-value
Intercept
1.614***
<0.001
Distance from the Assembly
-0.118
0.345
Distance from the Executive
-0.264***
<0.001
Caucus Size
0.011
0.906
Party
1.182
0.26
To begin with, the variance of the party-level random intercept (1.182) suggests that parties account for a substantial portion of the variation in legislative success within the ALEAC. To further assess the extent to which parties contribute, on average, to legislators’ legislative success, we calculated the Intraclass Correlation Coefficient (ICC) based on the variance of the party-level random intercept:
The ICC value of approximately 0.26 indicates that around 26% of the variance in the probability of legislative success is attributable to differences between parties.
As for the remaining parameters in the model, only distance from the Executive remains statistically significant. In other words, much of what had previously been attributed to caucus size or ideological distance from the rest of the assembly is actually the result of party-level heterogeneity. The only ideological dimension that remains significant after controlling for party effects is proximity to the governor’s party: the closer a party is to the governor, the greater the likelihood that one of its members will secure approval for a bill. In this model, internal polarization and caucus strength proved less important than the party affiliation of the bill sponsor, reinforcing the idea that party culture and organizational resources – captured through the random effects – play a crucial role.
Finally, we arrived at the main model of this study. We estimated the model at the level of individual bills rather than relying on party-level aggregates. The goal was to assess whether, and to what extent, party affiliation and the individual characteristics of bill sponsors influenced the legislative success of each bill. To this end, we introduced the following modifications to the model: 01. we added a binary variable indicating membership in the ALEAC leadership board during the corresponding legislature in order to control for the sponsor’s greater institutional capacity; 02. we removed the indicators of party ideological distance – given the absence of individual-level ideology measures and to avoid collinearity between party affiliation and ideology – but included a variable indicating membership in the governor’s party; 03. we included individual legislators as random effects, capturing unobserved differences among legislators (e.g., legislative skill, experience, prestige, and political networks) that may systematically affect their approval rates; and 04. we included electoral coalitions as a proxy for governing coalitions, assuming that legislators belonging to the governing coalition are more likely to secure approval for their bills. We opted to use electoral coalitions because data on governing coalitions themselves were unavailable.
A few observations about these choices are warranted. First, one would normally expect a regression model to include individual-level control variables for legislators, such as gender, education, professional background, political experience, and similar characteristics. Although these factors are not directly related to the mechanisms explored in our analysis, they could still affect the results and would therefore typically be included as controls. Our decision not to include individual-level variables in the main model, however, stems from the multilevel design itself. Logistic regression with random effects allows us to account for unobserved heterogeneity at the legislator level through the random intercept for bill authorship, which absorbs systematic individual differences that cannot be directly observed. Because the purpose of the article is to assess the relevance of political, partisan, and institutional factors – rather than explain individual variation in legislative performance – these characteristics were treated as part of the unstructured individual-level variance. Even so, we present two additional models below including individual-level controls as a robustness check.
Notes:
Model 04
Model 05
Fixed effects
Estimate
p-value
Estimate
p-value
Intercept
0.868**
0.002
-0.992***
<0.001
Caucus Size
-0.114
0.393
-0.003
0.979
Member of the Governing Coalition
0.312
0.141
0.590*
0.012
Member of the Governor’s Party
1.257**
0.006
1.263*
0.012
Member of the Leadership Board
-0.026
0.878
0.660**
0.006
Bill Sponsor
0.503
0.126
0.497
0.131
Party
0.211
0.053
≈0.000
0.000
Residual (π2/3)
—
0.822
—
0.869
In Model 04, which includes all bills, one notable finding is that caucus size appears to have a negative effect on the likelihood of a bill being approved, contrary to our initial expectations. However, although the estimated coefficient is negative, the effect is uncertain and statistically insignificant. The only statistically significant effect is membership in the governor’s party: the odds of approval are more than three times higher for members of the governor’s party (odds ratio = 3.51). We also observe a decline in the importance of party affiliation relative to the previous aggregated model. In this specification, party affiliation accounts for only 5.4% of the variation in the probability of bill approval, whereas 12.8% is explained by heterogeneity among bill sponsors. Overall, among the variables considered here, membership in the governor’s party stands out as the strongest predictor of legislative success.
Everything changes, however, once we exclude honorific and symbolic bills in Model 05. Although the coefficient for caucus size remains negative, the effect itself is substantively negligible and statistically insignificant. Membership in the governing coalition and in the governor’s party, by contrast, both have positive and statistically significant effects on the likelihood of approval. Members of the governing coalition were 80% more likely to secure approval for their bills, consistent with hypothesis H2c. At the same time, the independent effect of party affiliation disappears. Thus, the relevance of parties in the approval of substantive legislation ultimately comes down to whether legislators belong to the governor’s party or governing coalition. Once again, this suggests that the main distinction between parties may lie in their alignment with the Executive party. Future research dedicated to identifying governing coalitions over time – and across different stages of the legislative term – may help provide a more accurate account of legislative outcomes in the state.
In addition, membership on the ALEAC leadership board exerts a positive and statistically significant effect, indicating that legislators who hold these positions are more likely to secure approval for their bills because of the institutional advantages associated with the office (odds ratio = 1.93). This is an important finding and the strongest effect in the model. The institutional prerogatives of the leadership board largely explain this advantage in the legislative process. Members of the leadership board are responsible for organizing legislative business and overseeing the institution’s proceedings, including setting the voting agenda. This privileged position gives these legislators greater control over the progress of their own bills, making it easier to place them on the agenda and prioritize debate at strategically favorable moments.
This finding is also consistent with the literature on subnational legislatures, which has emphasized the importance of internal power structures in determining legislative outcomes. More specifically, the results underscore the importance of examining the professionalization of state legislatures and the career trajectories typically associated with them – a dimension absent from the typology proposed by
In both models, approximately 13% of the variation is explained at the legislator level, indicating substantial heterogeneity among bill sponsors. Both models also demonstrate that symbolic and substantive bills differ significantly in the factors that affect their likelihood of approval, a distinction that should be considered when studying subnational legislatures with high volumes of symbolic bills.
Taken together, the results indicate that parties do matter in the legislative dynamics of the state of Acre, even if their overall impact is limited. As for the type of Executive–Legislative relationship observed, the evidence suggests that relations within the ALEAC more closely resemble the politicized model, in which legislative outcomes are structured by party dynamics rather than by legislative autonomy or a rubber-stamp arrangement.
We found evidence contrary to both hypotheses associated with legislative autonomy (H1a and H1b). The Executive achieved higher approval rates for its bills while contributing relatively little to symbolic legislation, an area largely dominated by the Legislature.
Among the hypotheses associated with ultra-presidentialism, only H3b receives partial support, insofar as the Executive dominates the approval of substantive legislation. Although the Executive remains the central actor in the legislative process, legislators’ participation was far from negligible. With respect to H3a, legislators introduced slightly more substantive than symbolic bills, even if the difference between them remained modest. We also identified meaningful differences across parties, particularly among those belonging to the governing coalition, pointing to the importance of party negotiations – contrary to the expectations of the ultra-presidentialism thesis, according to which negotiations occur primarily at the individual level (H3c). Finally, we found variation in the Executive’s approval rate over the course of one gubernatorial term, a result that does not support hypothesis H3d.
Thus, although the Executive is clearly strong and faces little difficulty in securing approval for its agenda, our findings suggest that this process depends on negotiation with parties in the Assembly rather than on a rubber-stamp arrangement.
Overall, the evidence lends greater support to the hypotheses associated with the politicized model. We found that, when symbolic bills are included, membership in the governing coalition has little predictive power – a pattern that changes substantially once the analysis is restricted to substantive legislation (H2b and H2c). With regard to H2a, the main differences between parties appear to revolve around their level of support for the government. And, whether they belong to the governor’s party.
Beyond the article’s main questions, once individual-level heterogeneity among bill sponsors is taken into account, variables related to the Assembly’s degree of institutionalization and professionalization emerge as potentially important explanatory factors, pointing to promising avenues for future research.
Lastly, we present two additional models that incorporate individual-level controls for legislators as a robustness check for the results presented thus far (
Notes: * *** p < 0.001; ** p < 0.01; * p < 0.05
Model 06
Model 07
Fixed effects
Estimate
p-value
Estimate
p-value
Intercept
1.197*
0.048
-1.917**
0.008
Caucus Size
-0.218
0.139
-0.030
0.862
Member of the Governing Coalition
0.285
0.210
0.578*
0.016
Member of the Governor’s Party
1.622**
0.001
1.344*
0.019
Member of the Leadership Board
0.122
0.503
0.703**
0.004
Sponsor Gender (*male)
-0.141
0.698
0.069
0.852
Secondary Education
0.156
0.778
1.170
0.051
Higher Education
-0.155
0.736
0.960
0.079
Age
-0.055***
< 0.001
-0.006
0.670
Bill Sponsor
0.616
0.149
0.374
0.102
Party
0.216
0.052
0.022
0.006
Residual (π2/3)
—
0.798
—
0.892
Educational attainment was recoded into three broader categories (primary, secondary, and higher education) in order to avoid sparse categories and ensure the statistical stability of the model while preserving substantively meaningful distinctions in legislators’ educational backgrounds.
We obtained the data from the electoral records of the Superior Electoral Court. Missing information for substitute legislators serving in office was manually completed using official sources for the corresponding legislative year, without overwriting the original observations.
Overall, the results of the main models remain largely unchanged. In the model including all bills (Model 06), legislator age emerges as a statistically significant factor, suggesting that younger legislators are more likely to secure approval for their bills – an effect that disappears once the analysis is restricted to non-symbolic legislation.