Sie sind auf Seite 1von 24

The Dynamics of Lawmaking

in a Bicameral Legislature

Comparative Political Studies


Volume 41 Number 12
December 2008 1583-1606
2008 Sage Publications
10.1177/0010414007308536
http://cps.sagepub.com
hosted at
http://online.sagepub.com

The Case of Brazil


Taeko Hiroi
The University of Texas at El Paso

This article analyzes legislative performance in a nascent presidential bicameral democracy, taking Brazil as a case. The author argues that the timing and
outcomes of legislative production are functions of bicameral incongruence,
types of bicameralism, sequence of examination, and legislative bargaining.
These hypotheses are tested using a new legislative data set from Brazil that
covers over 3,000 bills submitted to the National Congress since 1988. Event
history analyses of these bills show that presidential bicameral (coalitional)
majorities, presidential elections, initiation by the lower house, and bills
proposing provisional changes raise the chances of a bills approval. The
results also indicate that the effects of many of these variables are time
dependent. In contrast, bicameral incongruence, symmetric bicameralism,
and legislative elections either raise the risks of a bills rejection or delay the
timing of its approval. Economic crises increase legislative activities in general
in both approving and rejecting bills.
Keywords:

bicameralism; legislative politics; gridlock; Congress; Brazil

his article analyzes the dynamics of lawmaking in a nascent presidential bicameral democracy, taking Brazil as a case. It makes at least two
contributions to comparative legislative politics. First, much of the influential quantitative legislative research both in American and comparative legislative studies has focused on either legislative behavior, usually analyzing
patterns of roll-call voting and party discipline (e.g., Ames, 2001; Desposato,
2006; Morgenstern, 2002; Poole & Rosenthal, 1997), or the propensity for
Authors Note: I would like to thank Barry Ames, Christopher Carman, Gaspare M. Genna,
Mark Hallerberg, William Keech, Anibal Perez-Linan, Lucio Renno, and two anonymous
reviewers for helpful comments. An earlier version of this article was presented at the annual
meetings of the American Political Science Association and Midwest Political Science
Association. This research was funded by the National Science Foundation Dissertation
Improvement Grant (Grant No. 0315126) and the Andrew Mellon Predoctoral Fellowship, the
University of Pittsburgh.
1583

1584

Comparative Political Studies

gridlock at the aggregate system level (e.g., Binder, 1999, 2003; Cheibub,
Przeworski, & Saiegh, 2004; Krehbiel, 1998; Mayhew, 1991). Studies of lawmaking per se are surprisingly rare in legislative research except when individual pieces of legislation are analyzed as case studies (Binder, 2003, ch. 5).1
To address this shortfall, I compiled a legislative data set that includes over
3,000 bills submitted to the Brazilian Congress since the promulgation of the
countrys 1988 Constitution. Using this data set, I take individual bills as a
unit of analysis and examine how various legislative rules, legislative actors
preferences, and environments that shift over time affect the timing and outcomes of lawmaking activities in the Brazilian Congress.
The second contribution of this research pertains to bicameralism. Many
scholars have argued that bicameralism is a key institution that increases
legislative stability or gridlock (e.g., Binder, 1999, 2003; Bottom, Eavey,
Miller, & Victor, 2000; Tsebelis & Money, 1997). Substantial corroborating
evidence has been accumulated in recent years for the U.S. presidential and
European parliamentary bicameral systems.2 However, there is virtually no
research to date on the effects of bicameralism in recently democratized
countries. Prior studies of legislative politics in new democracies have
instead concentrated on the executive and presidentiallower chamber relations or modeled bicameral legislatures as if there were only one chamber
(e.g., Ames, 2001; Cox & Morgenstern, 2002; Shugart & Carey, 1992).
However, modeling bicameralism explicitly is particularly important for
the study of new democracies. First, many newly democratized presidential
countries adopted bicameral legislatures. Legislative gridlock that bicameralism presumably causes may be detrimental to unconsolidated democracies, because it may contribute to democratic instability and breakdown,
especially in presidential systems (Linz, 1994). Furthermore, many new
democracies, where the status quo is not as favorable as that in the
advanced industrial countries, face multitudes of economic and social problems that require immediate and dire changes. If bicameralism indeed has
the propensity for increasing legislative deadlock as is often argued, it is
crucial to understand when and under what institutional configurations
bicameralism is likely to generate legislative gridlock.
Second, theories of bicameralism have been developed in advanced
industrial countries with consolidated democracies. Although their propositions have found strong support in the empirical studies of OECD countries,
we want to know whether they can travel to countries with very different
political and economic environmentsfor example, where democracy has
not yet been deeply rooted in society, political parties are inchoate and
amorphous, and economic stability is rare but much desired.

Hiroi / Lawmaking in a Bicameral Legislature

1585

This article examines legislative performance in a new presidential


bicameral democracy: Brazil. I argue that the timing and outcomes of legislative production are functions of bicameral incongruence, types of
bicameralism, sequence of examination, and legislative bargaining. Event
history analyses of Brazilian legislative data (1988-2004) show different
dynamics for bill approval and rejection. Specifically, presidential bicameral (coalitional) majorities, presidential elections, initiation by the lower
house, and bills proposing provisional changes raise the chances of a bills
approval. The results also indicate that the effects of many of these variables are time dependent. In contrast, bicameral incongruence, symmetric
bicameralism, and legislative elections either raise the risks of a bills rejection or delay the timing of its approval. Evidence also shows that acute economic crisesbut not mere economic problemsstimulate policy-making
activities both in approving and rejecting economic bills, a finding that is
especially relevant for crisis-prone developing countries.
This article proceeds as follows. The next section lays out theoretical
expectations derived from the existing literature on lawmaking in a bicameral legislature and legislative bargaining. Section 3 outlines the research
design and methodology. Section 4 discusses independent and dependent
variables and Brazilian legislative data and the fifth section presents their
estimation results. The final section concludes.

Lawmaking in a Bicameral Legislature


This section generates theoretical predictions about the timing and outcomes of legislative activities. Much of the contemporary debate on legislative politics focuses on the conditions that constrain a political systems
ability to act promptly and decisively (e.g., Weaver & Rockman, 1993).
Past research has shown that a concurrence of the preferences of key legislative actors (both collectives and individuals) is crucial for a change of a
prevailing policy (Tsebelis, 2002). Where this condition is absent, legislative gridlock is expected. Prior research has also revealed the importance of
bargaining in the timing of legislative enactment (Cox & Kernell, 1991,
p. 243). How, then, do bicameralism and bargaining affect the speed and
outcomes of legislation?
Bicameral incongruence and gridlock. Social choice scholars have
argued that bicameralism is an institution that imposes a more stringent
condition for a change of a status quo policy than unicameralism (Hammond

1586

Comparative Political Studies

& Miller, 1987; Riker, 1992; Tsebelis & Money, 1997). The gridlockinducing quality of bicameralism stems from the requirement that concurrent majorities of two distinct chambers be needed to implement a policy
change.3 In the social choice literature, therefore, whether bicameralism
induces policy immobility depends critically on the locations of the two
chambers ideal points. In general, the larger the distance between the ideal
points of the two chambers, the greater the size of the bicameral corethe
set of all points that cannot be defeated with a given decision-making rule
(Hammond & Miller, 1987). As a result, gridlock is more likely as the
bicameral incongruence of preferences grows:
Hypothesis 1: Bicameral incongruence increases the likelihood of gridlock.

Generally, the two houses of a bicameral legislature are incongruent if they


have different partisan compositions and member characteristics.
Legislative bargaining. Even when there is a consensus among legislative actors that some change is preferable to the status quo, they may disagree on the details of the new legislation. This leads to bargaining among
relevant actors. Prolonged bargaining causes significant delays in the passage of new legislation even when the prevailing policy is in no ones interest. In fact, delay is a prominent property of legislative bargaining (Cox &
Kernell, 1991, p. 243).
One can approach bicameral bargaining as a timing game where the lower
and upper houses decide on the timing of concession given its and the other
houses degrees of impatience to reach an agreement.4 Consider the following scenario. The lower and upper houses both loathe the status quo but
they want different changes. The lower house prefers a change to x while
the upper house desires a change to y. The passage of legislation requires
that both houses come to an agreement. However, agreeing on a point closer
to the lower houses ideal point, x, incurs greater costs to the upper house,
and agreeing on a point closer to the upper houses ideal point, y, makes the
lower house assume greater costs. Therefore, each house bargains hard to
minimize the costs of legislation to itself, the costs with which it will have
to live once the bill is passed. In bargaining between the two chambers, the
greater the distance between their ideal points, the larger is the degree of
concession required to pass legislation (Tsebelis & Money, 1997, p. 102).
One of the variables that affect the length of bargaining and its outcomes
is the time discount factor (denoted ) of the actors. The higher the , the
lower is the cost of waiting another round of bargaining to obtain a better

Hiroi / Lawmaking in a Bicameral Legislature

1587

deal. Thus, reflects the relative bargaining strengths of the actors involved.
In the bargaining literature, actors that value an immediate agreement more
than an agreement in a later period are considered impatient. In a game with
complete information, an agreement is struck immediately and the passage
of new legislation is instant (Rubinstein, 1982). Legislative delays occur
due to uncertainties about actors impatience levels (Alesina & Drazen,
1991; Tsebelis & Money, 1997). Generally, the more impatient a player is,
the sooner will a concession be made and therefore the faster the speed of
legislation.
Hypothesis 2: Impatience reduces the propensity for legislative delay.

Another factor that may affect the length of prelegislation bargaining is


the expected duration of new legislation. Remember that the difference
between x and y represents the degree of concession required to make to
pass legislation. Concessions are the costs of an agreement, and because
both houses must live with the consequences of new legislationincluding
those concessions madethat affect the anticipated values of their future
payoffs, ceteris paribus, an uneven distribution of compromises gives
strong incentives to bargain harder, resulting in more time before an agreement is concluded.5
However, if a new policys expected duration is shortfor instance, if it
has a preset expiration date, then an expected postlegislation utility loss
from compromise (i.e., the costs) would be smaller than in a case where
new legislation is expected to last infinitely. In other words, the difficulty
of bargaining does not only depend on x-y; it also depends on some sort of
coefficient representing the expected lifespan of the new legislation, say t
(as in t[x-y]), that affects future payoffs from new legislation. As t increases,
the expected costs of compromise increase, inducing the actors to bargain
harder. Conversely, as t decreases, the total expected utility loss decreases
and hence it is less difficult to reach an agreement. Therefore,
Hypothesis 3: Legislative delay is positively related to the duration of legislation proposed.

Symmetric and asymmetric bicameralism. Bargaining in a bicameral


legislature occurs within a set of rules and procedures binding bicameral
conflict resolution. Conferring on one chamber (typically the lower house)
the power to be decisive in the case of bicameral disagreements is a conflict resolution mechanism adopted in many bicameral systems. Another

1588

Comparative Political Studies

exemplary procedure for bicameral conflict resolution is navette in which


bills shuttle between the two chambers until they reach an agreement or
the bills are aborted (Tsebelis & Money, 1997). There is also a conference
committee in which select members of the two houses draft a compromise
bill to be voted in each house under a closed rule (i.e., without amendments). Conference committees are often adopted in the United States and
France.
Lijphart (1999) calls two chambers with equal constitutional prerogatives
and democratic legitimacy (i.e., whether members are appointed or selected
through popular elections) symmetric and ones that lack these qualifications
as asymmetric. The navette procedure without a stopping rule is an instance
of symmetric bicameralism, and rules that give one of the chambers prevalence over the other is asymmetric bicameralism. If, in symmetric bicameralism, the composition of the two chambers differs with respect to member
characteristics and their preferences (i.e., if the two chambers are incongruent), policy immobility is a likely result. It is also likely to take longer to
strike a deal under symmetric bicameralism than under asymmetric bicameralism. By contrast, an asymmetric bicameral procedure facilitates legislation
even if the two houses are incongruent because the more powerful chamber
can overshadow the less powerful one. In short,
Hypothesis 4: Legislative delay and gridlock are more likely in symmetric
bicameralism than in asymmetric bicameralism.

Bicameral sequence. Another factor that may affect the speed and outcome of legislation is bicameral sequence. Rogers (1998) argues that
sequential moves are strategic choices that bicameral legislatures face.
Because information acquisition is costly, in an environment of high uncertainty, the chamber that has higher expected payoffs specializes, given that
the net expected payoffs (gains from specialization minus costs of specialization) are positive. Rogers maintains that the costs of information acquisition are negatively associated with chambers sizes such that the cost
function is usually smaller for the lower house than the upper house
(assuming that the lower house has more members than the upper house).
Hence, the probability that the lower house introduces legislation that will
be adopted is higher than the probability that the upper house introduces
such legislation. Furthermore, because the upper house desires to take
advantage of the lower houses informational expertise, bills initiated by the
lower house should enjoy relatively swifter approval.

Hiroi / Lawmaking in a Bicameral Legislature

1589

Hypothesis 5: Bills initiated by the lower (more populous) house are more
likely to have speedy approval than bills initiated by the upper (less
populous) house.

Note that this argument challenges a conventional wisdom of bicameralism that stresses senatorial expertise. According to this view, a senate (or
an upper house) promotes superior legislation relative to a lower chamber
because of longer terms of office, higher minimum ages, and finer career
trajectories that make senators professional legislators capable of studying
legislation relatively objectively (Tsebelis & Money, 1997, p. 40). Therefore,
this perspective implies that senate-initiated bills should entertain higher
approval rates than those proposed by the lower house. However, Rogerss
informational model predicts the contrary. These two opposing arguments
can be tested empirically.

Research Design and Method


I test these hypotheses with legislative data from Brazil for the period
1988-2004. Brazil is an ideal case to examine the effects of bicameralism
and various decision rules on legislative production. First, Brazil has
unique legislative rules that in effect make it possible to study various
types of bicameralism while holding country-specific factors constant.
Constitutionally, the Brazilian Chamber of Deputies (the lower house) and
the Senate (upper house) are coequal. Unlike many countries in which the
lower house dominates the upper house, there is no area of legislation that
is granted to the Chamber of Deputies but denied to the Senate.6 However,
in the making of statutory regimes, the house where a bill is first introduced
prevails in the final decision. That is, although the reviewing house has the
right to amend, so long as it approves the bill, there is no formal mechanism
for the reviewing house to enforce such amendments. In this case, the
Brazilian bicameral system can be considered asymmetric. What makes
Brazilian bicameralism particularly unique is the fact that, by initiating bills
first, both the Chamber and Senate can be the house with the power to make
a final decision.
On the other hand, proposals for constitutional amendments represent
legislative activities in the context of symmetric bicameralism. The Brazilian
Constitution requires that each house of Congress approve an identical text
of an amendment by three fifth majorities, voted on two separate rounds
(Article 60), and the bill shuttles between the two houses until both houses

1590

Comparative Political Studies

Table 1
Rules and Procedures in the Brazilian Congress
Mode of
Deliberation
Constitutional
amendment
Complementary
law
Ordinary law

Navette until
agreement is reached
Navette (initiating
house is decisive)
Navette (initiating
house is decisive)

Type of Vote
Roll call
Roll call
Symbolic

Number of
Votes Required
3/5 of each chamber
on 2 rounds
Absolute majority
in each chamber
Simple majority
in each chamber

Type of
Bicameralism
Symmetric
Asymmetric
Asymmetric

Note: This table excludes budgetary procedures and deliberation of presidential decrees and vetoes.
A simple majority refers to votes by a majority of members present in the session, and an absolute
majority refers to votes by a majority of each chambers membership.
Source: The Brazilian Constitution (Senado Federal, 2002).

adopt exactly the same text, or else the bill is aborted. Therefore, neither house
is dominant in the making of constitutional amendments. Table 1 summarizes decision rules and procedures in the Brazilian Congress.
Second, the question of legislative efficiency has spurred widespread
discussions both among political scientists and policy makers in Brazil. To
many observers and practitioners of Brazilian politics, Brazil suffers governability problems. Governability refers to the efficiency of a nations
executive and legislative branches in the making of programs and policies
(Ames, 2001, p. 1). Scholars such as Mainwaring (1999), Ames (2001), and
Stepan (2000) argue that the Brazilian political system has been unable to
carry out major socioeconomic and political reforms despite their dire
needs. Where changes were made, moreover, they often came too late (usually punctuated by some sort of crises) and/or too little. These scholars contend that the lower chambers open-list electoral system and strong
subnational forces fragment the polity and hinder governability in Brazil. In
contrast, Figueiredo and Limongi (2001) argue that Brazilian presidents
have been able to marshal coalitional majorities in Congress and able to
govern effectively using their enormous agenda-setting powers in an otherwise fragmented Congress. My expectation is in the middle of the two
strands of arguments, akin to the argument by Amorim Neto, Cox, and
McCubbins (2003). I expect that Brazils governance patterns are more
complex than those depicted by either position but are influenced by each
presidents legislative strategy and other political and socioeconomic landscapes that prevail in particular periods.

Hiroi / Lawmaking in a Bicameral Legislature

1591

One of the key political configurations that are likely to influence governability is bicameralism. It is surprising that none of the major studies on
Brazilian legislative politics has examined bicameralism as a potential
source of legislative gridlock. Ames (2001), Figueiredo and Limongi
(2001), and Amorim Neto et al. (2003) all focus on executivelower chamber relations. This scarcity of scholarly attention to Brazilian bicameralism
may stem from the widespread perception that the Senate is simply the
house of review and tends to be pro-executive, making it appear to be a nonsignificant actor in the legislative process (Figueiredo & Limongi, 1996,
p. 8).7 Although it is true that many bills originate from the lower house (due
to the constitutional requirement that executive proposals be submitted to
the Chamber of Deputies) and the initiating house has the last word on the
bills, the Senate still retains veto power on such proposals. Moreover, the
Senate holds the same last word prerogative as the Chamber in relation
to the bills originating from that house. With respect to constitutional
amendments, there is no advantage for being the initiating house. Hence,
any analysis of legislative politics in Brazil should explicitly treat the powers and preferences of the Senate as variables to be studied rather than
assume its nonimpact. Understanding the relationships between the two
houses will shed new light into the study of legislative politics in Brazil and
contribute to solving the governability puzzle.
In the past, many scholars estimated legislative delays and gridlock using
cross-tabulations, linear regressions, and/or logit or probit models (e.g.,
Binder, 1999, 2003; Figueiredo & Limongi, 1996; Krehbiel, 1998; Mayhew,
1991). I use an event history analysis to examine legislative dynamics in a
bicameral congress. Event history analysis is concerned with the time until
an occurrence of an event. The hazard rate, h(t), is the instantaneous rate of
an events occurrence (i.e., the probability that an event occurs at a particular point in time), given that the case has survived until t:
Pr(t t + t T t)
h(t) = lim ________________________
t0
t

where T is a nonnegative random variable denoting the time to an events


occurrence. The following analyses use a semiparametric Cox model,
because it estimates the effects of the covariates on the hazard rate without
specifying the distribution of the baseline hazard function or duration
dependency (Box-Steffensmeier & Jones, 2004).
The modes of bill termination in Congresswhether congressional
deliberation ended in the approval or rejection of a billand not just the

1592

Comparative Political Studies

durations of deliberation, are of great interest for students of legislative politics. When multiple modes of events are of substantive interest, competing risks models can be used. The novel element of competing risks
models is risk-specific hazard rates hk(t|k, Xk) where k = 1, . . . , r possible
events that a case is at risk of experiencing. A competing risks model can
be specified using the latent survivor time approach.8
The goal of the quantitative analysis is to estimate risk-specific hazard
rates of bill deliberation in the Brazilian Congress by identifying a set of
covariates that influence them. There are multiple modes of terminating bill
examination. A bill may be approved or rejected. Yet many bills actually do
not reach the final stage of deliberation but are discarded in the process. For
example, some bills are withdrawn by their sponsors; others are declared
impaired due to the approval or rejection of related bills; and yet others are
simply dispatched to the archive at the beginning of a new Congress.9 The
subsequent analyses will focus on the causes and timing of bill approval
and rejection. Clearly, why certain bills are tabled and not acted on is an
interesting and important question. As the discussion of legislative bargaining pointed out, inaction is often a result of bargaining deadlock, and any
study of legislative production should not exclude tabled bills. The competing risks event history model can readily incorporate them because the
information on all forms of bill termination as well as pending bills is used
to estimate the hazards of approval and rejection. It explains both why an
event occurs and does not occur at specific times.

Dependent and Independent Variables


The dependent variable is the timing and outcome of bill examination in
the Brazilian Congress. Central controversies on Brazils governability lie in
not only whether a particular proposal passes (i.e., an outcome) but also how
soon the measure passes (i.e., timing). As Ames (2001) aptly points out, many
bills that are proposed to Congress never reach a vote, and long delays are
common for many others that are eventually voted up or down. Some important and controversial bills stay in Congress for many years, even decades.10
Furthermore, Congresss inability to move quickly often waters down the
effects of the policy change even if a proposal ultimately passes.11 Therefore,
we need to look at the time to a decision as well as the decision itself to accurately capture the essence of the governability debate.12
Timing is indeed of crucial concern to many policy makers and political
scientists. Many care not only whether a bill will be passed but also how

Hiroi / Lawmaking in a Bicameral Legislature

1593

fast it will be passed. The aggregation method based on the quantity of bills
passed then misses an important piece of information in an analysis.
Furthermore, counting approved bills per Congress may not be an appropriate way to capture the dynamics of bill approval and gridlock in legislatures like Brazils whose archival rules are more complex than their U.S.
counterpart. Unlike the U.S. Congress, where all bills not voted within a 2year legislative period are terminated, in Brazil bills that do not reach final
voting within a 4-year legislative period may be terminated or carried over
to the next legislative period.
The legislative data set I constructed for this analysis includes proposals
for constitutional amendment and two types of statutory billsordinary and
complementary.13 It includes all executive and judicial proposals submitted
to the Brazilian Congress from October 1988 (the promulgation of the current constitution) through December 2003. With respect to congressional
proposals, the data set includes all bills submitted during the same period
and subsequently approved at least by the house of origin. Specifically, all
the congressional bills considered here were approved by the Chamber of
Deputies if proposed by a deputy or approved by the Senate if proposed by
a senator. In other words, I consider only those bills that cleared the hurdle
of approval at least in the legislative body of origin, be it the Executive,
Judiciary, Senate, or Chamber of Deputies. This method allows me to analyze only those bills that are regarded as important enough to pursue at least
by a house or branch of origin given its legislative agenda. At the same time,
I can eliminate the bills that were proposed for the sake of proposing (which
many members of Congress do). In total, the data set contains 3,066 bills
that meet those criteria.
Using the Senates and Chambers online legislative databases (Cmara
dos Deputados, 2003-2004; Senado Federal, 2003-2004) and information
obtained through the Senates Subsecretaria de Informaes, I recorded the
history of each of these bills. The data include information on (a) the date
in which the bill was introduced in Congress, (b) its sponsor, (c) the type of
the bill (i.e., ordinary bill, complementary bill, or constitutional amendment
proposal), (d) the sequence of its examination, (e) the time (in days) it spent
in Congress, and (f) its final outcome (approved, rejected, impaired, or
pending). The histories of these bills were traced from their introduction to
Congress to their termination, or in the case of pending bills until July 31,
2004, on which date they exit the data set.
Table 2 provides descriptive statistics of the proposals used in the subsequent analyses. The figures include both censored and noncensored
cases. There are 2,842 bills of ordinary law with the median time of 1,162

1594

Comparative Political Studies

Table 2
Life Span of Bills (in Days)
Number of Bills
By type
Ordinary
Complementary
Constitutional amendment
All bills
By outcome
Approved
Rejected
Other termination
All billsa

Median

Min

Max

2,842
119
105
3,066

1,498
1,417
1,204
1,485

1,162
713
964
1,140

2
20
11
2

5,777
5,630
4,629
5,777

1,344
416
373
3,066

759
1,376
1,864
1,485

509
1,257
1,828
1,140

2
41
7
2

5,155
4,282
4,830
5,777

a. The figures include censored and noncensored cases. There are 933 censored bills.

days (or over 3 years), the minimum of 2 days, and the maximum of 5,777
days (or nearly 16 years). Those figures demonstrate that there is much variation in the bills life span. The bills with 5,777 days of recorded time were
still pending in Congress as of July 2004, indicating that certain bills stay in
Congress for a tremendously long time without any decision. Both bills of
complementary law and constitutional amendment follow similar patterns.
There are 119 complementary law bills and 105 constitutional amendment
proposals with their respective median times being 713 and 964 days.
Table 2 also shows the life span of bills by their outcomes. The data
reveal that approved bills have the shortest life span of the three outcomes.
There are 1,344 approved bills with the mean and median survival times of
759 and 509 days, respectively. Although the Brazilian Congress passed
more than half of the approved bills in less than 2 years, it took much more
time to pass some billsit could take as long as 5,155 days (or 14 years)!
Compared to approved bills, the bills that were eventually rejected or terminated for other reasons tend to stick in Congress much longer. For
example, the median number of days in Congress for rejected bills is 1,257,
which is 2.5 times as much as the median number of days for approved
bills. Finally, the data set contains 933 censored cases.
The covariates included in the analysis are as follows. Hypothesis 1 predicts a greater likelihood of legislative gridlock as bicameral divergences
increase. Brazils fluid multiparty system poses a challenge to a researcher
in measuring the degree of bicameral divergence over time. For example,
there is no simple and meaningful way to calculate the difference between
the partisan compositions of the Senate and the Chamber as it would be

Hiroi / Lawmaking in a Bicameral Legislature

1595

possible in a two-party system, and party switching is common and frequent. However, Brazilian legislative parties do tend to cluster together into
two camps: one supporting the president and the other acting as opposition
to the presidents government.14 I therefore use the percentage difference in
seat shares of the parties in the presidential coalition in the Senate and
Chamber (labeled incongruence) to measure the degree of bicameral divergence. Admittedly, this is a crude measure of bicameral divergence and
assumes that the members of the parties in the presidential coalition
have similar preference portfolios with respect to important legislation.
However, there are reasons for political parties and their members to be in
the presidents support coalition, and many of them do try to vote with the
government whenever possible. In addition, this measure permits me to
record changes in the partisan compositions at much more frequent observation points than using alternative methods (e.g., estimated ideological
locations based on roll-call data), thus allowing for more accurate estimates
of legislative timing.15
Because presidents are important legislative actors in Brazil, I created
another measure of legislative congruence that represents presidents bases
of support in Congress. Presidential Bicameral Majority is a dummy variable and takes the value of 1 when the president has (coalitional) majorities
in both the Senate and Chamber, and 0 otherwise. Both incongruence and
presidential bicameral majority are measures of legislative preferences. I
expect that the larger values of incongruence increase the risk of bill rejection and that presidential bicameral majorities raise the chance of bill
approval. I used deputies and senators changes in party affiliations to construct the partisan composition of the Chamber and Senate using information obtained through the Chambers Secretaria-Geral da Mesa and Centro
de Documentao e Informao (Cmara dos Deputados, 2004), and the
Senates Relatrio da Presidncia (Senado Federal, various issues). In
1988-2004, no presidents achieved legislative majorities solely with their
own parties; however, many presidents did attain coalitional majorities. The
only minority presidents were President Collor (1990-1992) and President
Lula (2003-Present) during his first two quarters. The largest bicameral discrepancies in the sizes of presidential coalitions occurred during the Franco
and Lula presidencies and the beginning of Cardosos first term.
The data analysis examines two factors that may elevate the level of
impatience among legislative actors. First are legislative and presidential
elections. Elections may increase legislative activities because legislators
and presidents allies in Congress are likely to wish to deliver results to
their constituencies when facing upcoming elections. I coded 1 for the

1596

Comparative Political Studies

quarter in which legislative and presidential elections took place and three
quarters prior to it. I expect that legislative productivity generally increases
in pre-electoral periods compared to nonelection years. However, there is a
possibility for an alternative hypothesis. Because most Brazilian legislators
campaign locally and prefer to stay close to their constituencies during
election years, they tend to disappear from the capital during campaign seasons. Consequently, legislative productivity may actually decline during
election years. If such is the case, the hazards of legislative elections will
be lower with respect to both approval and rejection of bills.
The second operationalization of impatience uses an economic variable.
Many developing countries experience severe economic problems rather frequently. I expect that politicians and policy makers imperatives to promptly
cope with economic difficulties become elevated during times of economic
crises. One of the chronic economic problems that Brazilian governments
have faced is inflation. As such, I use monthly inflation rates as a measure
of economic problems and interact it with bills proposing economic policies.
Economic policy bills hazard rate should be significantly positive only
when inflation rates become very high. Rather than arbitrarily deciding the
threshold for very high inflation, I leave the data to determine such levels
and limit my prediction to describing a general pattern. At low inflation
rates, an economic policy bill does not increase its chance of approval.
However, as inflation rates rise, the probability that economic policy bills
will be approved increases. Inflation data are obtained through the Brazilian
Central Banks online database (Banco Central do Brasil, 2004).
Hypothesis 3 pertains to the anticipated duration of legislation. I assess
the impact of the expected duration of legislation by including a dummy
variable representing bills proposing a temporary policy change rather than
a permanent one. Temporary policy changes include those with predetermined expiration dates (e.g., provisional taxes) and those that are subject to
periodic re-examinations (e.g., minimum wages). In total, 103 bills are
coded as provisional legislation. The relative ease of bargaining is expected
to facilitate the approval of those bills compared to bills proposing permanent legislation.
To analyze the impact of asymmetric and symmetric bicameralism
(Hypothesis 4), I use dummy variables that represent supermajoritysymmetric bicameralism (i.e., constitutional amendment proposals) and
absolute majority-asymmetric bicameralism (i.e., bills of complementary
law). The base category is simple majority-asymmetric bicameralism (i.e.,
bills of ordinary law). I refer the reader to Table 2 for the descriptive statistics of those bills.

Hiroi / Lawmaking in a Bicameral Legislature

1597

Hypothesis 5 is about the impact of the sequence of deliberation. It states


that the bills whose examination begins in the lower house have higher
chances of approval than the bills initiated by the upper house. The bills
that were considered by the Chamber first are coded 1 and the bills initiated
by the Senate are coded 0.
In addition, the data analysis takes into account the following factors.
First, the extant literature (e.g., Figueiredo & Limongi, 2001) on Brazilian
legislative politics indicates that the executive is the principal agent of legislation. Thus, I include dummy variables for congressional and judicial
proposals, having executive bills as the base category. Second, the beginning of each legislative period may represent changes in the preferences
with respect to status quo policies as new members join Congress and many
old ones depart. Turnover rates of the members of Congress in Brazil are
quite high, especially in the Chamber of Deputies, where re-election rates
have been around 50% in the last two decades. Hence, I coded 1 the first
year of each 4-year legislative period and 0 otherwise. Third, in addition to
economic policy, the data analysis includes policy areas concerning administrative issues, rights and codes (such as civil and penal codes), political
and institutional questions, and tributes (such as renaming a highway in
honor of a famous politician). The base category is social policy. Finally,
the models are estimated with a series of dummy variables representing different administrations since the promulgation of the 1988 Constitution.
Last, incongruence and inflation variables are time-varying covariates.
For computational convenience, these continuous time-varying covariates
are monthly averages over a quarter. Incongruence has a mean of 9.11 ranging from 0.36 to 24.83. Monthly inflation rates range from 0.28% to
75.22% with an average of 10.3%.

Estimation Results
Table 3 presents the results of Cox competing risks models of bill
approval and rejection in the Brazilian Congress. As discussed before, a hazard rate is the instantaneous rate of an events occurrence, that is, the probability that an event (bill approval or rejection) occurs at time t, given that
the case has survived until t. Tests of the proportional hazard assumption
based on Schoenfeld residuals indicate potential nonproportional hazards for
government bicameral majority, supermajority-symmetric bicameralism,
congressional and judicial bills, and temporary change in the case of the
approval model, and for supermajority-symmetric bicameralism in the

1598

Comparative Political Studies

Table 3
Cox Competing Risks Models of Bill Approval and Rejection

Bicameral incongruence
Presidential bicameral majority
Presidential bicameral majority bicameralism*Time
Supermajority/Symmetric bicameralism
Supermajority/Symmetric bicameralism*Time
Absolute majority/Asymmetric bicameralism
Congressional election
Presidential election
Beginning of legislature
Congressional proposal
Congressional proposal*Time
Judicial proposal
Judicial proposal*Time
Chamber of Deputies first house
Temporary change
Temporary change*Time
Inflation
Inflation*Economic
Economic
Administrative
Rights and codes
Tribute
Political-Institutional
Log pseudo-likelihood
N

Approval

Rejection

0.01
6.17****
0.78****
3.36****
0.54****
0.19
0.22
0.30**
0.35****
3.18****
0.36****
0.41****
0.03
0.97****
3.96****
0.51****
0.002
0.01***
0.03
0.23***
0.41****
0.93****
0.15
9483
52,950 (1,344)

0.04**
0.50

9.53****
1.49***
0.47
0.64*
0.90**
0.91****
1.29****

-0.53****
-0.47

0.01
0.02**
0.08
0.19
0.11
-0.05
0.12
2863
52,950 (416)

Note: Entries are coefficients. Entries in the parentheses are the numbers of failures. Significance
tests used a two-tailed test and robust standard errors. There is no rejected judicial proposal.
Coefficients are estimated with various government dummies (not shown in the table).
*p .1. **p .05. ***p .01. ****p .001.

rejection model. I interacted these variables with the natural logarithm of


time to correct for the potential violations.
As predicted, bicameral incongruence has a positive and significant effect
on bill rejection. The hazard ratio can be obtained by exponentiating the coefficient, exp(0.04) = 1.04. Because the values of incongruence range from 0.36
to 24.83 and their corresponding hazard ratios from 1.01 to 1.58, going from
the minimal value of bicameral incongruence to that of the maximal value
raises the risk of bill rejection by 56.4%. In other words, when the difference
between the shares of seats held by presidential coalitional parties in the two
houses grows from 0.36% to 24.83%, the probability that a bill gets rejected

Hiroi / Lawmaking in a Bicameral Legislature

1599

at a given moment increases by 56.4%. Bicameral incongruence has no statistically significant impact on bill approval, however. This finding suggests
that bicameral incongruence is an important determinant of bill rejection, but
it alone has no influence on bill approval.
What does affect the chances of bill approval is the congruence of the
three institutional legislative actors: the president and the lower and upper
houses of Congress. When presidents hold bicameral majorities, the likelihood that Congress will approve of the bill increases significantly. This may
imply that Congress anticipates presidents preferences and responses when
considering bills. However, the impact of presidential bicameral majorities
diminishes over time without the bills approval. For example, a bill that has
stayed in Congress for a year is 4.71 times more likely to be approved when
presidents sustain majorities in both houses than when they lack such majorities. In contrast, approval under presidential bicameral majorities is only
twice more likely if a bill has been pending in Congress for 3 years. Finally,
such majorities (or lack thereof) do not influence the hazard of bill rejection.
As expected, deliberation under symmetric bicameralism with a supermajority rule delays bills approval significantly and increases the risk of rejection compared to deliberation under an asymmetric bicameral setting.
However, these effects also lessen as bills stay longer in Congress. Moreover,
under asymmetric bicameralism, there is no significant difference in either
bill approval or rejection when only voting quotas are compared, as nonsignificant coefficients of the absolute majority/asymmetric bicameralism
variable indicate. Although it is impossible to separate the effects of symmetric bicameralism from the supermajority rule, these findings combined
suggest that types of bicameralism do have significant impact on the timing
of bill approval and rejection.
Turning to the impatience variables, the estimation results of electoral
periods indicate that the hazard of bill approval rises and that of bill rejection
decreases during presidential election years.16 That is, the Brazilian Congress
is more likely to approve bills in a year running up to presidential elections
and delays the rejection of bills that are eventually rejected. In contrast, the
hazard of bill rejection rises in legislative election years but that of bill
approval is not statistically significant. These findings provide mixed evidence for the impatience hypothesis. The coefficient of bill approval during
presidential elections is as predicted by the theory. However, estimation
results of legislative elections do not render strong support for either the impatience hypothesis or the alternative hypothesis. If the impatience hypothesis
were correct, legislative elections coefficient would be positive and significant for bill approval. If the alternative hypothesis (that legislative activities

1600

Comparative Political Studies

Figure 1
Conditional Hazard Ratios of Economic Bills at Various Inflation Rates

Conditional Hazard Ratios

7.0
6.0
5.0

Rejection

4.0

Approval

3.0
2.0
1.0

80

75

70

65

60

55

50

45

40

35

30

25

20

15

10

-5

0.0
Monthly Inflation Rates

Note: The solid lines indicate where hazard ratios are significant at least at p = .1. The discontinuous lines are those that do not achieve statistical significance.

decline during legislative election years because legislators are on campaign


tours) were right, then the coefficient for bill rejection should also be negative. One of the possible explanations may be that legislators find it more
electorally useful to orchestrate position-taking by rejecting bills than by
approving them. This finding calls for more research in the future.
Another measure of impatience is inflation rates. The interaction of
inflation rates and economic policy bills is positive and significant, suggesting that inflation raises the hazard of bill approval and rejection in the
economic policy areas. Figure 1 shows conditional hazard ratios of economic bills (the y-axis) at various inflations rates (the x-axis). Note that the
hazard ratios are only significant at higher inflation rates (monthly inflation
rates of 10% or higher for rejection and 15% or higher for approval), confirming the hypothesis that only economic crises, but not mere economic
problems, compel politicians and policy makers to work more efficiently in
approving and rejecting bills. The results also show that the chances of economic policy bills being approved and rejected grow progressively higher
as inflation rates become elevated. For instance, the likelihood that economic

Hiroi / Lawmaking in a Bicameral Legislature

1601

policy bills will be approved is 24% higher than that for social policy bills
(i.e., the base category) when a monthly inflation rate is 25%; however, at
a monthly inflation rate of 75%, such likelihood is 120% greater for economic policy bills compared to social policy bills.
As expected, provisional legislation has a higher hazard of being
approved than permanent legislation. This finding suggests the relative ease
of bargaining in striking an agreement when one proposes a short-term
change rather than a long-term change and has tremendous implications for
legislative agenda-setting. It also implies the difficulty of pursuing a longterm reform through legislation. But the advantage of a provisional policy
change in a bills passage declines with time. For instance, a 3-month old bill
proposing a provisional change is five times more likely to be approved than
an equivalent bill proposing a permanent change. However, this advantage
diminishes to 2.5 to 1 in a year and 1.8 to 1 in 2 years. There is no advantage of it being a provisional reform if a bill has been pending for 3 years.
With respect to the effects of bicameral sequence, when the Chamber of
Deputies is the first house to deliberate, a bill has a much higher chance of
being approved on a given day (163% higher) and a lower risk of being
rejected (41% lower) compared to when the Senate is an initiating house.
These results hold even though the estimated models control for the effects
of the executive bills (as a base category comparing bills institutional origins) whose consideration always begins in the lower house. These findings
support Rogerss (1998) argument about a sequential advantage of a bill
being examined in the lower house first and counter the conventional wisdom of bicameralism that emphasizes senatorial expertise. The estimation
results also indicate, however, that congressional bills as a whole do not
fare very well compared to executive bills.
Finally, the beginning of a legislative period induces greater legislative
activities in both approving and rejecting bills. These findings are consistent with the expectation that fundamental changes in the institutional legislative preferences occur at the beginning of each Congress due to the
departure of many former incumbent legislators and the arrival of new ones.

Conclusion
This article examined the dynamics of lawmaking in a new presidential
bicameral democracy taking Brazil as a case. It contributes to comparative
legislative research by analyzing the causes of bill approval and rejection
with new legislative data. Rather than using aggregate data on legislative

1602

Comparative Political Studies

gridlock as much previous work has done, I performed more nuanced


analyses of individual bills submitted to the Brazilian Congress using an
event history method.
This article also analyzed the impact of bicameralism on legislative
performancean often overlooked but important institutional feature of
lawmaking in many new democracies. Although legislative research in new
democracies has increased in recent years, most of these studies tend to
model bicameral legislatures as single-chamber congresses. However, this
article demonstrated that bicameralism is consequential for policy-making
even in nascent democracies. In particular, bicameral incongruence raises
the risk of bill rejection, and presidents bicameral (coalitional) majorities
increase the chances of bill approval. The data analysis also showed that
passing bills under symmetric bicameralism with a supermajority rule faces
significant challenges. Furthermore, a bicameral sequence matters: Bills
that are introduced in the lower house first enjoy greater legislative successes than the Senates bills.
In addition to the effects of bicameralism, this study also confirmed the
importance of bargaining in a legislature. The bargaining literature has
shown that the timing of concession, or reaching an agreement, critically
depends on the levels of actors impatience to conclude such a deal. The
data analysis indicated that acute economic crisesa situation in which
policy makers are likely to be impatienttend to stimulate policy-making
activities both in approving and rejecting economic bills. But evidence also
revealed that elevated levels of economic distress (i.e., crisis), and not mere
problems, were necessary to increase legislative productivity. This finding
is also ironic: If a legislature needs an economic crisis to act, it is not a competent legislature. A more incisive legislature would address problems at
their initial stages before they escalate into crises.
The empirical analysis also indicated that the likelihood of approval is
higher for bills proposing provisional changes rather than permanent
changes. This finding suggests a relative ease of bargaining in passing provisional changes compared to long-term changes. It also has important
implications for legislative agenda-setting. Particularly, if legislators and
government are interested in a quick approval of certain measures, the best
way to accomplish this is to propose temporary legislation. This finding,
unfortunately, also implies a difficulty of pursuing long-term reforms.
Perhaps that is why postdemocratization Brazil has had a series of mini
reforms in the economic areas but its significant economic reforms have
been very few.

Hiroi / Lawmaking in a Bicameral Legislature

1603

Future research should take a step further and identify and include legislative priorities in a quantitative analysis. As Ames (2001) correctly points
out, there are many important legislative issues that never even reach
Congress. On the other hand, there are many nonpriority and/or noncontroversial bills that do get proposed and approved. This study did not distinguish priority bills from nonpriority ones. However, it would be important
to understand how priority bills fare in a bicameral legislature.

Notes
1. Examples of prior research on lawmaking in Latin America include Amorim Neto and
Santos, 2003; Ricci, 2006; Ricci and Lemos, 2004; Saiegh, 2003; Taylor-Robinson and Diaz,
1999.
2. For example, Tsebelis and Money (1997), Binder (1999, 2003), and Druckman and
Thies (2002) deal with U.S. presidential or European parliamentary bicameral systems.
3. This statement assumes that both chambers are endowed with veto rights. However,
even if one of the chambers has only the power to delay, modeling legislative politics in a
bicameral legislature as if it were unicameral would be fallacious (Tsebelis & Money, 1997).
4. Alesina and Drazen (1991) uses a war of attrition model to explain rational delay in economic stabilization policies. Tsebelis and Money (1997) models bicameral bargaining as a
divide-the-dollar game.
5. Alesina and Drazen (1991) and Fearon (1998) made similar arguments with regard to
bargaining over economic stabilization and creation of international regimes, respectively.
6. Neivas (2006) study comparing constitutional powers of upper houses indicates that the
Brazilian Senate is one of the most powerful upper chambers in the world.
7. During President Luiz Incio Lula da Silvas first term (2003-2006), the Senate defeated
certain government projects, thereby challenging the perspective that the upper house is
always a pro-executive chamber.
8. The latent survivor time approach assumes that there are K specific events that a case is
at risk of experiencing but only the incidence of the first event is observed. For example, if we
observe a rejection of a bill in Congress, we would not observe its approval that might eventually occur were it continue to stay in Congress. See Diermeier and Stevenson (1999).
9. The author of a bill may appeal the decision to the executive board of the house.
10. The median time that a bill spends in the Brazilian Congress without a decision is 509
days if eventually approved, 1,257 days if rejected, and 1,828 days if terminated for other reasons (see Table 2). As a comparison, the median time approved bills spent in the Chilean
Congress during the 1990-1993 legislature is 149 days for presidential proposals and 448 days
for congressional proposals (Siavelis, 2002, p. 87). In the United States, all bills not voted
within a 2-year legislative period are terminated.
11. For example, there is a cost in delaying the adoption of legislation aimed to improve
the fiscal balance as deficits grow in its absence.
12. Legislative delays are indeed an important measure of gridlock alternative to a more
commonly used measure of gridlock in terms of the quantity of bills passed. See Binder, 2003,
pp. 75-79.
13. I thank Fernando Limongi for sharing Cebraps legislative data set. I benefited from
their coding schemes in constructing my data set.

1604

Comparative Political Studies

14. See Amorim Neto and Santos (2001) and Amorim Neto, Cox, and McCubbins (2003).
15. As an alternative operationalization of bicameral incongruence, I also used the difference between the Chambers and Senates median voters W-NOMINATE scores. However,
W-NOMINATE scores are not comparable across different houses or different congresses.
Furthermore, these scores are available only for the 49th and 50th Congresses. Perhaps due
to these problems, this variable had no significant impact in the event history analysis and
is therefore not included in the further discussion. I thank Scott Desposato for sharing his
W-NOMINATE data.
16. Presidential and legislative elections became concurrent from 1994.

References
Alesina, A., & Drazen, A. (1991). Why are stabilizations delayed? American Economic Review,
81(5), 1170-1188.
Ames, B. (2001). The deadlock of democracy in Brazil. Ann Arbor: University of Michigan Press.
Amorim Neto, O., Cox, G. W., & McCubbins, M. D. (2003). Agenda power in Brazils Cmara
dos Deputados, 1989-98. World Politics, 55, 550-578.
Amorim Neto, O., & Santos, F. (2001). The executive connection: Presidentially defined factions and party discipline in Brazil. Party Politics, 7(2), 213-234.
Amorim Neto, O., & Santos, F. (2003). The inefficient secret revisited: The legislative input
and output of Brazilian deputies. Legislative Studies Quarterly, 28, 449-479.
Banco Central do Brasil. (2004). Sries temporais [Time series]. Retrieved October 1, 2004,
http://www.bcb.gov.br
Binder, S. A. (1999). The dynamics of legislative gridlock, 1947-96. American Political
Science Review, 93(3), 519-533.
Binder, S. A. (2003). Stalemate: Causes and consequences of legislative gridlock. Washington,
DC: Brookings Institution.
Bottom, W. P., Eavey, C. L., Miller, G. J., & Victor, J. N. (2000). The institutional effect on
majority rule instability: Bicameralism in spatial policy decisions. American Journal of
Political Science, 44(3), 523-540.
Box-Steffensmeier, J. M., & Jones, B. S. (2004). Event history modeling: A guide for social
scientists. New York: Cambridge University Press.
Cmara dos Deputados. (2003-2004). Projetos de lei e outras proposies [Bills of law and other
propositions]. Retrieved August 1, 2004, from http://www2.camara.gov.br/proposicoes
Cmara dos Deputados. (2004). Changes in party affiliation [Information obtained through
Centro de Documentao e Informao]. Brasilia, Brazil: Author.
Cheibub, J. A., Przeworski, A., & Saiegh, S. M. (2004). Government coalitions and legislative
success under presidentialism and parliamentarism. British Journal of Political Science,
34, 565-587.
Cox, G. W., & Kernell, S. (1991). Conclusion. In G. W. Cox & S. Kernell (Eds.), The politics
of divided government (pp. 239-248). Boulder, CO: Westview.
Cox, G. W., & Morgenstern, S. (2002). Epilogue: Latin Americas reactive assemblies and
proactive presidents. In S. Morgenstern & B. Nacif (Eds.), Legislative politics in Latin
America (pp. 446-468). Cambridge, UK: Cambridge University Press.
Desposato, S. W. (2006). Parties for rent? Ambition, ideology, and party switching in Brazils
Chamber of Deputies. American Journal of Political Science, 50(1), 62-80.

Hiroi / Lawmaking in a Bicameral Legislature

1605

Diermeier, D., & Stevenson, R. T. (1999). Cabinet survival and competing risks. American
Journal of Political Science, 43(4), 1051-1068.
Druckman, J. N., & Thies, M. F. (2002). The importance of concurrence: The impact of
bicameralism on government formation and duration. American Journal of Political Science,
46(4), 760-771.
Fearon, J. D. (1998). Bargaining, enforcement, and international cooperation. International
Organization, 52(2), 269-305.
Figueiredo, A. C., & Limongi, F. (1996). Congresso Nacional: Organizao, processo
legislativo e produo legal [National Congress: Organization, legislative process, and law
production]. Sao Paulo, Brazil: Cadernos de Pesquisa.
Figueiredo, A. C., & Limongi, F. (2001). Executivo e legislativo na nova ordem constitucional
[Executive and legislative in the new constitutional order]. Rio de Janeiro, Brazil:
Editora FGV.
Hammond, T. H., & Miller, G. J. (1987). The core of the constitution. American Political
Science Review, 81(4), 1155-1174.
Krehbiel, K. (1998). Pivotal politics: A theory of U.S. lawmaking. Chicago: University of
Chicago Press.
Lijphart, A. (1999). Patterns of democracy: Government forms and performance in thirty-six
countries. New Haven, CT: Yale University Press.
Linz, J. (1994). Presidential or parliamentary democracy: Does it make a difference. In J. Linz
& A. Valenzuela (Eds.), The failure of presidential democracy: The case of Latin America
(pp. 2-87). Baltimore: Johns Hopkins University Press.
Mainwaring, S. (1999). Rethinking party systems in the Third Wave: The case of Brazil.
Stanford, CA: Stanford University Press.
Mayhew, D. R. (1991). Divided we govern: Party control, lawmaking, and investigations,
1946-1990. New Haven, CT: Yale University Press.
Morgenstern, S. (2002). Patterns of legislative politics: Roll-call voting in Latin America and
the United States. Cambridge, UK: Cambridge University Press.
Neiva, P. R. P. (2006). Os determinantes da existnica e dos poderes das cmaras altas:
Federalismo ou presidencialismo? [Determinants of the existence and powers of upper
chambers: Federalism or presidentialism?]. DADOS, 49(2), 269-299.
Poole, K. T., & Rosenthal, H. (1997). Congress: A political-economic history of roll call
voting. New York: Oxford University Press.
Ricci, P. (2006). O Contedo da produo legislativa brasileira: Leis nacionais ou polticas
paroquiais? [The content of Brazilian legislative production: National laws or parochial
policies?] DADOS, 46(4), 699-734.
Ricci, P., & Lemos, L. B. (2004). Produo legislativa e preferncias eleitorais na Comisso
de Agricultura e Poltica Rural da Cmara dos Deputados. [Legislative production and
electoral preference in the Committee on Agriculture and Rural Policies in the Chamber of
Deputies]. Revista Brasileira de Cincias Sociais, 19(55), 107-130.
Riker, W. H. (1992). The justification of bicameralism. International Political Science
Review, 13(1), 101-116.
Rogers, J. R. (1998). Bicameral sequence: Theory and state legislative evidence. American
Journal of Political Science, 42(4), 1025-1060.
Rubinstein, A. (1982). Perfect equilibrium in a bargaining model. Econometrica, 50, 97-109.
Saiegh, S. M. (2003). Government defeat: coalitions, responsiveness, and legislative success.
Unpublished doctoral dissertation, New York University.

1606

Comparative Political Studies

Senado Federal. (2002). Constituio da Repblica Federativa do Brasil [Constitution of the


Federative Republic of Brazil]. Brasilia, Brazil: Author.
Senado Federal. (2003-2004). Atividade legislativa [Legislative activity]. Retrieved August 1,
2004, from http://www.senado.gov.br/sf/atividade/Materia/pesqAvancada.asp
Senado Federal. (Various issues). Relatrio da Presidncia [Report from the Presiding Office].
Brasilia, Brazil: Author.
Shugart, M. S., & Carey, J. M. (1992). Presidents and assemblies: Constitutional design and
electoral dynamics. Cambridge, UK: Cambridge University Press.
Siavelis, P. M. (2002). Exaggerated presidentialism and moderate presidents: Executivelegislative relations in Chile. In S. Morgenstern & B. Nacif (Eds.), Legislative politics in
Latin America (pp. 79-113). Cambridge, UK: Cambridge University Press.
Stepan, A. (2000). Brazils decentralized federalism: Bringing government closer to the citizens?
Proceedings of the American Academy of Arts and Sciences, 129(2), Brazil: The Burden of
the Past; The Promises of the Future. Boston: Metcalf.
Taylor-Robinson, M. M., & Diaz, C. (1999). Who gets legislation passed in a marginal legislature and is the label Marginal Legislature still appropriate? A study of the Honduran
Congress. Comparative Political Studies, 32(5), 589-625.
Tsebelis, G. (2002). Veto players: How political institutions work. Princeton, NJ: Princeton
University Press.
Tsebelis, G., & Money, J. (1997). Bicameralism. Cambridge, UK: Cambridge University Press.
Weaver, K. R., & Rockman, B. A. (Eds.). (1993). Do institutions matter? Government capabilities in the United States and abroad. Washington, DC: Brookings Institution.

Taeko Hiroi is an assistant professor of political science at the University of Texas at El Paso.
Her research interest is in the areas of democratic institutions, legislative politics, political
economy, and Latin American politics. Her most recent publications appear in Latin American
Perspectives, Journal of Developing Societies, International Interactions, and Brazilian Journal
of Political Economy.