On the Effects of Legislative Rules
Author(s): Gary W. Cox
Source: Legislative Studies Quarterly, Vol. 25, No. 2 (May, 2000), pp. 169-192
Published by: Comparative Legislative Research Center
Stable URL: http://www.jstor.org/stable/440367 .
Accessed: 27/06/2014 19:31
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
.
Comparative Legislative Research Center is collaborating with JSTOR to digitize, preserve and extend access
to Legislative Studies Quarterly.
http://www.jstor.org
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
GARY W. COX
Universityof California,San Diego
On
of
the
Effects
Legislative
Rules
In this essay, I consider how a legislature's rules of procedurecan affect both
the process and the outcome of legislation. I begin by asking whetheror not rules of
procedure should have any effects at all, given that they can often be changed by
simple majorities of legislators. The second part of the essay classifies the effects
that rules have. Rules can change the set of bills that plenary sessions of the legislature consider;they can change the menu of amendmentsto any given bill considered
in the plenary;they can affect how membersvote; and-putting the first three effects
together-they can affect which bills pass. I review evidence that rules do in fact
have the suspected effects.
Introduction
In this essay, I considerhow a legislature'srulesof procedure
canaffectboththeprocessandtheoutcomeof legislation.By legislative
rulesof procedureI meanboththestandingordersthelegislaturemay
establishforitselfandthosestatutoryorconstitutional
provisionsthat
materiallyaffectthe legislature'sprocessingof bills (e.g., urgency
provisionsin LatinAmerica).Thediscussionis dividedintotwomain
parts.
First,I considerwhetheror not rulesof procedureshouldhave
any effects at all, given thatthey can often be changedby simple
majoritiesof legislators.Onewaythatthis concerncanbe expressed
is intermsof policyinstability.Inmultidimensional
spatialmodelsof
legislativedecisionmakingunderpuremajorityrule,thegenericresult
is instability-thatis, therealmostneverexistsa policythatcannotbe
defeatedin a pairwisemajorityvoteby someotherpolicy(McKelvey
and Schofield1986;Saari1997).Riker(1980) has arguedthatrules
arevaluedprimarilyfor theiranticipatedeffect on policy outcomes,
so thatgenericallytherewill be no stabilityin the choiceof rules,just
as there is none in the choice of policies. By this argument,one cannot
point to the rules as playing any systematic role in determining
LEGISLATIVESTUDIESQUARTERLY,XXV, 2, May 2000
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
169
170
Gary W. Cox
legislative outcomes. Here, I do not consider Riker's "inherited
instability"argumentbased on the multidimensionalspatialmodel but
insteadfocus on a relatedargumentbasedon the unidimensionalspatial
model.
The second partof the essay classifies the effects thatrules have.
Rules can change the set of bills that plenary sessions of the legislature consider;they can change the menu of amendmentsto any given
bill consideredin the plenary;they can affecthow membersvote; andputting the first three effects together-they can affect which bills
pass. I review evidence thatrules do in fact have the suspectedeffects.
Can Legislative Rules Have Any Effects,
If They Can Be Changed by the Assembly?
In studying how rules of procedureaffect legislative outcomes,
we mustmake an importantinitialdistinctionbetweenexogenous rules
and endogenous rules. Exogenous rules are those that cannot legally
be changedby the legislatureitself; changesrequirethe assent of some
other actors separatefrom or externalto the legislature.Endogenous
rules, in contrast,are those that can legally be changedby the legislature itself.'
Most studies of the effects of legislative rules take them to be
exogenous. For example, in the 1970s and 1980s, much of the literature stressing the importance of committees in the U.S. Congress
assumed, either implicitly or explicitly, thatthe committees'jurisdictions were stableandthatthe senioritynormwas essentiallyinviolable.
Studyingthe effects of rules by assumingthatthey are exogenous is a
useful endeavor,butit can be analogousto studyingpresidentialvetoes
in the U.S. under the assumptionthat vetoes cannot be overridden.
When rules can be overturnedby the very actors whose behavior is
supposedly constrainedby the rules, doubts are legitimately raised
about how effective those rules can be.2
Suppose,for example,thatthe currentrulesof procedurein some
legislature promote a particularpolicy choice, X, but that a majority
exists in the legislaturethat would prefera differentpolicy, Y. In this
case, the majority preferringY could replace the currentrules with
ones consistent with passing Y, and then pass Y. Knowing this, any
agents empoweredunderthe currentrules might acquiesce in Y's passage ratherthanobstructingit or amendingit, as they would be able to
do were the rules exogenous. Thus, one might argue that stable rules
of proceduremust either be inherentlyneutralas far as policy choice
is concerned or they must be consistent with the equilibriumpolicy
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
171
choice under majorityrule. For example, in the case of a unidimensional policy space, the equilibriumpolicy choice undermajorityrule
is the medianvoter's ideal point(Black 1958). It shouldnot be possible
to maintain rules that potentially interferewith the selection of the
median voter's ideal policy unless that potential is never actualized.
Although scholars used to take the effectiveness of rules of
procedure for granted, arguments such as the one just given have
suggested that rules might have no independentcausal force, at least
when they can be changedby a simple majorityof those subjectto the
rules (Krehbiel 1991, 1993). Reacting against this sort of argument,
scholars have suggested a variety of ways in which the causal impact
of rules, which continuesto be widely acceptedas an empiricalmatter,
can be theoreticallyexplained. I consider three such argumentsnext.
TheRules Are in the Constitution(or OtherwiseEntrenched)
A first possibility is thatthe legislature's internalrules of procedure are stable because they are stipulatedin the constitutionand the
constitutioncannotbe amendedby a simple majorityof the assembly.
If some legislative rules are entrenchedin the constitution, and they
lead to a policy choice, X, will that be stable?If a coalition (of legislators and nonlegislators) existed that was large enough to amend the
constitution,preferredY to X, andcould find a set of rulesthatyielded
Y, then X would not be stable and neitherwould the rules leading to
it.3 But, if one continues to assume a unidimensional policy space,
then constitutionallyentrenchedrules will enable the stabilizationof
nonmedianpolicy outcomes. For example, if the constitutionrequires
a three-fourthsmajorityof the assembly to amend, then the range of
stable policies expands from the median to the interquartilerange; if
the constitution requires another body to concur by majority, then
anythingbetween the medians of the two bodies would be stable, and
so forth.
Empiricalexamplesof entrenchedrulesof procedureincludeboth
constitutionalstipulationssuch as the vote bloque in France(cf. Huber
1996) or the urgency procedure in Brazil and Chile and
nonconstitutionalstipulationssuch as the U.S. Senate's filibusterand
cloture provisions (Binder 1997; Binder and Smith 1997). More
generally, Doring (1995b, 36) reportsthat, of 16 countriescovered in
a multiauthor study of Western European legislatures, 6 required
supermajoritiesto changethe legislative rules of procedure,while Cox
and Morgenster (2000) reportthat in 7 LatinAmericancountriesthe
presidenthas entrenchedagenda-settingpowers of one sortor another.
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
172
Gary W. Cox
TheRules Suit the MajorityParty and It Can Protect Them
If party leaders can expel members from legislative caucuses,
deny them renomination,or deny them future office opportunities,
then the majorityparty(or coalition) may be able externallyto enforce
a given set of rules. Cox and McCubbins(1994) tell such a story for
the U.S. House of Representatives.Whateverone thinks of their story
in the U.S. case, something like it seems plausible in countries such as
Costa Rica, the United Kingdom, Venezuela, and Taiwan.
TheLegislators Are Too Busy to Change the Rules
Legislation is more like researchand developmentthan it is like
choice among already known alternatives.Members of most major
legislatures are thus constantly strappedfor time.
But overridingany legislative decision made pursuantto the rules
takes time and effort, and overturningthe rules themselves takes even
more time and effort. The costs entailed in overturningrule-based
decisions and rules are thus largely exogenous: they are the costs in
terms of time and effort needed to constructa strategyfor overturning
the decision andto assemblethe requiredcoalitionin supportof change.
If one believes these claims-that it is costly to overturnrule-based
decisions and more costly to overturnrules, andthatthe costs entailed
are largely exogenous opportunitycosts-then one must also believe
that rules have consequences.
Consider, for example, the ability to "veto"bills that U.S. committees are routinely alleged to possess. As bills can be discharged,
one might arguethat committee vetoes can be overriddenwhenever a
floor majoritywishes. But supposea memberfiles a dischargepetition
on a particularbill. She now tries to get other members to sign her
petition but finds that hardly anyone knows what the bill is about.
They are too busy with their own issues. When asked to sign, the other
members face a choice. They can take the petitioner's word that it is
in their intereststo sign. They can take the committee's word that it is
in their interestsnot to sign. Or, they can allocate scarce resources to
investigate the matterfor themselves. Unless the petitioner can find
enough memberswho eitheralreadyknow aboutthe issue at stake and
agree with her or are willing to take her cue over the committee's, the
petition may stall because the other potential signatoriesare too busy
to investigate the matter.
Suppose that the petitionerthinks she might get enough signaturesfrom memberswho alreadyknow theirpreferencesor arewilling
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
173
to take her cue. She now faces anotherhurdle. There is not enough
time on the floor to pass every bill on the calendars,much less every
bill in committee. Does a majorityof the House preferthat the to-bedischargedbill take an incrementof time on the floor, at the cost of
not being able to proceedwith some otherbill (or bills) in the queue?
Which other bill(s) will be sacrificed? Unless these issues can be
negotiated as well, the petition will still fail, even if in a world of
complete information(all members costlessly know their operative
political preferences) and zero transaction costs (no time budget
constrainton the floor) it would pass.
For importantenough bills, the two costs suggested above will
be lower: more members will already know about the bill and have
clear preferences, and more will be willing to bump something else
off the floor in orderto proceed with it. But for less importantbills,
the costs can be prohibitive.Thus,even when vetoes can be overridden
by a simple majority,and even when the bill vetoed would improve
policy for some majorityon the floor, the veto can stand.If one wishes
to reject this conclusion, one must believe that every bill that would
improvethe statusquo policy for some majorityis identifiedandpassed
in each Congress.
The general point is this: Even "suspensoryvetoes" are important and consequential in an assembly in which time is short. As I
shall argue below, positive agenda power-the ability to initiate the
next step in a bill's progresstowardpassage, at a given time-is even
more importantand harderto overturn.4
If Rules Have Effects...
Supposethat one believes that rule-baseddecisions are costly to
overturnand that the rules themselves are even more costly to overturn. This is enough to motivate a belief that rules will have causal
effects butnot enough to say much aboutwhat those effects might be.
In the next sections, I consider some of the effects that rules have.
Rules can have proximal, intermediate,and final effects. The
proximal effects of rules are to distributeresources (e.g., staff) and
agenda power (e.g., a suspensory veto subject to discharge).5 The
distributionof resources and agenda power stipulatedby the rules in
turnaffects the menu of policy choices with which membersare faced
and how members vote on any given policy choice (intermediate
effects). Finally, the ability to set the menu of policy choices and to
affect voting behavior may lead to an impact on the policy actually
chosen.
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
174
GaryW. Cox
The proximaleffects of rulesarethe easiest to describe,although
even herethere is a wide rangeof specific techniques(see, e.g., Doring
1995c). In the next two sections, I will consider the intermediate
effects-how rules affect the menu of choices and voting behavior.
How Rules Affect the Menu of Choices
Agenda powers can be classified in two broadcategories:6
1. The power to put bills on or keep them off the floor agenda
(thereby determiningwhetheror not the floor has the chance to alter
policy along a given dimension);
2. The power to protectbills from amendmenton the floor.
Each of these species of agendapower is differentlyallocated in
different assemblies. In each case, the question is whether the rules
that allocate agenda power produceagendas that are consequentially
different from those that would have been constructed by floor
majorities(whetheror not these agendasthen lead to different policy
outcomes is considered later).
Setting the Floor Agenda
In consideringhow the plenaryagenda is set, I shall build on the
work of Shepsle (1979) and Cox and McCubbins(1999). Shepsle's
model begins with two main elements: a set of w issues, or policy
dimensions, thatthe legislaturemust decide, and a set of n legislators,
each with strictly quasi-concave preferencesdefined over the policy
space. The issues are partitionedinto a numberof jurisdictions. The
membersare divided into a numberof committees (with membership
on more thanone committee possible), each with its own jurisdiction.
Committees are given the exclusive right to propose bills in their
jurisdictions, but a committee'sjurisdictionmay itself be the union of
smaller subjurisdictions(possibly, thoughnot necessarily, attachedto
subcommittees).A committeewith a complexjurisdictionmay propose
a separatebill for each of its subjurisdictions,but it cannot propose
one omnibus bill dealing with all at once.
The sequence of events in Shepsle's model is (1) the committees
propose bills-i.e., put them on the plenary agenda,7(2) the plenary
amends each of the bills put on its agenda by the committees, as its
members see fit, and (3) each bill, as amended,is put to a final up or
down vote. Although this particularsequence is not followed in all
legislatures, something like it is followed in many.
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
175
Cox and McCubbinsalter Shepsle's model in three main ways.
First, they consider fully strategicactors instead of"sincere" legislators. In this respect, they follow Denzau and Mackay (1983) and
Krehbiel(1985). Second, they divide Shepsle's committee stage into
separateagenda-setting,amendment,andfinal passage substages.Just
as the floor legislative process can be divided into these stages, so the
committee legislative process can, too. Third,they considertwo polar
distributionsof agendapower-the flooragendamodel andthe partisan
agenda model.
In the firstmodel,the floor agendais determinedas if by majority
vote in the plenarysession. One way to interpretthis model is literally.
In some cases (e.g., Denmark in the 1980s-see Damgaard and
Svensson 1989; the U.S. in the 1790s), much of the plenaryagenda is
decided by the plenary itself. Shepsle's model can formally accommodate such cases by assuming that there is just one committee-a
committee of the whole. Another way to interpretthe floor agenda
model is to say that there are in fact a numberof distinct committees
but thatthe floor can extractbills from any committee it chooses (via
a dischargepetition in the United States, for example);thus, the committees cannot bottle up bills that a floor majoritywishes to consider.
In the second model, the floor agenda is determined as if by
majorityvote in the majoritypartycaucus.8Again one might interpret
this literally-as a model of those few periods in the U.S. House's
experience when the majoritypartycaucus seemed to rule the roostor indirectly-as a model of committees which must anticipate the
reaction of control committees or majority party leaders (Cox and
McCubbins 1993).
A key considerationin the model is the location of the statusquo
point on each dimension.Normalizing,one can take Fj,the location of
the median legislatoron the floor on thejh dimension,to be F== F= 0.
That is, we locate the zero point on each dimension at the floor median on that dimension. This is done without loss of generality since
the scale is arbitrary.The location of the status quo point, SQj,varies
withj. Positive values of SQjindicateright-of-centerstatusquo points,
while negative values indicate left-of-centerstatus quo points.
Now consider how the plenaryagenda is set. Suppose first that
there is a separatefloor vote on whetheror not to consider each of the
w dimensions of policy. Suppose also that every membervotes his or
her own constituents' interests (or his or her personal beliefs) rather
than following the partyline. If the motion to consider dimensionj is
passed, the ultimateconsequencewill be that policy on dimensionj is
moved from SQjto F.9 If the motion fails, then policy will remain at
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
176
GaryW. Cox
SQ.. Thus, a memberwill vote to consider dimensionj if and only if
she prefersF to SQ.. An agenda formed by a sequence of floor votes
on what to consider next would thus producean agendathat consisted
of all dimensions with status quo points not equal to F.
Nota bene thatconstructingagendasby puremajorityvote in the
plenarydoes not offer any inherentadvantageto the majoritypartyor
coalition. Suppose that a majorityparty exists and is left-leaning, so
thatmost of its membersareto the left ofF, while most of the minority
party's membersare to the right ofF. The bills that the median legislator decides to proceed with concernthe dimensions with status quo
points not equal to F. These status quo points can be put in four basic
categories. First, "farleft" status quo points are so far left that even a
majorityof the left-leaning majoritypartyprefersF to the status quo
(as does a majority of the minority party). Second, "nearleft" status
quo points are such that a majorityof the majorityprefers the status
quo to F (with a majorityof the minoritypreferringthe reverse).Third,
"nearright"status quo points are such that a majorityof the majority
prefersFto the statusquo, but a majorityof the minoritypartyprefers
the status quo to F. Finally, "farright"statusquo points are such that
majoritiesof both partiespreferF to the status quo.10
Fromthese fourpossible locationsfor the statusquo emergethree
possible voting patterns.If the status quo is either far left or far right,
then majoritiesof bothpartieswill supportplacinga bill on the agenda.
If the status quo is near left, then a majorityof the majoritypartywill
oppose placing the bill on the agenda,but lose. If the statusquo is near
right,then a majorityof the minoritypartywill oppose placing the bill
on the agenda, but lose. Thus,in thefloor agenda model, whetherthe
majoritypartydoes betteror worse thanthe minorityin termsofsetting
the agendafor plenary action depends entirely on the location of the
status quopoints on the variousdimensions.If the distributionof status
quo points is symmetricaboutF, then the majorityparty should lose
as often as the minority party loses. If the distributionof status quo
points is skewed to the right, then the majority party (assumed leftleaning) should lose less often thanthe minority.Finally, if the distribution of status quo points is skewed to the left, then the majority
should lose more often thanthe minority.
An alternativemodelof how the plenaryagendais set, the partisan
agenda model, assumes that the majority party has a veto over the
placement of any issue on the floor agenda. More specifically, if a
majorityof the majoritypartyopposes placing a particularbill on the
agenda, it can prevent its appearance.I shall be less concerned with
how such an agendaselection mechanismmight arise1 thanwith what
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
177
the consequences would be, were it to exist. The most obvious consequence would be that no status quo preferredby a majority of the
majority party to the floor median-no near-left status quo in the
example above-could gain a place on the floor agenda. Thus, under
this model, one should never observe the majority party unsuccessfully opposing the placement of a bill on the plenaryagenda.
ComparingPoint Estimates.Cox andMcCubbinspose two questions to assess the success of the majoritypartyin controllingthe floor
agenda. First, how frequentlyis the majoritypartyrolled on agendasetting votes? A roll is countedwhen a majorityof the majorityparty
opposes the placement of a bill on the floor agenda, but loses. If the
majoritycan veto agendaitems,then its agenda-settingroll rateshould
be zero. Second, how frequently does a majority of the majority
coalition oppose the final passage of a bill, but lose? If the only issues
thatwere placed on the agendaby a left-of-centerpartywere those on
which the status quo was right-of-median,then the worst that could
happen to these bills would be that they were amended to the floor
median, in which case the median legislator and all to her left would
still prefer passing the bill to rejecting it. Thus, the majority's final
passage roll rate should also be near zero.
To addressthe first question, Cox and McCubbinsexamined all
5,789 bills that originated in the House in eight selected congresses
and were reportedout of committee. They found that in only four of
these cases (or 0.07%)did a majorityof the majorityparty'scommittee
membersdissent from the committee report.Indeed, majoritydissent
of any magnitude was quite rare. Thus, in the House at least, the
majority'srecordof avoidingissues thatproduceserioussplits on initial
reportwas nearly perfect.
What about the majority coalition's final passage roll rate? In
the 45th to 99th Congresses, Cox and McCubbinsfind that the modal
roll rate for the majoritypartyon final passage votes is zero, with an
average of 3%.
The majority's low roll rates-0.07% at the agenda-settingand
3% at the final passage stage-contrast with ratherhigher figures for
the minority: about 5% at the agenda-setting and 25% at the final
passage stage. Thus, in the U.S. case, the preliminaryevidence tends
to reject the floor model in favor of the partisan(or "cartel")model.
A similar study of the Japanese Diet (Cox, Masuyama, and
McCubbins2000) finds a similar pattern.Governingparties in Japan
arevirtuallynever rolled.Oppositionparties'roll ratesgenerallyrange
from 7% to 50%. Although similar studies have not been performed
for other legislatures, for many of them the expected result of a
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
178
GaryW. Cox
comparablestudy seems clear. In the U.K. House of Commons, the
French National Assembly, or the GermanBundestag, for example,
the government seems clearly to control the agenda. Issues do not
come out of committee unless the governmentwants them to come
out, nor do governmentssustaindefeats on the floor. There are, however, assemblies where the agenda-settingand final passage roll rates
for the government are much higher, such as Denmark (De Winter
1995, 145).
ComparativeStatics. The conflicting point estimatesof the floor
and partisanmodels are importantin indicatingthe centraldifferences
between them. However, the equilibriumpoint estimates of the two
models could be perturbedby any of a numberof small changes in the
model, such as introducinga small amount of uncertaintyabout the
location of status quo points or a small opportunitycost to action.
Such amendmentsare less likely to affect the two models' comparative statics predictions than their point estimate predictions, and so
the former are more importantfrom the point of view of empirical
testing.
To explain the two models' comparativestatics, we will let M
denotethe majorityparty'smedianandm.the minorityparty'smedian.
Underthe floor agendamodel, Cox and McCubbins(1999) show that
the probabilityof the majoritypartylosing an agenda-settingor final
passage vote increaseswith the distancebetweenM andF = F= 0 for
all j, while the probabilitythat the opposition loses an agenda-setting
or final passage vote increaseswith the distance between m.andF for
all j. Underthe cartel agenda model, the second of these comparative
staticsexpectationsholds-the oppositionshouldlose more often since
its median member is more distant from the floor median-but the
first does not-the majoritypartyshould never lose and any fluctuations in its roll rate should be unrelatedto the distance between the
majoritypartyand floor medians.
Cox and McCubbins(1999) find supportin the U.S. case for the
cartel model and reject the floor agenda model. In particular,while
the majorityparty's roll rate is unrelatedto the distance between the
party's median (as measured by Poole-Rosenthal W-NOMINATE
scores)12 and the floor median (measured in the same way), the
minority's roll rate is significantly relatedto the distance between the
minority and floor medians.
In the United States,it is often said that it is not surprisingto find
the majoritypartywinning votes-after all, they are the majority.But
this observationis specious in the context of a unidimensionalspatial
model in which membersareassumedto vote theirown interestssolely.
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
179
Suppose the Democrats are in the majority. If the status quo on a
particulardimension is to the left of the floor median, then the vote
will go against the majorityparty(the Democrats).The median voter
and all to her right will favor pulling the leftward status quo to the
floor median,while many of those to her left will oppose such a move.
Precisely how many oppose the move depends on how far left the
statusquo is, but for statusquo points thatare preferredby the median
Democratto the floor median, a majorityof the Democratswill vote
against passing the bill. In that sense, the party will lose. If, on the
other hand,the statusquo is to the right of the floor median, then the
vote will go against the minority party. So, in a world in which all
issues are unidimensional,most of the battle is to decide which issues
to consider at all. Once a given issue is chosen, and assuming that
nongermaneamendmentscan be beatenoff, the amendmentstructure
and the twisting of armsat vote time are marginalconsiderations.
The predictionthat governmentroll rates will not vary with the
distancebetweenthe government'sandthe assembly's medians,while
opposition roll rates will vary with the distance between the
opposition's and assembly's medians, is also borne out in the case of
Japan(Cox, Masuyama,and McCubbins2000). While there are thus
far no other similar studies, it seems clear that for some, such as the
U.K. House of Commons, the results will be comparableto those in
the United States and Japan.
Regulating Debate andAmendmenton the Floor
But what happens if exogenous events (e.g., SaddamHussein's
invasion of Kuwait) force an issue onto the agenda?What happens if
the Senate is held by the otherparty,which does its best to force issues
onto the agenda?Whathappens,in otherwords, if the majoritycoalition in a particularchambercannotperfectlycontrolits plenaryagenda
and ensure that only unidimensionalbills of a coalition-friendlysort
appearon the agenda?
In these cases, controllingthe flow of amendmentson the floor
may be especially important. The reaction of the majority to an
unavoidable but unpleasantissue may be to package it with several
otherissuesto createan omnibusthatis at leastpalatableto the majority.
But then the majoritywill need to ensurethatthe packageis not picked
aparton the floor, which will entail restrictingthe range and natureof
amendmentsoffered.
Do majority parties or coalitions actually have the ability to
restrict amendmentson the floor in ways that the median voter (on
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
180
GaryW. Cox
some particulardimension)mightnot agreewith?It seems clearenough
that the French governmenthas such an ability via the package vote
(Huber 1996), that the U.K. government has such an ability via the
guillotine (Dion 1997), and that several Latin American presidents
have a similar ability (Cox and Morgenster 2000). Even in the U.S.
House, where there arenot constitutionallyentrenchedagenda-setting
powers, there is a wide arrayof case-study evidence that the majority
partyuses restrictiveprocedureson the floor to protectmajorityparty
legislation. More systematically, Sinclair (2000) shows that bills
reportedfrom committee on partisanvotes have been more likely to
receive restrictiverules in the House.13
How Rules Affect Voting Behavior
Rules affecting the distributionof resources and agenda power
can affect the voting behaviorof members in two broadways: (1) by
allowing agenda settersto manipulatewho can monitorvotes, and (2)
by providingthe wherewithalto make side payments.Let us consider
each of these points in turn.
One way to view legislators' voting behavioris as the net result
of variousdifferentconsiderationsthat can be divided into three main
categories: constituents'preferences,personal preferences,and party
preferences.If a legislator'sconstituents'interests,his or her personal
beliefs, andhis or herparty'sdesirescoincide,voting decisionsareeasy
(Kingdon1989).If these considerationsconflict,however,then manipulatingthe observabilityof members'actionsbecomesparticularlyimportant(Arnold1990).Governingor majoritycoalitionswithagendacontrol
can frame issues in ways thatprotecttheir membersfrom the scrutiny
of theirconstituentsor expose themto the scrutinyof theirpartyleaders.
An exampleof protectionfromconstituents,or "providingcover,"
is an omnibusbill thatincludes a controversialprovisionor two, along
with many popularor necessaryprovisions. Memberscan vote for the
whole andjustify their votes as ways to secure the (locally) popular
bits, while decrying or disowning the (locally) unpopular bits.
Restrictive procedures can be used to ensure that the members are
never faced with an amendment that proposes to remove just the
unpopularparts.Anotherexample of protectionfrom constituentsis a
vote of confidence. In the unlikely event that she is challenged, a
member can apologetically explain that the vote was really aboutthe
continuationin office of the government,so that she was not at liberty
to vote simply on the merits of the issue (on which, of course, she is
100% in accordwith her aggrieved constituents).
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
181
An example of exposure to party leaders' scrutiny is Willy
Brandt's"stay in your seat"confidence vote. Worriedthat some of his
backbenchersmight supportthe no-confidencemotion confrontinghis
governmentif allowed to vote anonymously,Brandtorderedhis troops
to stay in their seats and took advantageof the Germanrequirement
thatan absolutemajorityof legislatorsmust supportthe no-confidence
motion in orderto bring the governmentdown.
Agendapowerandotherresources(e.g., staff) distributedby rules
can also providethe wherewithalto make side paymentsto members.
For example, if party leaders control assignment to committees or
portfolios,desiredassignmentscan be held out as inducementsto good
behavior. In the United States, Cox and McCubbins (1993) provide
and review evidence showing how parties use party assignments to
shore up loyalty on the margin. A similar story seems to play out in
Europe,both in termsof the initialtradingof ministerialportfolios for
party supportof a governmentand in terms of party leaders handing
out plum assignments in the committee or party hierarchyto more
loyal followers(Damgaard1995).Even in Brazil,Santos(1999) reports
that party loyalty enhances a member's chance of being appointedto
the more importantcommittees for some parties.
Two furtherpoints aboutside paymentsought to be made. First,
side paymentsareoften usedto clinch deals or clear legislative hurdles.
Particularisticbenefits may be ends in themselves for the ordinary
member,but for partyleadersthey are also means to the accomplishment of broadergoals. Second, positive and negative side payments
intended to influence members' voting behavior are a part of any
healthylegislative leviathan,butthey areof quitevariableimportance.
In electoral systems that foster personal votes, it is not even in the
party'sbest interestto "force"theirmembersto vote in particularways,
sincethis substantiallyreducestheirprobabilitiesof reelection,thusdamagingthe party'sprospectof attainingor retainingmajoritystatus.In the
UnitedStates,the typicalprocedureis formajoritypartyleadersto "buy"
no morevotes thanthey mustin orderto securea legislativevictory.
The bottom-line question, of course, is whetheror not attempts
by parties to manipulate who can monitor members' votes and to
distributeside paymentsdo in fact influencemembers'voting behavior.
In parliamentary systems, the ability of parties to control their
members' voting behaviorby making issues mattersof confidence is
widely accepted (althoughquestions have been raised aboutthe credibility of governmentthreatsto resign). In contrast,legislative parties
in presidentialsystems do not havethe big gun of confidenceto enforce
discipline at the voting stage. Strong correlationsstill typically exist
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
182
Gary W. Cox
between voting behaviorand partisanaffiliation in such systems, but
it is hard to say whether these correlations arise simply because
membersjoin partieswith which they tend to agree or whetherthere is
also some partyinfluence above and beyond what would be expected
on the basis of members' preferences.
So, how can one detect the influence of partyon roll-call voting?
Some headwayhas recentlybeen made in solving this methodological
problem.If House membersvote with theirpartiesonly when the party
position is congenial to their constituents and/or to their personal
ideologies, then (I) theirvoting behaviorshouldnot change when they
switch parties (unless their constituency or ideology changed when
they switched);(2) there shouldbe no systematicrelationshipbetween
members' reelection success and how frequentlythey vote with their
partyon key partyvotes (by assumption,no one is casting "tough"i.e., electorally costly-votes for their party;thus, if one member is
more loyal to the party than another on key votes, this can only be
becausehis or herconstituencyis more in tunewith the party'sposition,
and such memberswill be no more likely to suffer larger-than-average
vote dropsor actualdefeat than less loyal members);and (3) members
should be no more likely to supporttheirpartythanwould be expected
on the basis of the positionsthey advocatein elections. Yet, (1) several
analyses of party-switchersover the postwarerahave foundsignificant
changes in voting behaviortoward greateragreementwith the newly
joined party (see, e.g., Ansolabehere, Snyder, and Stewart 1999;
McCarty, Poole, and Rosenthal 1999); (2) several studies find
systematiccorrelationsbetween partyloyalty andelectoralrisk (Brady
et al. 1996;Jacobson1996);and(3) Ansolabehere,Snyder,and Stewart
(1999) find that membersof the U.S. House are more likely to support
their party than would be expected from their responses to electoral
questionnaires.It is hard to explain these results on the assumption
that partiesdo not influence their members' voting behavior.
How Rules Affect Final Policy Choices
Inthis section,I considerhow rulesmightaffectfinalpolicychoices.
I first consider how the rules enable the majoritypartyor coalition to
get what it wants, then how they can simplify legislative negotiations.
Helping the Majority "GetIts Way"
Rules can empowerthe majoritypartyor coalition, when there is
at
one, three broad stages of the legislative process: at the agenda-
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
183
setting stage, at the amendmentstage, and at the voting stage. I shall
considereach of these stages,in reverseorder.In each case, the question
addressed is how power at a particularstage can lead to final policy
outcomes favoredby the majority.
The WhipModel. Perhapsthe most frequentlymentionedway in
which a majority party or coalition can get its way is by exerting
discipline over its memberswhen they vote. While this techniquemay
be importantin parliamentarysystems with strongparties,in some of
the presidentialsystems of LatinAmerica and the United States, this
technique is importantonly on the margins.Ratherthan insisting on a
solid block of partyvotes to ensurepassageof their legislation, parties
in some presidentialsystemsallow theirmembersto dissentmoreoften
but still seek to corrala few pivotal votes when needed. The evidence
thatpartiescan actuallypressuretheirmembersintovoting has already
been reviewed. Are the votes in fact pivotal, so that they make the
difference between winning and losing?
The evidence here is largely anecdotal, but there are a lot of
anecdotes, at least for the United States. The standardoperatingprocedure for Speakers in the U.S. House on close votes is to have a
certain number of "vest pocket" or "just-in-case"votes lined up in
advance. If the bill appearsready to fail, the Speakerthen calls in as
many of these votes as are needed to ensure victory, in some cases
even stopping the clock to providetime to find the needed votes.14
The Restrictive Rule Model. One step earlier in the legislative
process than votes on final passage are votes on amendmentsto bills.
And beforethe actualamendmentscome, decisionsmustbe madeabout
which amendmentswill be in order.In cases where the government
has the power to preventall amendmentsand hence presentmembers
with take-it-or-leave-itchoices, it can clearly affect the final outcome.
Theoretically,the impactof this sort of power on the legislative
outcome is captured in the widely known setter model (Rosenthal
1990). In the United States, the partisaneffect of restrictiverules on
outcomes is contested (Krehbiel 1991, 1997a, 1997b). But there is
certainly evidence that the majority party uses restrictive rules on
partisan issues and that the minority complains about it. More
compellingly, members' voting behavior on (rule/bill) pairs is what
one would expect on the hypothesisthatmembersbelieve thatthe rule
will have an effect on the outcome.
To explain this last point, due to Sinclair (2000), we must
rememberthat underHouse procedurea "special rule" is sometimes
adopted to regulate plenary debate and amendment activity on a
particularbill. Members vote first on the issue of whether to accept
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
184
GaryW. Cox
the rule, then on the substanceof the bill. Suppose the majorityparty
fashions the proposedrules in orderto preventamendmentsthat the
minoritywould like to move andthatall membersknow this. Suppose
furtherthat the electoral consequences of votes on rules are murky:
members can always talk about fairness or efficient transaction of
business or minority stalling and majority bullying to shield themselves from any claim that a vote for a rule, R, that facilitates passage
of a bill, B, is really a vote for B, or that a vote against R is really a
vote against B. Suppose finally thatall membersbelieve that R will in
fact facilitate passage of B and that both parties are pressuringtheir
membersto supportthe partyposition.
Giventhese threeassumptions,one expectsthe following patterns
in voting behavioron (rule/bill) pairs. Some majoritypartymembers
will have constituencies that supportthe substance of the bill. They
will vote bothforthe ruleandforthe bill. Othermajoritypartymembers
will have constituencies that oppose the substanceof the bill. These
members are likely to vote against the bill because to do otherwise
incurs electoral risk. Put anotherway, it will be relatively expensive
for the party to buy these members' votes on the bill. In contrast, it
may be considerablycheaperto buy theirvotes on the rule, depending
on how securelyshieldedfromelectoralretributionthey feel. The party
need offer no side paymentat all to memberswho personallyfavorthe
bill. Members who are personally indifferentor mildly opposed can
be bought by compensatingthemjust for their personaldistaste;they
need not be compensatedfor electoral risk (or, at least, the electoral
risk is reducedand hence the needed compensation).
All told then, one expects majoritypartymembers whose constituentsoppose the bill eitherto vote againstboth the rule andthe bill
or to vote for the rule but against the bill (with only a few voting for
both). A similarargumentleads one to expect minoritypartymembers
to fall mostly into two camps:those voting against both rule and bill,
andthose voting againstthe rule butfor the bill. Sinclair(2000) shows
that these opposed expectations about majorityand minority voting
on (rule/bill) pairs do obtain, suggesting supportfor the underlying
assumptionthatmembersbelieve rulesto have causalforce. At present,
this is probablythe best statistical evidence we have that rules affect
outcomes.
The Partisan Agenda Model. Before a bill gets to the stage at
which permissible amendmentson the floor are decided, it has to get
out of committee. The powerto decide which bills make it to the floor
is arguablythe least appreciatedbutmost fundamentalpower in terms
of influencing final outcomes.
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
185
Recall thatunderthe partisanor cartelagendamodel, the majority
coalition monopolizes the plenaryagenda,in the sense thatit is able to
prevent the appearanceon the floor agenda of any bill with which a
majority of the majority would prefer not to deal. With this sort of
agenda selection power, the issues that actuallymake it to the plenary
will all have status quo points that are either to the minority's side of
the floor median or so far to the majorityparty's side that a majority
of the partywould preferpulling them backto the median. Thus, even
if every separatedecision taken by the House is unidimensionaland
ends up at the floor median, the policy location in the multidimensional space in which the whole sequence of decisions takes place will
be substantiallybiased from what it would have been undera neutral
agenda structure.If the majority is left-leaning, movements will be
made leftwardon a good numberof dimensions with relatively fewer
rightwardmovements (and those only to correct "far left" status quo
points). Few rightwardmovements from the "near left" are passed
because no resources are devoted to finding such moves, and if they
do crop up, considerableresourcesare deployed againstthem. The net
result is that policy will be more leftist than would have been the case
undera neutralagenda.
SimplifyingLegislative Negotiations
Besides privileging the majoritycoalition over other legislative
majorities, rules of procedurecan also make commitments to pass
complex packages of legislation more credible, hence less costly to
negotiate.
ConcentratedAgendaPower Leads to Fewer butMore Complex
ConflictualActs.Doring(1995a) andHenning(1995) considera model
in which the government is a monopoly supplier of legislation. The
government can choose either to produce fairly simple and
nonconflictualbills, whose passage will not requirethe use of agenda
power,or to producemorecomplexandconflictualbills, whose passage
will be facilitated by agenda power. They argue that increasing the
concentration and amount of agenda power in the hands of the
governmentlowers the "price"of the more complex and conflictual
bills, while leaving the price of simple nonconflictualbills unchanged.
Accordingly, a governmentwill increase its "purchase"of complex
conflictual bills when its agenda power increases, assuming that all
else is equal and that complex conflictual bills are normal goods (in
the economist's sense). Since (1) complex andconflictualbills always
requiremoretime to processthansimple andnonconflictualbills, even
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
186
GaryW. Cox
when the governmenthas strong agenda powers, and (2) the government faces a fixed, hardtime constraint,any increase in the percentof
complex bills thatthe governmentundertakeswill necessarily mean a
lower volume of legislation overall. Thus, all told, one expects fewer
but more complex conflictual bills and acts (given the high success
rate of governments)when agenda power is higher.
Doring (1995a) provides some cross-nationalevidence for part
of this story in a study of WesternEuropeanlegislatures.Ranking 18
European assemblies in terms of the agenda powers given to the
government,then correlatingthis rankingwith the volume of legislation enacted in each country,Doring finds a negative and significant
relationship.This finding addressesthe issue of the overall volume of
legislation but not the issue of how conflictual the acts are, which
Doring leaves for a latertime.
Some evidence that agenda powers are in fact broughtto bear
mostly on complex and conflictual bills can be gleaned from singlecountrystudies. Many studies of the use of restrictiverules in the U.S.
House find that they are more often used on omnibus or multiplyreferred-which is to say, more complex-bills (Sinclair 1995, 2000).
Similarly, some studies find these rules used more often in cases of
partisanconflict (Sinclair 2000). Huber's (1996) study of the use of
restrictiverules in Francefinds similar tendencies.
ConcentratedAgendaPower Leads to QuickerNegotiationsover
GovernmentFormation. De Winter(1995, 143) arguesthat "in countries where institutional arrangementsgive the executive extensive
legislative agendasettingpowers,governmentformationwill consume
less time, as most matterscan be settled 'on the road'. On the other
hand, in countrieswhere the legislaturehas a relatively stronghold on
its own agenda,coalitionpartieswill preferto settle all or most potential
disputesbefore a cabinet is formedand,thus, formalize agreementsin
a writtencontract...." De Winterfindssupportforthis ideain the factthat
D6ring's index of governmentagendacontrolcorrelatesnegativelyand
significantlywith the lengthof coalitionnegotiationsin WesternEurope.
Therearedifferentpossible interpretations
of this argument.First,
perhapspartieswithout agendapower view themselves as not having
any ability to implementagreementson the road.Thus, they negotiate
more in advance.But why aretheirinitialnegotiationsableto be implemented? Second, perhapsit is possible for partiesto agree on a distribution of portfolios, ratherthana detailedpolicy platform,in systems
with strong agenda power. Third,perhapssome of the contingencies
thatmustbe negotiatedin systems withoutconcentratedagendapower
can be preventedfromever arisingby appropriateuse of agendapower.
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
187
Conclusion
Legislative rules have effects because they distribute real
resourceswhose effects cannotbe undonewithout incurringreal costs
in time and effort, because they confer benefits thatpartiesfind worth
preservingthroughextralegislativemeans,andbecausethey are sometimes entrenchedlegally. In a world in which time is short,the power
to delay or expedite can be crucial, even if decisions to decelerate or
accelerate a particularbill can be appealedto the plenary. In a world
in which the effects of rules on final outcomes are obscure to voters,
members fear electoral retributionfrom their constituents less than
they would on straightforwardvotes on substance. Moreover, the
obscurity of proceduralrules' effects makes it easier for parties to
maintain control of procedural votes than to maintain control of
substantivevotes.
The actualeffects rules have are sortedhere into threecategories:
effects on the menu of choices (which bills areconsideredon the floor?
which amendments are allowed?), effects on voting behavior, and
effects on the final legislative outcome (which bills pass?). Evidence
of such effects is visible in most of the legislaturesthat have actually
been studied, althoughthe qualityof the evidence varies widely, since
it is difficult to make the necessary counterfactualcomparisons(e.g.,
whatwould the agendahave been hadthe rulesbeen different?)cleanly.
Typically, the effect of rules is most visible in conjunctionwith
a majority party or coalition's efforts to push through its legislative
agendaagainstopposition.If one orderedthe world's legislative parties
from those with the greatest incentives to push throughtheir legislative agendas to those with the fewest incentives, the orderingwould
putparliamentarypartiesat the top (the governmentfalls if its program
fails); proactive congressional parties, such as those in the United
States, next (where the legislative parties are sometimes viewed as
having electoral incentivesto prosecuteagendas[Cox andMcCubbins
1993]); and reactive congressional parties, as in Latin America, last
(where the party label means less and, relatedly, there are fewer
incentives to prosecutea partyagenda).
It is interestingto examine the structureof agenda power across
these three levels of incentive. Controlof the agenda is firmest in the
parliamentaryregimes, especially the more Westminsterianones. It is
lodged, moreover, in the handsof legislative partyleaders(who typically assume executive office). Controlof the agenda is next firmest
in the United States. It is again lodged in the handsof legislative party
leaders(who, due to the separationof powers,do not assume executive
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
188
GaryW. Cox
office). Controlof the agendais weakest in LatinAmerica.It is lodged
partly in the hands of the president, an executive official who often
attempts to be "above parties,"and partly in the hands of the mesa
directiva, typically composed of legislative partyleaders.
It is interestingto note also thatthe meaningfulnessof partylabels,
in terms of the policies that each party is likely to pursue, is ordered
roughly the same way as are the incentives to legislate: European
parliamentarydemocracies first, U.S. presidential systems second,
LatinAmericanpresidentialsystemslast.Thereappears,in otherwords,
to be a positive correlationbetween (1) legislative parties' incentives
to legislate (and the frequencywith which they actually do legislate),
(2) the extent to which agenda power is centralized in the hands of
legislative partyleadersof the majoritycoalition, and (3) the electoral
meaningfulnessof the partylabel in policy terms.
Gary W. Cox is Professor of Political Science, University of
California,San Diego, 9500 GilmanDrive,La Jolla, California920930521.
NOTES
I thankthe participantsat the ShambaughConference,University of Iowa, for
theircommentson an earlierdraftof this paper,andChrisDen Hartogfor assistancein
preparingthe manuscript.
1. The best examplesof exogenous rulesarethose stipulatedin constitutions,at
leastwhen the constitutioncannotbe amendedby the legislaturealone butalso requires,
say, presidentialapproval, approval by state assemblies, or approval in a national
referendum.Clearexamples of endogenousrules would be the standingordersof the
U.S. House of Representatives,which can be changedor maintainedat the sole discretion of the House (the courts have repeatedlydeclined to review these rules, citing
Article I, Section 5 of the Constitutionwhich stipulatesthat "EachHouse may determine the Rules of its Proceedings....").
2. An analogous concern has frequentlybeen raised about electoral rules. See
Cox 1997, chapter2.
3. This would be the typical result if the policy space were of high enough
dimensionality,since then even super-or concurrentmajorityrequirementswould not
suffice to stabilize policy (see Cox and McKelvey 1984; McKelvey and Schofield
1986; Saari 1997). But note that these results from cooperativegame theory assume
that it is costless to change the rules-no time elapses, no one needs to examine the
consequences of changing to a particularnew set of rules as such consequences are
utterlytransparentto all, and so forth.Note also that noncooperativeapproachescan
give quite differentresults-consider Diermeierand Feddersen 1998.
4. If rule-baseddecisions-such as a committeedeciding not to proceedwith a
bill-are costly to overturn,what about ripping up the rules themselves? Instead of
discharging the committee of a particularbill, would it be cheaper to abolish the
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
189
committee,pack it, or stripit of jurisdiction?One supposes thatthe answer is "no"in
this particularexample and more generally.
5. Anotherimportantproximaleffectthatrulescan have,which I do not consider
here, is to delineateinformationalrequirements-e.g., the requirementthatcommittees
file writtenreportsto accompanybills, with documentationof those supportingand
dissenting from the report.
6. Anotherimportantpowerthatmight be classified as an agendapower is that
of settingthe reversionarypolicy outcome.
7. In some legislatures,such as the U.S. House, bills proposedby committees
are not guaranteeda hearingon the floor; they are simply eligible for such a hearing
but must still pass otherhurdles.
8. The model assumes the existence of a majoritycoalition. Here I consider
the case of a majorityparty.
9. This follows if one assumes that nongermaneamendments(i.e., amendmentsthatbringin otherpolicy dimensions)arenot allowed andthatan amendmentto
substitutethe policy F for whateverpolicy the bill proposes is in order.
10. The boundariesbetween these four regions can be defined as follows. Let
Mjbe the median ideal point of the majoritypartyand m. be the medianideal point of
the minoritypartyand assumethatM. < F< mj.Let R(MA)be the point to the majority
median's left that gives utility equal to F (given symmetricutility functions, this will
be the point 2M.-F). Similarly, let R(m.) be the point to the minority median's right
that gives utility equal to F (given symmetricutility functions, this will be the point
2m.-F). Then every point to the left of R(Mj)is "farleft," every point between R(Mj)
and F is "nearleft," every point betweenF and R(mj)is "nearright,"and every point
to the right of R(mj)is "farright."
11. In the U.S. case, likely institutionalbases for a majoritypartyveto over the
placement of items on the plenary agenda include: the scheduling power of subcommitteeandcommitteechairs,the requirementthatevery"spendingbill"passthrough
bothan authorizingcommitteeandthe appropriations
committee(on whichthe majority
partyis disproportionatelyrepresented),the requirementthatmost importantbills pass
throughthe Rules Committee(on which, again,the majoritypartyis disproportionately
represented),andthe Speaker'sandMajorityLeader'scontrolover floor scheduling.
12. See Poole and Rosenthal 1997.
13. As Sinclair 2000 points out, this patternis precisely opposite to what a
majoritarian(or "neutralrules")model would expect (cf. Krehbiel 1991).
14. The only argumentthat this does not representevidence that the majority
partyis "buying"pivotal votes is Krehbiel'ssuggestionthatbothpartiesbuy aboutthe
same numberof votes on any given bill, thus cancelingout each other's efforts. There
is no evidence for this "canceling conjecture,"however, and there is considerable
reasonto doubt it. For example, the majoritypartyhas more resourceswith which to
rewardand punishthanthe minority,which should lead to bettersuccess in the votebuyingmarket.Also, if partyeffortsdo routinelycancelout, one would expectminority
party leaders to be just as prominentin last-minutevote buying as majority party
leaders, which is certainly not the picturepainted in the qualitativeliterature.Also,
why are there not as many anecdotes about minority party leaders pulling out lastminutevictories as thereare aboutmajoritypartyleaders?Eitherthe agendahas been
stackedor the majorityhas a bigger bag of side payments.
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
190
GaryW. Cox
REFERENCES
Ansolabehere,Stephen,JamesM. Snyder,andCharlesStewartIII. 1999. "TheEffects
of PartyandPreferenceson CongressionalRoll-CallVoting."Unpublishedtypescript. MassachusettsInstituteof Technology.
Arnold, R. Douglas. 1990. TheLogic of CongressionalAction. New Haven, CT: Yale
University Press.
Binder, Sarah. 1997. Majority Rule, Minority Rights. Cambridge: Cambridge
University Press.
Binder, Sarah, and Steven Smith. 1997. Politics or Principle? Filibustering in the
UnitedStates. Washington,DC: Brookings InstitutionPress.
Black, Duncan. 1958. TheTheoryofCommitteesandElections.Cambridge:Cambridge
University Press.
Brady, David W., John F. Cogan, Brian J. Gaines, and Douglas Rivers. 1996. "The
Perils of PresidentialSupport:How the RepublicansTook the House in the
1994 MidtermElection."Political Behavior 18:345-67.
Cox, Gary W. 1997. Making VotesCount.Cambridge:CambridgeUniversity Press.
Cox, Gary W., and Mathew D. McCubbins. 1993. Legislative Leviathan. Berkeley:
University of CaliforniaPress.
Cox, GaryW., andMathewD. McCubbins.1994. "Bonding,Structure,andthe Stability
of Political Parties: Party Government in the House." Legislative Studies
Quarterly19:215-31.
Cox, Gary W., and MathewD. McCubbins.2000. "AgendaPower in the U.S. House
of Representatives,1877 to 1986." In Essays on the History of Congress, ed.
David W. BradyandMathewD. McCubbins.Palo Alto, CA: StanfordUniversity
Press (forthcoming).
Cox, Gary W., and Richard D. McKelvey. 1984. "A Ham Sandwich Theorem for
GeneralMeasures."Social Choice and Welfare1:75-83.
Cox, Gary W., and Scott Morgenstern.2000. "Reactive Assemblies and Proactive
Presidents:A Typology of Latin American Presidentsand Legislatures."In
ComparativePolitics (forthcoming).
Cox, GaryW., MikitakaMasuyama,andMathewD. McCubbins.2000. "AgendaPower
in the JapaneseHouse of Representatives."Unpublishedtypescript.University
of California,San Diego.
Damgaard,Erik. 1995. "HowPartiesControlCommitteeMembers."InParliamentsand
MajorityRule in WesternEurope,ed. HerbertD6ring.Berlin:CampusVerlag.
Damgaard,Erik, and Palle Svensson. 1989. "Who Governs? Parties and Policies in
Denmark."EuropeanJournal of Political Research 17:731-45.
Denzau, ArthurT., and RobertJ. Mackay. 1983. "Gatekeepingand Monopoly Power
of Committees:An Analysis of Sincereand SophisticatedBehavior."American
Journal of Political Science 27:740-61.
De Winter, Lieven. 1995. "The Role of Parliamentin GovernmentFormation and
Resignation."In Parliamentsand MajorityRule in WesternEurope,ed. Herbert
DOring.Berlin: CampusVerlag.
Diermeier,Daniel, and Timothy Feddersen.1998. "Cohesionin Legislaturesand the
Vote of ConfidenceProcedure."AmericanPolitical Science Review92:611-22.
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
Legislative Rules
191
Dion, Douglas. 1997. Turningthe Legislative Thumbscrew.Ann Arbor:University of
Michigan Press.
Doring, Herbert.1995a. "FewerThough PresumablyMore ConflictualBills: ParliamentaryGovernmentActing as a Monopolist." In Parliaments and Majority
Rule in WesternEurope, ed. HerbertD6ring. Berlin: CampusVerlag.
Doring, Herbert. 1995b. "Institutionsand Policies: Why We Need Cross-National
Analysis." In Parliaments and MajorityRule in WesternEurope, ed. Herbert
Doring. Berlin: CampusVerlag.
Doring, Herbert. 1995c. "Time as a Scarce Resource: GovernmentControl of the
Agenda." In Parliaments and MajorityRule in WesternEurope, ed. Herbert
Doring. Berlin: CampusVerlag.
Henning, Christian.1995. "A FormalModel of Law Productionby Governmentas a
NaturalMonopoly."In Parliamentsand MajorityRule in WesternEurope, ed.
HerbertD6ring. Berlin: CampusVerlag.
Huber,John. 1996. RationalizingParliament:LegislativeInstitutionsand PartyPolitics
in France. Cambridge:CambridgeUniversity Press.
Jacobson,GaryC. 1996. "The 1994 House Elections in Perspective."Political Science
Quarterly111:203-23.
Kingdon, John W. 1989. Congressmen's Voting Decisions, 3d ed. Ann Arbor:
University of Michigan Press.
Krehbiel,Keith. 1985. "ObstructionandRepresentativenessin Legislatures."American
Journal of Political Science 29:643-59.
Krehbiel, Keith. 1991. Information and Legislative Organization. Ann Arbor:
University of Michigan Press.
Krehbiel,Keith. 1993. "Where'sthe Party?"BritishJournalof PoliticalScience23:23566.
Krehbiel,Keith. 1997a. "Rejoinderto 'Sense and Sensibility."'AmericanJournal of
Political Science 41:958-64.
Krehbiel, Keith. 1997b. "Restrictive Rules Reconsidered." American Journal of
Political Science 41:919-44.
McCarty, Nolan, Keith Poole, and Howard Rosenthal. 1999. "The Hunt for Party
Discipline in Congress."Unpublished manuscript.Carnegie Mellon University.
McKelvey, RichardD., and Norman Schofield. 1986. "StructuralInstability of the
Core."Journal of MathematicalEconomics 15:179-98.
Poole, KeithT., andHowardRosenthal.1997. Congress:A Political-EconomicHistory
of Roll Call Voting.New York: OxfordUniversity Press.
Riker, William H. 1980. "Implicationsfrom the Disequilibriumof MajorityRule for
the Study of Institutions."AmericanPolitical Science Review 74:432-46.
Rosenthal,Howard. 1990. "The SetterModel." In Advances in the Spatial Theoryof
Elections, ed. James Enelow and Melvin Hinich. Cambridge: Cambridge
University Press.
Saari,Donald. 1997. "TheGenericExistence of a Corefor q-rules."Economic Theory
9:219-60.
Santos, Fabiano. 1999. "Party Leaders and Committee Assignments in Brazil."
Presentedat the annualmeeting of the AmericanPolitical Science Association,
Atlanta,Georgia.
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions
192
Gary W. Cox
Shepsle, Kenneth. 1979. "InstitutionalArrangementsandEquilibriumin Multidimensional Voting Models."AmericanJournal of Political Science 23:27-59.
Sinclair, Barbara. 1995. Legislators, Leaders and Lawmaking. Baltimore: Johns
Hopkins University Press.
Sinclair, Barbara.2000. "Do PartiesMatter?"In Essays on the History of Congress,
ed. David W. Brady and Mathew D. McCubbins. Palo Alto, CA: Stanford
University Press (forthcoming).
This content downloaded from 209.91.173.106 on Fri, 27 Jun 2014 19:31:22 PM
All use subject to JSTOR Terms and Conditions