Industrial Organization Congress

The Industrial Organization of Congress; or, Why Legislatures, Like Firms, Are Not Organized as Markets Author(s): Barry
R. Weingast and William J. Marshall Source: The Journal of Political Economy, Vol. 96, No. 1 (Feb., 1988), pp. 132-163 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/1830714 . Accessed: 08/02/2011 13:56
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=ucpress. . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. 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.
The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Political Economy.
http://www.jstor.org
Organizationof Congress; The Industrial or, Why Legislatures,Like Firms, Are Not Organized as Markets
Barry R. Weingast
Stanford University
William J. Marshall
Goldman, Sachs and Company
that parallels This paper provides a theoryof legislativeinstitutions Like and the theoryof contractualinstitutions. the theoryof the firm reflecttwo key compolegislative institutions market institutions, nents: the goals or preferencesof individuals (here, representatives seeking reelection) and the relevant transactionscosts. We present enthree conclusions. First,we show how the legislativeinstitutions forcebargains among legislators.Second, we explain why,given the specific peculiar formof bargaining problems found in legislatures, formsof nonmarketexchange prove superior to marketexchange. Third, our approach shows how the committee systemlimitsthe typesof coalitions that may formon a particularissue.
The organization of Congress meets remarkably well the electoral needs of its members. To put it another way, if a group of planners sat down and tried to design a pair of American national assemblies with the goal of serving members' electoral needs year in and year out,
acknowledge the helpful commentsof Lee Alston, Harold Demsetz, We gratefully Thomas Gilligan,Tom Hammond, Douglass North, Michael Riordan, Brian Roberts, Kenneth Shepsle, and George Stigler. We also thank Elizabeth Case for her editorial assistance.Weingastthanksthe National Science Foundation forpartialsupport (Grant SES-86 17516).
[JournalofPoliticalEconomy, 1988, vol. 96, no. 1] ?) 1988 by The University of Chicago. All rightsreserved. 0022-3808/88/9601-0010$01.50
132
INDUSTRIAL
ORGANIZATION
133
theywould be hard pressed to improve on what exists. 1974, p. 81] [MAYHEW The new economics of organization holds that explicit market exfor a transaction.The change is not the universallyideal institution most successfulapplication of this approach, the theoryof the firm, attemptsto explain, for example, why some transactionstake place withina firmunder certain circumstancesand across a market(e.g., between firms)under others.' This theoryalso focuses on the structure of the corporation, notably the separation of ownership and control (Alchian and Demsetz 1972; Jensen and Meckling 1976; Fama 1980; Fama and Jensen 1983; Demsetz and Lehn 1985; Grossman and Hart 1986). Withfewexceptions,however,it has not considered other typesof organizations,such as public bureaucracies,political parties,or legislatures.2The purpose of this paper is to extend thistheoryto the studyof politicalorganizationsand, in particular,to withinthe legislaturethatfacilitates explain the patternof institutions decision making. emphasize the dependence of politStudies of public policy-making While participation. ical decisions on interestgroup and constituency this approach is consistentwith outcomes in many individual policy areas, it fails to explain how so many diverse interestsare provided with policy benefitssimultaneously.A huge varietyof interestsare represented in the legislature,and almost none is represented by a majority.For most intereststo gain policy benefits,representatives must agree to exchange support. Put anwithdifferent constituents other way, the diversityof interestscreates gains from exchange assumes that within the legislature. While the literature implicitly these gains are captured, it fails to explain how trades are accomplished and enforced. If public policy reflectsa series of bargains among various interests,how are these bargains maintained over time? As we know from the modern literatureon contracts,the ansince not all agreeswer to thisquestion is not always straightforward mentsare enforceable.
' Typical applications focus on the various formsof verticalrelations(Coase 1937; Williamson 1975, 1985; Klein, Crawford,and Alchian 1978). Besides these more genof other formsof' there are excellenttreatments of verticalintegration, eral treatments (Rubin 1978), resale price maintenance (Gilligan verticalrelationssuch as franchising 1986), and long-termcontracting(Joskow 1985). 2 The exceptions include Goldberg (1976), Moe (1984), Weingast (1984), Miller and Moe (1986), Tirole (1986), Milgrom and Roberts (1987), and some of the topics in North (1981). The program forwide application of the approach is discussed in Jensen (1983). Fama and Jensen (1983) extend the analysisof marketorganizationsto include some nonprofitones, though their analysis only begins the study of this important organizations. categoryof widelydifferent
134
JOURNAL OF POLITICAL
ECONOMY
To address these issues, we develop a theoryof legislativeinstitutions that parallels the theoryof the firmand the theoryof contractual institutions.Like market institutions,legislative institutions reflect two keycomponents: the goals and preferencesof individuals, and the here legislators seeking reelection from their constituents, opporcosts that are induced by imperfectinformation, transactions tunism, and other agency problems. But the enforcementmechanisms underpinning exchange in market settingsare typicallyunavailable to or inappropriate for the legislature. Solutions to contractualproblems that arise in the market (e.g., verticalintegration) do not directlytranslateinto solutions to similar problems found in enforcebargains legislatures.We show how the legislativeinstitutions problems bargaining the peculiar why, given among legislatorsand found in legislatures, specific nonmarket exchange mechanisms prove superior to marketexchange. From a policy perspective,these have importantimplications.Durabilityof bargains leads institutions both to the durabilityof policies that these bargains are designed to implementand to the coalition supportingthese policies. Our model thus has importantimplicationsfor coalition formationand maintenance. Section I summarizesthe new economics of organization.Section II begins the analysis by presentingseveral assumptions on which our approach is based. Section II describesmodels of the marketforvotes and focuses on enforcementproblems. Section IV presentsour thesolve and suggestswhythese institutions ory of legislativeinstitutions problems that arise in simple markets.Section V provides empirical evidence on several propositions that follow from our model. This evidence, from a varietyof contexts involving the U.S. Congress, support for the model. Section VI derives some provides significant comparative staticresults that provide some additional evidence for the approach and suggest some important avenues for additional tests.A discussion section, Section VII, followsin which we explore alternative explanations forenforcinglegislativeexchange along with possible extensions of our approach. I. The New Economics of Organization
holds thatproductionand exchange take place The theoryof the firm (contractualpatterns,organizationalforms)that throughinstitutions reflectthe specificpattern of transactioncosts found in trade. The emphasis of this theoryis on how specificorganizationalor contracresultsfromthis tual formsreduce these costs. Some of the important literaturewill prove useful in our discussion of legislatures. The seminal paper in this tradition(Coase 1937) asserts that the
INDUSTRIAL
ORGANIZATION
135
firm emerges not simplyto take advantage of specializationor economies of scale but to avoid the costs of using marketsand the price system:"The main reason whyit is profitable to establisha firm would seem to be thatthereis a cost of using the price mechanism.The most obvious cost of 'organising'production throughthe price mechanism is that of discoveringwhat the relevantprices are" (p. 390). In other words, the firmprovides a set of contractualmechanismsthatsubstitutesforthe price mechanism,in part because the price mechanismis too costlyto use in certain circumstances.3 A major theme in the literatureis that the institutions of the firm are designed, in part, to reduce the costs of assuring contractualperformance. In the words of Williamson (1985, pp. 48-49), "Transactionsthatare subject to ex post opportunismwillbenefitif appropriate safeguards can be devised ex ante. Rather than replyto opportunism in kind, therefore,the wise [bargainingparty]is one who seeks both to give and receive 'credible commitments.'Incentives may be realigned, and/orsuperior governance structures withinwhichto organize transactions may be devised." This principleis one of the central lessons of this body of work; it underlies much of institutional and organizational design.4 The costsof assuringcontractualperformanceare high in a variety of circumstances. Two settingsconcern us. The first centerson problems of observability(Holmstrom 1979) or measurement (Barzel 1982), for example, when it is difficult to separate out an agent's contributionfrom that of random events or when an agent has private informationabout, say, the quality of the good being sold. Imperfectobservabilitygenerates well-knownproblems such as moral hazard, adverse selection,and shirkingthat plague simple spot marketexchange. A large part of the literature spells out ex ante contractual formsdesigned to mitigatethese problems. The second setting centers on incomplete contracts,for example, when it is impossible (or too costly)for contractingparties to plan for all possible contingencies. Several scholars have studied these settings and the attendant of problemsof ex post opportunismthatarise when ex post incentives the bargaining parties are inconsistentwith performing ex ante agreements (e.g., Klein et al. 1978; Kreps 1984; Williamson 1985; of mechaGrossmanand Hart 1986). Those worksalso studya variety nismsthatare used to mitigatethese problems,typically some formof verticalrelations.
3See also the discussion in Cheung (1983). 4Virtually every paper cited on the theoryof the firmmakes this argument. For
particulardetails, see, e.g., Barzel (1982), Fama and Jensen (1983), Kreps (1984), or Williamson(1985).
136
ECONOMY
We emphasize that the literatureis not simplyan analysis of contractualfailures.As suggested by Williamson in the quote above, ex post problems lead to the design of organizational formsto mitigate these problems. The literatureon verticalintegration,for example, argues that this organizational form is largely an endogenous response to ex post contractualproblems of the sort we have just mentioned. This example illustratesthe argument that a particularform of internalorganization proves superior to marketexchange. A major limitationof the new economics of organization is that it remains largely tied to market settings.Though the principles are obviously more general (as clearly articulated in Jensen [1983] or Milgrom and Roberts [1987]), applications to other settingsare just beginning. Indeed, developing a general theoryof organizationsreapplying this theory to types of organizations bequires effectively yond those included in the set studied to generate it. II. Representatives and Their Constituencies
In thispaper, we take up thischallenge by showinghow thisapproach illuminatesphenomena that take place in legislatures.The perspective developed in this paper restson three assumptions. interests responsive) 1. Congressmen represent the(politically ASSUMPTION locatedwithintheirdistrict.-While rational ignorance pervades the of constituents politicalsystem, that does not implythat the interests or that the latterare free to pursue are irrelevantfor representatives their own interests. Rather, rational ignorance underpins interest group advantage in politics. Because most voters have only a dim awareness of an incumbent'sactions, rational ignorance biases political response toward those who do form impressions.Thus interest groups, because they have greater individual stakes in particular issues, monitor congressmen and provide them with information. congressGroups also mobilize theirmembers in support of friendly men. have distributed.They typically Interestgroups are not uniformly concentrationsof voters in particular locations. Farm organization members,for example, are concentrated in specificdistricts;so too are consumersof food stamps and membersof welfarerightsorganizations. The elderly,to take another example, have a disproportionate presence in Florida and Arizona (medicare and social security) while minersare found in West Virginia,Pennsylvania,and southern Illinois (mine safety,black lung disease). In the competitionfor interestgroup support, specificrepresentativeshave a comparative advantage. The lack of complete fungibility of votes implies that legislatorsare advantaged in attracting support
INDUSTRIAL
ORGANIZATION
137
(see Denzau and Munger frominterestgroups located in theirdistrict attracts 1986). This advantage arises because serviceto local interests representative. both votes and organized resources for the district's attracts onlythe latter Serviceto thisgroup by an outsider,in contrast, and may lose votes. Electoral competitioninduces congressmen,at least in part,to repBecause groups are not uniof theirconstituents. resentthe interests formlydistributedacross constituencies,differentlegislatorsrepregroups.5 sent different ofindividual on thebehavior 2. Parties place no constraints ASSUMPTION were strong around the turn of the century representatives.-Parties and sanctionmechanismsto conwhen theypossessed reward systems Specifically, partyorganizationsdetertrolthe behavior of members. mined entryintocompetitionforthe local seat, the positionsof power (e.g., of legislativebenefits withinthe legislature,and the distribution a representativeobtained legislative benefitsonly if he supported partymeasures). None of these conditionsnow holds. In whatfollows, we thereforetreat the individual as the decision-makingunit.6 constraint.-Proposedbills 3. Majority rule is a binding ASSUMPTION (alterationsin the statusquo) mustcommand the support of a majorityof the entire legislaturein order to become law. III. The Gains from Exchange: The Problem to Be Solved
Legislators pursue their reelection goals by attemptingto provide benefitsto theirconstituents(assumption 1). Acting alone, theycanof not succeed (assumption 3). This, in combinationwiththe diversity interests theyrepresent,generates gains fromexchange and cooperation among legislators. But what institutionsunderlie-and enforce-this cooperation?
5 Evidence for this view abounds in the literature.For a recent summary in the systematic political science literature,see Fiorina (1981b). In the economics literature, over ideological votingin Conevidence has been provided as part of the controversy behavior gress. While the empirical issue concerns the degree to which representative evidence thatthe all studies provide substantial interests, can diverge fromconstituents' lattersystematically-thoughnot necessarilycompletely-affects congressionalvoting (see Kau and Rubin 1979; Kalt and Zupan 1984; Peltzman 1984). 6 Substantialevidence for this assumption is provided in the politicalscience literaonce a tool of the ture (see, e.g., Mayhew 1966). To take one example: the whip system, leadership to keep partymembers in line, now operates as a serviceorganizationprovidinginformation to the leadership and to the members.To quote one popular texton Congress,it "operates not as much as a device to coerce or even persuade membersas it does simplyto informthe leadership of the dispositionof memberstowardlegislation" (Polsby 1984, p. 129).
138
JOURNAL
OF POLITICAL
ECONOMY
The new economics of organizationsuggeststhatinstitutions evolve to ensure deliveryof benefits.In order to understand why one exchange mechanismsurvivesinstead of another,we need to studythe potentialagency and transactionscost problems faced by legislators, given the types of trades they seek to make. It is useful to begin by focusingon previous approaches to legislativeexchange that explicitly rely on marketlike mechanisms. By studying the enforcement problemsencountered in thissetting, we can determinethe characteristicsa more appropriate legislativeexchange mechanism must possess. Previous workhas focused on votetrading, also knownas logrolling, centralizedlegislativeexchange, or legislativeIOUs. The major proponents of particularversions include Tullock (1967, 1981), Wilson (1969), Telser (1980), Koford (1982), and Becker (1983). While there are significant differencesamong these approaches, fundamentalto each is an explicit or implicitmarketin votes. Under the most wellknown logrollingversion, legislatorsbegin with proposals to benefit themselves at the expense of others, but none of these proposals commands a majority(Buchanan and Tullock 1962; Tullock 1967, 1981). Legislatorsthereforesearch out tradingpartners.In exchange forsupport,each gets his proposal passed and, on net,is betteroff.In the explicit market versions, votes are bought and sold for a price, withthe "equilibrium" prices determiningvote trades and hence the set of bills passed (see also Wilson 1969; Koford 1982). The motivationunderlyingthese marketmodels is clear. By giving away votes on issues thathave lower marginalimpacton theirdistrict (and thereforeon their electoral fortunes)in exchange for votes on issues having a larger marginal impact, legislators are better off. Whether or not they incorporate an explicit auction, models of the legislativemarketfor votes have considerable appeal. A careful inspection,however, reveals that this approach assumes away some of the deepest problems plaguing legislativeexchange. It assumes, for example, that all bills and their payoffsare known in advance; that is, there are no random or unforeseen future events thatmay influenceoutcomes or payoffs.Either the timedimensionis suppressed or enforcementof agreements over time is left exogenous. Because these models study a legislaturewith no future,they cannot address how legislatorscope withagreementsthatcover more than one legislativesession. A varietyof exchange problems arise because the value of today's legislation significantly depends on next year's legislative events. fromthose faced Membersof futuresessions face incentivesdifferent when the trade occurred and may seek, forexample, to amend, abolish,or simplyignore previous agreements.Because currentlegislators
INDUSTRIAL
ORGANIZATION
139
cannot bind a futurelegislativesession, problemsof enforcetypically importantforunderstandinglegislatures mentover timeare critically Moreover, as we will see, these settings away. assumed be and cannot inhibit the ability of noninstitutionalenforcement of cooperation (e.g., reputation)as the sole means of policingbargains. In the face of legisover the futurestatusof today's bargain, therefore, uncertainty for long-termdurabilityof agreements lators will devise institutions beyond thissession of the legislature. thatensure the flowof benefits To begin our analysis,we observe that most models of the legislative market apply to only a subset of problems faced by legislators, the pork barrel. Pork barrel programsare an importantpart typically but theyhave special characterof everymajor Westerngovernment, istics that do not hold for other types of legislation. For example, benefit flows are contemporaneous to differentlegislators (in this the project),and consummationof tradingis case, the fundsfinancing simultaneous (see, e.g., Buchanan and Tullock 1962; Tullock 1981; Koford 1982). Focusing solely on pork barrel-type programs rules all the importantissues studied in the regulatoryliteraout virtually programs.7We consider ture as well as the major U.S. redistributive the problems generated by noncontemporaneous benefitflowsand in turn. nonsimultaneity
Flows Benefit A. Noncontemporaneous patterns of benefitflows potentiallyinhibit To see how differential trading,consider the followingexchange problem. Suppose that a group of legislators seeking pork, for example, dams and bridges, attemptsto find some other group of legislatorswith whom to exchange votes. Suppose furtherthat one potentialset of tradingpartners is a group of legislators who seek a flow of services from a regulatoryagency. If the two sides exchange votes, the firstgroup obtains its dams and bridges while the second obtains its regulatory group agency. Once the dams are built,however,what stops the first fromrenegingon the agreement,for example, fromworkingduring a futurelegislativesession to revoke the regulatorybenefits?Simple do not adequately protectagainst this marketexchange institutions form of reneging (and, as we will see, repeated interactionalone is to prevent this problem). Rational coalition partners, insufficient discount the potentialgains froma proposed trade by the therefore, thatthese benefitflowswillbe curtailedby reneging.Conprobability sequently,the second group of legislatorsmightnot accept the trade
7 For several surveysin this literature, see the articlesin Fromm (1981).
140
ECONOMY
(e.g., if the trade yields them positive net benefitsonly if reneging does not occur). B. Exchange Nonsimultaneous
A second exchange problem arises because many potential trades concern bills that do not come up for a vote simultaneously.In the pork barrel area, legislatorsare able to limitthis problem by packaging all projects into an "omnibus" bill containingall elements of the trade. This simple device limitsthe opportunitiesfor ex post reneging.8 But it is not always possible for all bills in a legislativesession to come up for a vote simultaneously.9 Consider a trade negotiation taking place just prior to a vote. In exchange for a vote, some legislator promises to support another legislator's bill thatis due to come up sometimelater in the session. In other words, he extends an IOU to the second party.But problems withIOUs occur in part because theyare not a medium of exchange. They require that one individual rely on the futurebehavior of another. Were votes a medium of exchange, thisreliance would not be
necessary. 1()
Consequently,exchanges relyingon IOUs are plagued by the two and of problemsnoted in Section I, namely,problemsof observability the existence of contingenciestoo numerous (or too costly)to anticipate fully.Many eventsmay occur between the twovotes. First,public perception of the issue may change, and the electoral effectof this change is observable solely to the representativeit affects.This inlegislatormay claim that duces a formof moral hazard. Thus the first he can no longer support the bill and so attemptto renege. Since the for state of the world is observed only by one legislator,it is difficult the second legislator to verifythe first'sclaims about whether he should be required to hold up his end of the bargain. Second, in the billitself mayevolve. response to changing politicalcircumstances, This introducesa double-sided formof moral hazard. Since the elec8 Because the omnibus mechanism for ensuring against reneging is more readily available for bargains between members of the same committee(e.g., across subcomjurisdictionsdepends on the expected patmittees),the optimal patternof committee discussionof thisand similarissues. ternof trading.See Ferejohn (1986) fora further 9 Nonetheless,when the volume of legislationwas sufficiently low to allow all bills to be passed in a short period, legislaturesin factdid so. Thus, for the U.S. Congress in the nineteenthcentury,it was common that a major portion of legislationwas passed during the lame-duck session afterthe election of the next congress.This also appears to hold today for states whose legislaturesmeet for only short periods. by this 10 As a consequence, the so-called double coincidence of wantsis not satisfied transaction.More generally, IOUs have none of the properties of' a medium of' exchange: a store of value, a unit of account, and ready transferability.
INDUSTRIAL
ORGANIZATION
141
toral effectsof this change are observable solely by the legislatorit affects, the first legislatormay argue that,while he could support the originalbill, he cannot support the new version. On the other hand, the drafters of the legislation, having gained additional support throughtrades, may opportunistically rewritethe legislationso as to increase theirown benefits(and impose greater costs on others). Trading in legislative IOUs thus poses considerable contractual problems of the sort studied in the theoryof the firm.Either IOUs must be for a specificform of a bill withoutany alterationsor they must provide for hundreds of contingencies,many of which are not observable to both parties. Neither form of IOU is likelyto prove useful. The formerseverelylimitsthe tradingpossibilities. Since most legislationis altered at several stages before it is passed, this formof IOU exchanges one vote for sure against one vote under relatively rare circumstances-an unlikely basis for a transaction.11 Further, different contingenciesare importantto different legislators, and the market for specific,contingentIOUs is likelyto be extremelythin, perhaps requiringa different price foreach potentialtrade. As Coase (1937) observed, thisobviates the benefitsof a price system.But perhaps more important, the observabilityproblems associated with many contingenciessuggest that IOUs are unenforceable: how are the parties to agree ex post when the number of possible events is larger than the number of specified contingenciesand when both parties cannot observe the outcome? This discussion reveals that marketformsof exchange are limited as a means of capturing the gains from trade. As noted in Section I, problemswithobservability and ex post enforceability are fundamental to understanding the motivationfor internalizinga transaction witha firm. Justas these problems lead to the emergence of vertical integrationto replace marketexchange, they motivatethe design of institutions withinthe legislature that substitutefor explicit market exchange. In the discussion so far,there has been littlementionof the role of repeat play. Repeated interactionprovides incentivesfor individuals to adhere to agreementsthis period so as to maintaina flowof benefitsover time.'2 This formof endogenous cooperation surelyplays a
" See Ferejohn (1974b) for a furtherexploration of' the peculiar properties of a marketin votes. This stems in part fromresultsin the collectivechoice literaturethat show that when one set of vote trades is feasible, so are many others (e.g., Schwartz 1981). This preventsthe logic of the standard argumentsabout supportingprice systems fromholding in this context. 12 See, e.g., Axelrod (1984) and Calvert (1985). There is, of course, a growingliterature in economics on this topic (e.g., Telser 1980; Klein and Leffler1981; Kreps and Wilson 1982; Roberts 1986). A furtherproblem limitsthe workability of this solution, thatof legislativeturnover.Even in currenttimeswhen incumbentsare reelected with
142
JOURNAL
OF POLITICAL
ECONOMY
to it alone may be sufficient and forsome settings, role in legislatures, police bargains. It is well known, however,that "the long arm of the future"is inadequate in settingsin which agents have private informationand in which it is impossibleor too costlyto specifyall contingencies in advance.13 It is preciselythese problems that we have arThe gued motivate the need for alternativelegislativeinstitutions. both noncontemin contingencies of unanticipated importance poraneous and nonsimultaneous trading combined with private informationand moral hazard in the lattersuggeststhe need for additional mechanismsto maintain bargains. Perhaps another way of puttingthe argument of this section is as to preventthe breakdownof follows.Repeat play alone is insufficient cooperation under certain circumstances.Legislators thereforehave that reduce the circumstancesin an incentiveto devise institutions whichbreakdown occurs. In thissense, legislativerules are not substicommonlyused in tutesforreputationbuilding and triggerstrategies and, repeat play. Rather,rules complementthe use of these strategies in particular,prevent the breakdown of cooperation at preciselythe circumstancesunder which these other strategiesfail. in which This argumentcloselyparallels thatof verticalintegration to police cooperation between reputationeffectsare also insufficient firms. In both cases, potentialcontractualproblemslead to the design that substitutefor marketexchange; in so doing, they of institutions of agreements. This does not imply improve ex post enforceability that thatreputationbuilding is unimportantin legislaturesor in firms are vertically tied,just thatit is not the solemeans of enforcingagreeof the legislatureundoubtedly ments. Indeed, the other institutions facilitate its use as a means to complement other devices. C. Implications of bargains are and enforceability Problemsconcerningthe durability limitingthe value of explicitmarket ubiquitous in legislativesettings, under formsof exchange.14 Put another way,coalitionslack durability
high frequency,the average net turnoverin Congress is 10 percent per term. Morepreferencesif only replaced withmemberswithdifferent over, the losers are typically because the latter,in order to beat the former,had to devise a separate support constituency. on the theoryof the firmis builton the premise thatthe incentives 13 The literature to police incentiveproblems.Examderived fromrepeat dealings alone are insufficient ples are the verticalintegrationor the optimal structureof' financialclaims. See the referencesin n. 2. do 14 Moreover, the problem of non-pork barrel programs and lack of simultaneity not exhaust the situationsin whicha legislativemarketis a poor providerof durability. benefits, For example, even iftwo groups of legislatorsboth seek permanentregulatory
INDUSTRIAL
OR(GANIZATION
143
an explicit market exchange system.In the face of these problems, legislatorswill devise alternativeinstitutions that provide exchanges witha greaterdegree of durability(see Ferejohn 1986). We now turn to a discussion of how this is accomplished. IV. The Legislative Committee System
This section develops a model of an idealized legislativecommittee system.The types of policies (i.e., legislativebargains) that emerge fromthis model parallel those predicted by the vote-trading models; but it is not plagued by problems of enforcementof exchanges. The legislativecommittee systemis defined by the followingthree conditions. 1. Committees are composed of a number of seats or CONDITION positions,each held by an individual legislator.Committeespossess the following properties: (a) associated with each committee is a specific subset of policy issues over which it has jurisdiction (e.g., commerce,energy,banking,or agriculture); (b) withintheirjurisdiction, committeespossess the monopoly rightto bring alternativesto the statusquo up for a vote before the legislature;and (c) committee proposals mustcommand a majority of votes against the statusquo to become public policy. CONDITION 2. There existsa propertyrightssystem over committee seats called the "seniority system."It has the following characteristics: (a) a committeemember holds his position as long as he chooses to remain on the committee; subject to his reelection, he cannot be forced to give it up; (b) leadership positions withinthe committee (e.g., chairmanship) are allocated by seniority, that is, the length of continuousserviceon the committee;(c) rightsto committeepositions cannot be sold or traded to others. 3. Whenever a member leaves a committee (e.g., by CONDITION transfer, death, or defeat),his seat becomes vacant.There is a bidding mechanismwherebyvacant seats are assigned to other congressmen. Condition 1 definesthe source of committeepower and value, condition 2 definesthe propertyrightssystemassociated withcommittee positions,and condition 3 establishes an exchange mechanism over the rightsestablished under 1 and 2. Let us explore the consequences of the legislative committeesystem to determine its enforcementproperties,how new policies are prochanging electoral fortunesmay promote growthin one and shrinkthe other; to the extentthat thischange appears reasonably permanent,it provides the conditionsfostering a revocation of the latter group's benefits.When the once and for all gains exceed the cost potentially imposed bythe (now smaller)other side, renegingis likelyto occur.
144
ECONOMY
vided, its controlof the agency problems that arise fromthe delegation of power to a particular subset of members, and the types of policies that are likelyto emerge fromit. Bargains ofLegislative A. Enforcement The committeesystemprovides substantialprotectionagainstopportunistic behavior, therebyprovidingdurabilityto policybargains. To see this,consider the settingdescribed above in which one group of legislatorsseeks dams and bridges and the second seeks a regulatory market, thisagreeIn the legislative itsconstituents. agencybenefiting mentis vulnerable to ex post renegingof the followingform:the first group, after building its dams, might form a coalition with other excluded fromthe original deal) to legislators(perhaps the minority the second group. pass a new bill revokingthe regulationbenefiting But now consider the same bargain assuming that it was forged under the committeesystemand that the firstgroup controlled the committeewithjurisdiction over pork barrel programs,the second, the committee with the jurisdiction over the relevant regulations. Under the committeesystem,the second group retainscontrolover the agenda withinitsjurisdiction. Suppose that,once the dams and group introduceslegislationto revoke bridgesare completed,the first the benefitsflowingto the second group, and, further,a majority supports this legislation.However, only the committeewithjurisdiction can bring it to the floorfor a vote. This controlover the agenda withinitsjurisdiction implies that a committeehas veto power over the proposals of others. Since this proposal would make the committee worse off (and since, by assumption,a majoritywill support it on the floor),the committeewould not allow it to come up for a vote. In access to the agenda servesas a mechanism otherwords,the restricted to preventex post reneging. Moreover, because exchanges in influence are institutionalized is through the property rights system,the absence of simultaneity is system considerablyless troublesome.As long as the propertyrights for maintained,the agenda power held by each committeesubstitutes outstandingIOUs withuncertaincontingencies.The problemsassociated withdevisingcontingentclaims over futureeventsare relatively absent under the legislativecommitteesystem. B. Capture (or How Committees Providing New Benefits theGainsfrom Exchange)
The agenda rightsaffordcommitteemembersconsiderable influence over policy choice withintheirjurisdiction. This followsbecause the
INDUSTRIAL
ORGANIZATION
145
set of points that command a majorityagainst any given status quo, W(sq), is generally quite large (McKelvey 1976, 1979; Shepsle and Weingast 198 1). Typically,W(sq) includes a wide range of policyalternatives,some making committeemembers worse offand some makagenda power ing them better off. Given this range of alternatives, they allows committeesto bias the outcome in favorof the alternative most prefer.'5 a trade among all the legisThe committeesysteminstitutionalizes lators,policy area by policy area, for the rightto select which points from W(sq) replace the status quo. But this is neither accomplished nor enforced by an explicitmarketexchange. Rather,a legislatoron committeei gives up influenceover the selection of proposals in the j's giving area of committee in exchange for membersof committee up their rights to influence proposals in area i. Institutionalizing rightsover agenda power-that is, controlover the design and selection of proposals thatarise for a vote-substitutes for purchasingthe votes of others in an explicit market.Since any element of W(sq) will it is the influenceover elementsof thisset afforded pass by definition, committeesby agenda power that eliminates the need for explicit exchange of votes. C. WhoGains Influence (or How Are theGainsfrom ExchangeDistributed)?
This question concerns the types of policies chosen under the committeesystem.Since committeesaffordtheirmembersdisproportionate influenceover policy choice withintheirjurisdiction,it also concerns the mechanism that assigns legislatorsto committees. Condition 3 provides thatthe legislatureuses a bidding mechanism to assign members to committee positions. Since a representative's electoral fortunesdepend on his obtaining benefitsfor his constituents and since constituent interests differ, legislatorsseek assignment to those committeesthathave the greatestmarginalimpactover their costs of bidding forcommitelectoralfortunes.The real opportunity of holding a tee i are that the representativegives up the possibility seat on committeej. Thus representativesfrom farm districtsare than they much more likelyto bid forseats on agriculturecommittees are for seats on urban, housing, or merchantmarine committees.A potential problem arises, however, because some committees are valued by all (e.g., the spending or taxingcommittees).However, here too the bidding mechanism determinesassignment.The more com'5 The details of this process are beyond the scope of this paper. For an in-depth analysis,see Shepsle and Weingast (1984, 1987).
146
JOURNAL
OF POLITICAL
ECONOMY
petitionfor seats, the less likelythe bid will be successful.Suppose each potentialbidder for a highlyvalued committee(e.g., one concerning taxes) also values some specificpolicycommitteewithmuch less competition (e.g., housing, agriculture,or public works). The increased competitionfor seats on the tax committeesimplies that only those with the greatestdifferential value between the tax committeeand theirnext-best alternative will pay the opportunity cost of bidding (i.e., giving up a higher probabilityof gettingtheir policy committee). D. Implications for Coalition Formation
The legislative committeesystem has two separate effects on coalition formation.First,agenda power held by committeemembersimplies that successfulcoalitions must include the membersof the relevant committee.Withoutthese members,the bill will not reach the floor for a vote. This, in turn,implies that certainpolicies are unlikelyto become law, forexample, those thatprovide benefits onlyto a majorityoffthe committee.In technicalterms,committee veto impliesthat, fromamong the set of policies that command a majority against the statusquo, onlythose thatmake the committee betteroffare possible (this issue is extensivelyexplored in Shepsle and Weingast [1987]). This significantly reduces the feasibleset of policies that may be implemented. have rightsto Along these lines, we also note thatsince committees bringa single bill to the floor,trades among committeemembersare more likely to succeed than those across committees.This follows because there is less chance for such a deal to fall apart. When a must coalitionformsbetween membersof twocommittees, legislators agree to exchange voteson two separate bills.When a coalitionforms among membersof the same committee, theymay bringa single bill to the floor.The latterallows a singleup or down vote on the package (whereas the formerdoes not), therebyaffording less chance for reneging. This suggeststhat drawing thejurisdictionalboundaries between committeesis an importantstrategicvariable that affectsthe patternof coalitions.'6 Ceterisparibus,expected tradingpartnersare better off if they are members of the same committeeso that the
16 See Ferejohn (1986) fora discussionof thisissue in the context of a trade between the urban members on the AgricultureCommittee (seeking food stamps) and the farmers on thiscommittee(seekingcontinued farmbenefits). He argued thatbeing on the same committeeadvantaged these urban membersover other potentiallegislative partnerswho were part of othercommittees thatmighthave broughtsome otherform of legislationprovidingsome subsidyforfood forthe poor (the lattercould have easily been written by, e.g., Ways and Means).
INDUSTRIAL
ORGANIZATION
147
optimal patternofjurisdictionsmust in part reflect the expected pattern of trades. The second effecton coalitionsconcerns durability. The durability afforded by the committee systeminduces some rigiditiesinto the coalition formationprocess. Under a market exchange mechanism, small changes in politicalcircumstanceswould lead to a small change in the optimal set of bargains and coalitions.But under the committee system,small changes in circumstancesdo not automaticallylead to changes in policy. To see this,consider the example explored above involvingdams, bridges, and regulatorybenefits.We showed above that committeeveto power prevents the proponents of dams from easilyrenegingonce theirdams are built or if,because of a change in politicalcircumstances, theyfind a more attractive coalition partner. This does not mean, however,thatthe dam-and-bridgeslegislators can never alter policy.Rather,it means thattheymustbid forseats on the committeeand wait until theyattain a majority.Small changes in political circumstancesare not likelyto make it worth the attempt. Therefore,the committeesystem impliesthatpolicywillrespond only to large changes in political circumstancesor to major shiftsin the 17 electorate. E. Controls overCommittees
Committees are decentralized decision-making units composed of those legislators with the greatest stake in theirjurisdiction. Their power to decide what proposals (ifany) are broughtto the floorplaces themin an agency relationwiththe restof the legislature.As withany form of delegation, this authorityprovides the potential for moral hazard. What preventsthe committeefromextracting too much surplus at the expense of other legislators? The committee systemconstrains the behavior of its subunits by restricting committeepower. In particular,the majorityrule condition precludes any one committeefromextractingtoo many gains at the expense of others. Suppose, for example, that one committee attemptsto extractthe entirebudget. The majorityrule requirement impliesthatthisproposal must get a majority of legislatorsto give up the opportunity to spend some of the budget in theirareas. They will do so only if the value of the last dollar fromthis proposal to them exceeds the value of the firstdollar spent withintheir own jurisdic17 We note that this phenomenon parallels vertical integration.There, long-term agreementsalso induce durabilityand rigidities:the contractis not renegotiatedwith each small change in economic circumstances (e.g., prices) and thereforedoes not respond to changes in the way a spot market(toes.
148
ECONOMY
tion. Since members value influence withintheir own jurisdictions, this situation is unlikely. Thus the voting rule plays an important constrainingrole over the opportunisticbehavior of particularcommittees. 18 F. Summary
Instead of trading votes, legislators in the committee system institutionalize an exchange of influenceover the relevant rights.Instead of bidding for votes, legislators bid for seats on committees associated withrightsto policy areas valuable for theirreelection.In contrastto policychoice under a marketforvotes,legislative bargains institutionalized through the committeesystemare significantly less plagued by problems of ex post enforceability. V. Evidence: The Distribution of Preference, Influence, and the Benefitsof Committees
In what follows,we provide evidence showing that choices and decision making in the U.S. Congress are consistentwith our view.19 (Thus thisis not a directtestbetween our model and the vote-trading approach.) The major feature of our model is that exchange takes place via institutionalization throughthe committees. By farthe strongest piece of evidence from the U.S. Congress in favor of our approach concerns the pattern of membership and benefitflows for the various committees(Fiorina 198 la). Members from farmingdistricts dominate the agriculture committeesand oversee programs that benefit farmers.Members from urban districts sit on banking,housing, and welfare committeesthat provide benefitsto an incredible array of urban constituents.Members with large defense installationsor industries dominate Armed Services committees.In each case, members mold policies in theirjurisdiction to their constituents'advantage. The model is based on a set of assertionsabout committeeoperation: (a) the assignment process operates as a self-selection mechanism; (b) committeesare not representativeof the entire legislature but instead are composed of "preferenceoutliers,"or those who value
18 In most legislatures,the amendment process places additional constraints on the behaviorof committees.For details of thisprocess forthe U.S. Congress,includinghow it qualifies this argument,see Shepsle and Weingast (1987). The problem of how this body places constraintson committeeshas never received systematic treatment. 19 Congress, unlike the BritishParliament,meets the conditionsset out in Sec. II. We briefly compare our findings forthe American case withthose of the Britishin Sec. VII.
INDUSTRIAL
ORGANIZATION
149
committeemembers the position most highly;and (c) most centrally, receive the disproportionate share of the benefitsfrom programs withintheirjurisdiction. Let us survey the empirical evidence supportingthese propositions. A. Committee Assignments At the beginningof each new congress,there are a number of vacant committeeseats in some 25 committeesand thereare incomingfreshmen without seats.20 They are encouraged to request only a small number of possible positions. Then party leaders attemptto match individual assignmentswith their freshmanrequests. There is, howfrombreakever,a potentialproblem here: What preventsthe system ing down because everyone requests seats on the best and most powerful committees?How does the bidding mechanism actually select those freshmenwillingto bid the most for particularcommittees? The mechanics of the assignment process are designed to work against breakdown. It turns out that there are certain committees (e.g., Post Office)thatno one wants.Those who failto get one of their requested slots are generally put on one of these committees.Requesting the most valuable slots,therefore,increases the probability of ending up with Post Office. Suppose each freshmanmay potentiallyrequest a particularsubstantivepolicycommittee(e.g., Agriculture, Housing and Welfare,or Public Works) valuable for his district of getting.Which ones will opt instead thathe has a high probability to request the more powerful committees?Since the latter option and one worth between the mostvaluable committee involvesa lottery virtuallynothing, only those freshmenwho value it most highlyin comparison with the sure thingof gettingon theirpolicy committee willbid for it.21This lottery implies that revealed preferencesreflect true preferencesand shows how the assignmentmechanismsucceeds
2() The following descriptionrelies on Shepsle (1975, 1978). While he did not discuss the preferencerevelationaspects of the assignmentprocess, it is clear thatthe process requests. Since few empirical contexts must relyon some means of inducing truthful that make use of these mechanisms have been studied, his data remain an untapped returningmembers source for furtherstudy.In what follows,we ignore for simplicity who wish to change committees.For details on how this works,see Shepsle (1978). 21 The following over the lengthsof request table reportsthe frequencydistribution lists(i.e., how many committeeseach freshmanrequests). Three-quartersof all freshmen (87th-93d Congress) ask for three or fewerout of 25. The number of observations is 231 (source: Shepsle 1978, p. 49).
Length Percentage
1 23
2 16
3 36
4 15
5 or More 10
150 TABLE
FRESHMAN
JOURNAL 1
SUCCESS
OF
POLITICAL
ECONOMY
AssIGNMENTr
PROPORTI First CONGRESS 87th 88th 89th 90th 92d 93d All SOURCE.-Shepsie (1978, Preference .474 .500 .591 .308 .750 .691 .585 p. 193).
ON Other
RECEIVING
No
Preference .159 .194 .155 .384 .106 .193 .172
Preference .368 .306 .254 .308 .144 .166 .243
N
19 36 71 13 28 26 193
in matchingmemberswithcommitteeswhosejurisdictionstheyvalue most highly. The evidence supportingthis interpretation is twofold.First,table 1 shows that the probabilityof a freshman'sgaining one of his top three is above .8.22 Second, and more important,table 2 shows that when there is no competition for a seat, the requester is virtually assured of gettinghis first choice (the probability is over .94); but the greaterthe competition,the less likelyis a freshmanto attainhis first choice. There is also considerable evidence that freshmanrequests take into account competition for seats.23 Competition of this sort appears necessary-though not sufficient-to ensure thatbids reflect underlyingpreferences. Overall, then, the patternof committeeassignmentslooks remarkably like an optimizationprocess that maps membersinto those committeestheyvalue the most. B. Committee Membership
To be more systematic about committeemembership,we have examined indexes of member preferencesover issues that correspond to
22 Moreover, it is not clear that this frequencycan be much higher because of the manyaccountingconstraints (see Shepsle 1975) imposed on the problem (e.g., onlyone freshmanper slot; each vacant slot must be filled). 23 Shepsle (1978) provided one more piece of evidence for our model. Using probit analysisto predict which freshmanrequests particularcommitteeslots,he estimateda set of simple demand equations. His resultsare consistent withour model, namely,that simple measures of constituency interest(e.g., numberof agriculturalworkers, military employees,or housing) are good predictorsof requests. Moreover,these estimatesalso show that freshmenrationallyanticipate competitionfor different seats: when other factorsare held constant,the estimatedprobability of a freshman'srequestinga certain seat goes down as the number of competitorsincreases.
INDUSTRIAL
ORGANIZATION
151
TABLE 2
EFFECTS OF COMPETITION ON ASSIGNMENTS NUMBER OF EFFECTIVE PER VACANCY
TOTAL FIRS T PREFERENC E ASSIGNMENT SUCCESS
REQUESTS
Less than 1 94.4 5.6
1-2
More than 2
Yes No
SOURCE.-Shepsle (1978, p. 201).
67.2 32.8
30.5 69.5
major committee jurisdictions.This exercise reveals thatmembersof the relevantcommitteeor subcommitteesignificantly differfromthe restof the House.24 Most indexes are computed by an interestgroup witha clear stake in the policy area being considered. Because they are constructed so as to indicatewhichcongressmenare supportersof the group, these indexes are good proxies for supporters of the group's interests.The scores computed by the AFL-CIO Committee on Political Education (COPE), for example, indicate pro- and antilabor congressmen; the American SecurityCouncil's National SecurityIndex (NSI) reveals supportersof a strongnational defense and, apparently,opponents of foreignaid.2' The model predictsthatrepresentatives of particularinterests gain policy benefitsthrough membershipon relevantcommittees.Hence we should observe that committeesare composed of members who are significantly above-average supporters of the relevant interest group and, in particular, have interest group scores significantly above the mean for the entire Congress. This pattern is borne out by the results reported in table 3. The differencein preferencesbetween committeemembers and the rest of policy of the House is highlystatistically significant. For a diversity areas-defense, foreign aid, consumer protection,labor, and the aboveenvironment-committee members are indeed significantly average supportersof benefitsto the relevantinterestgroup. Puttingthis evidence togetherwithresultsfromcommitteeassignments reveals that legislators opt for committeesrelevant to their constituents'interestsand that their doing so leads to committees
24 theyhave Though this would seem to be an obvious topic for political scientists, provides typically never systematically collected thistypeof data. Instead the literature anecdotal evidence, the best of which can be found, e.g., in Jones (1962) or Fenno (1973). 25 Foreign aid to other nations,under the jurisdiction of the Foreign RelationsComspending programs.The evidence formilitary mittee, seems to be a (political)substitute suggeststhatthose congressmenwho support thisaid tend to be against defense spending, and vice versa.
152
ECONOMY
TABLE 3
COMMITTEE MEMBERS ARE PREFERENCE OUTLIERS RELATIVE TO THE FULL HOUSE (1978)
Full House
Meana
Committee Mean 76.8 50.2 46.5
Nb
t-Statistic 17.87**
1. Armed Services: NSI 2. InternationalRelations: NSI t-test for mean NSI difference between Armed Services and InternationalRelations International Relations:International Economic Policy and Trade Subcommittee: NSI ADA InterstateCommerce: Consumer Protectionand Finance Subcommittee: ADA Education and Labor: Economic Opportunity Subcommittee:COPE Environmentalsubcommittees:LCV'
ADA`
59.1 61.7 37.5
38 37 37
10.23**
11.42**
3.
19.40**
4.
60.8 38.1
51.3 45.0
7 7
4.24** 3.50**
5. 6.
37.9 50.4 46.7
55.5 60.0 58.3
8 4 28
9.57** 3.33** 2.08*
a All non-committee members. b Committeeor subcommitteesize. ' Vote ratingsof the Americans for Democratic Action. Includes two of the major subcommittees with oversight responsibilityfor the Environmental Protection Agency,the Subcommitteeon Energy and the Environment(Interior Committee),and Subcommitteeon Health and the Environment(Commerce Committee). LCV is the League of Conservation Voter scores for 1977. * Significant at the .05 level. ** Significant at the .01 level.
composed of legislatorswithconsiderablyhigher support for policies withintheir jurisdiction.This patternis preciselythatexpected by the view that committeesinstitutionalize trades over influence so as to give their members greater control over policies with theirjurisdiction. C. Committee PolicyBenefits
Do committeemembers receive a disproportionateshare of the benefitsfrom their committees?The evidence on preferences provides indirectsupport for this since committeesdisproportionately attract representativesseeking to provide their constituentswith benefits.
INDUSTRIAL
ORGANIZATION
153
Here we summarize some direct evidence in favor of this proposi26 tion. 1. Ferejohn (1974a) in his now-classic study on the pork barrel testeda varietyof hypothesesabout committees.He showed thatthe number of new projects startedin each state is a functionof committee membership.His estimationsimply,forexample, thateach member on the Public Works Committee yields an additional 0.63 new projectsfor his state. Further,each 10 years of serviceby representativesfroma state yields approximatelyan additional project. Similar resultsare obtained regarding more than two dozen related hypotheses. base closings,water 2. Arnold (1979) studied three areas (military and sewage grants,and model citiesgrants)and providesresultssimi27 His contingency lar to Ferejohn's about the pattern of benefits. tables provide unambiguous evidence; we reproduce two. Table 4, part A, shows the frequencyof acceptance of an application for a water and sewage grant, depending on a congressman's position in the committee system: is he a member of the relevant appropriations subcommittee?the relevant authorizationcommittee (Banking and Currency)?of neither?The table shows that members of the relevant committeessystematically fare betterthan nonmemof acceptance of bers. Those on neithercommitteehave a probability .176. In contrast, membersof the AppropriationsSubcommitteehave a probability of acceptance of .313 (80 percent larger),and members of acceptance of .281 of the authorizingcommitteehave a probability at the .001 level. (60 percent larger). The differencesare significant Part B of the table shows thatthe same patternholds formodel cities projectselection.For these projects,congressmenwho are on neither of selectionof .29. The probarelevantcommitteehave a probability of acceptance for members of the Banking and CurrencySubbility committee, .62, is more than double thatfornonmembers;the proba.86, is nearly bility formembersof the AppropriationsSubcommittee, triple. 3. Several recent studies by economistsused similarmethodologies defense expenand yielded similarevidence. Malone (1982), studying
26 Unfortunately, by far the biggesteffort to support thispropositionin the political science literaturecomprises anecdotal or descriptivematerial rather than systematic data analysis. While this literaturesupports our proposition, it is no substitutefor systematic empirical investigation. 27 We do not reproduce his probit estimates here (nor discuss his concerns about whether congressmen manipulate bureaucrats or bureaucrats manipulate congressmen). These estimatessufferfromsignificant econometricproblemsand are therefore of questionable value. Simultaneity,much like that found in estimatingsupply and demand equations, plagues his design.
154
JOURNAL
OF
POLITICAL
ECONOMY
TABLE 4
FREQUENCY OF ACCEPTANCE OF APPLICATIONS
Application Represented
Applications Accepted A.
Not Accepted
Total Decisions
Probability of Acceptance
Water and Sewage Grant Selection (1970)
Subcommitteeof Committee Banking and Currency Committee Neither committee Total

Appropriations
21 27 261 309 B.
46 69 1,223 1,338
67 96 1,484 1,647
.313 .281 .176
Model Cities Project Selection
Subcommitteeof Committee Banking and Currency Committee Neither committee Total

Appropriations
6 5 38 49
1 3 78 82
7 8 116 131
.86 .62 .29
SOURCE.-Arnold (1979, pp. 139, 180). NOTE.-For t. A, x2 = 13.80 and significancelevel is .001. For pt. B, x2
10.81 and significance level is .01.
ditures,showed that members of the Armed Services committeesreceive a statistically significant greatershare of federal expendituresin this category, though Rundquist (1973) could find none. Faith, Leavens, and Tollison (1982) studied the geographic locationof firms that are the target of antitrustsuits brought by the Federal Trade Commission (FTC). They showed that firmslocated in districts represented on the FTC oversight subcommitteeswere systematically underrepresentedin the set of suits broughtby the commission.Cohen and Noll (1986), using an innovativemethodology,derived similar resultsfor federal R & D projects. 4. Weingastand Moran (1983) studied the influenceof Congress on the distribution of cases chosen by the FTC under the various statutes it administers.They found, for the Senate, that all members possess some influencebut that members of the relevantsubcommitteepossess more influence and that the subcommitteechairman possesses even more influence (see table 5). According to their estimates for textilecases (under the Fur, Wool, and Textile Labeling acts),a member of the subcommitteehad nearly three times the effectof a nonmemberwhile the chairman had 12 timesthe effect of a nonmember. Their resultsreveal a similarpatternfor the other case typesstudied (credit cases, Robinson-Patmancases, and mergercases). 5. The patternof campaign donations by firms provides additional
INDUSTRIAL
ORGANIZATION
155 TABLE. 5
CHANGE
IN THE PROBABILITY
ADA
OF OPENING SCORE
A TEXTILE
CASE WHEN A SENATOR'S
INCREASES
10 POINTS
Senator's Position Not on the subcommittee On the subcommitteebut not chairman Subcommitteechairman
SOURC.E.-Weingast
Change in Probability .005 .013 .060
and Moran (1983).
evidence. A firm'sdecision to donate money to a congressionalcampaign must pass the same test as any other investment made by the firm;namely,the expected value of the returnmust exceed the dollars invested.When deciding among politicians,firmsmust focus on those congressmenwitha marginal impact on theirfutureprofitability. If committee members have a disproportionate influence over policychoice in theirarea, then theyshould attract a disproportionate share of campaign contributions fronifirmsaffectedby the committee's policyjurisdiction. This predictionis clearly borne out in Munger's (1984) study. He estimated a probit model of the probabilitythat a certain legislator receivesa donation froma given firm.He showed thatpoliticalaction committeesare systematically more likelyto donate to members of committeesthat affecttheir firms:the probabilitythat a committee member will receive a donation is .34 higher than that of a nonmember. VI. Comparative Statics: Predictions and Evidence
In a simple marketfor votes, a small change in the relativecomposition of interestgroups leads to a small change in the demand for votes.This, in turn,leads to a small change in the equilibriumpattern of exchange and hence in the distribution of policycostsand benefits. However, our argument about the demand for durable policies and the evolution of institutions to provide them implies that policies are partiallyinsulated from small changes in member preferences. Because committeesretain a veto over policy change, we must look to how these changes affectcommitteemembers.If the change in interest groups affectsonly legislatorswho are not members of the comless likely.But our model mittee,then policy change is significantly also leads to an importantcomparative staticsprediction:a sufficient condition for policy change is that there is a substantialturnoverin committeemembershipso that the new holders of committeeprop-
156
ECONOMY
ertyrightshave preferencesthatdifferfromthose of theirpredecessors (see Weingast 1981; Weingast and Moran 1983). While comparative staticsresults are a primarytool of prediction and testingin economics, few studies of politicaleconomy have used this approach to test theories of politics. Nonetheless, there exists some evidence on the predictionnoted above in the empirical literature. We cite these studies and then suggest furthertests. A. Appropriations Ferejohn (1974a) again plays an important role here. During the 1950s and early 1960s, fiscal conservativesdominated the congressional appropriationsprocess. Further,during thisperiod, committee leaders had nearly absolute power of assignmentof membersto subcommittees.One way of enforcingfiscalrestraint was to assign members of the AppropriationsCommitteeto a subcommittee only ifthey jurisdiction. By the mid-1960s, had no stake in the subcommittee's however, this rule had gone by the wayside so that subcommittees came to be composed of members witha high stake in theirjurisdiction. Ferejohn showed that,for the Public Works Subcommittee, this led to a statistically significant increase in appropriations. B. Agencies Regulatory
A host of recent studies of regulatoryagencies has shown that committeemembershave substantialinfluenceover agencies withintheir jurisdiction (Barke and Riker [1982] on the Interstate Commerce Commission, Grier [1984] on the Fed, Moe [1985] on the National Labor Relations Board, and Weingastand Moran [1983] on the FTC). In nearly all cases, these statistical studies showed that,as committee preferenceschange, so too does agency policy. Large swingsin committeepreferenceslead to large swingsin policy. Weingastand Moran (1983), forexample, studied the recentpolicy change at the FTC. In 1979 and 1980, the commission'saggressive consumer activistpolicies were halted by Congress. While thisaction was hailed as Congress's finallycatching a runaway, out-of-control bureaucracy,Weingast and Moran showed that nothing of the sort happened. Instead, the FTC had been under the influence of the relevantsubcommittee all along. From the late 1960s throughthe mid to late 1970s, thissubcommittee both favoredand fosteredaggressive consumeristpolicies. However, followingthe 1976 election,a nearly complete turnoverin membershipbrought to power members with different substantially preferences.Weingast and Moran interpreted the 1979-80 episode as the new committee'ssimply reversing the
INDUSTRIAL
ORGANIZATION
157
policies of theirpredecessors ratherthan catchingan uncontrollable testssupport this interpretation. bureaucracy.Their statistical VII. Discussion
Representativesof different constituencieshave considerable incentivesto exchange support so as to provide benefits to theirsupporters. Because the value of today's legislativebargains depends on actions taken in futurelegislativesessions, legislatorsalso have incentivesto As in devise institutions that provide today'sbargains withdurability. all exchange settings,the institutions that evolve to support the exchange reflect the specificpatternof transactioncosts underlyingthe of conpotentialtrades. For legislaturesthese include the possibility tingenciestoo numerous (or costly)to specifyin advance and private information. This gives rise to a host of institutions underpinninga set of propertyrightsloosely referredto as the committeesystem. We showed that these institutions lower the riskof ex post opportunistic behavior that would plague explicitexchanges of votes. The legislative institutions thereforelower the agency costs associated with exchange. In addition we showed why this set of institutions is superior to a marketexchange mechanism. Instead of tradingvotes,legislatorsexchange special rightsaffordingthe holder of these rightsadditional influenceover well-definedpolicyjurisdictions.This influencestems from the propertyrightsestablished over the agenda mechanisms, that is, the means by which alternativesarise for votes. The extra influenceover particularpolicies institutionalizes a specificpatternof trades. When the holders of seats on committeesare preciselythose individuals who would bid for votes on these issues in a marketfor votes,policychoice under the committeesystem parallels thatunder a more explicitexchange system.Because the exchange is institutionalized, it need not be renegotiatedeach new legislativesession,and it is subject to fewerenforcementproblems. CommitThe committeesystemalso influencescoalitionformation. tee agenda power implies that successfulcoalitionsin the area of the jurisdiction must include the committee.This rules out, committee's forexample, policies thatbenefitsolelya coalitionof membersoffthis of committee, and thisholds even if thiscoalitioncontainsa majority the entire legislature.Unless a coalition of non-committee members is prepared to include or "buy out" the committee, veto power allows the committeeto block access of this coalition to the floor. We also showed thatpolicybargains,and hence coalitions,are more durable under the committeesystem.Thus the decision to enter into such an agreement is much like entering a long-termcontract,and
158
JOURNAL
OF POLITICAL
ECONOMY
legislatorswill take this into account. This implies thatcoalitionswill as they not alwaysrespond to small changes in politicalcircumstances would under a spotlikemarketexchange system.Rather theytend to Commitrespond only to large shifts or major politicalrealignments. tee veto power combines withthe propertyrightsystemover seats to play an important role in maintaining a political coalition-and a particular policy-for long periods. Policy in a particulararea may remain stable if committeemembership is relatively stable, and this can hold even withmajor changes in the preferencesof membersoff the committee.The abilityto veto the proposals of others is a subtle yet powerful tool used by committees to influence policy in their jurisdiction(Weingastand Moran 1983; Shepsle and Weingast 1987). This argument raises some interesting parallels and contrastswith those provided for vertical integrationin market settings.In both cases, institutions are designed to preventsimilar formsof incentive problems, for example, ex post opportunism. However, it appears that the source of these problems differs.For the case of vertical integration, it is relation-specific assets. For the legislature,however, incentiveproblems arise because there is no underlyingmedium of exchange so thattradingvotes requires futurereliance and hence the opportunitiesforreneging (see n. 10). Moreover,as Ferejohn (1974b) has shown, it is not clear whether one can exist, given the peculiar externalities associated withvote trading. We have pursued in this paper only one explanation for enforcing trades. It is useful,therefore, to discuss a numberof potentialalternais beyond the scope of tives,though a full-scale empiricalinvestigation this paper. The first alternativeis that ex post opportunismeitheris allowingexchange negligibleor is handled in some otherway,thereby to take place throughtrading.According to thisview,the existenceof committees is epiphenomenal, perhaps representing some formal (though unimportant) recognition of those legislatorswho have in fact "bought" influenceover particular issues. An empirical test between this explanation and our model might focus on the responsivenessof policychoice to membersof the committee.In an explicit exchange setting,large changes in the preferencesof members off the committeeshould lead to changes in policy.Under the legislative committee model, committee veto rightsimply that policy is more insulated from changes of this type, and hence we should observe policies to be less responsive. ParA second competing explanation is perhaps more interesting. ties,ruled out by assumption in our model, offeran obvious alternaand enforcing trades. The historicalevitive for institutionalizing dence for the U.S. Congress suggests that strong parties and strong committees,as institutionalunderpinnings of legislativeexchange, are substitutes. When partieswere more powerful(e.g., at the turnof
INDUSTRIAL
ORGANIZATION
159
the century), committees, though important, did not have such clearcut rightsas in modern times. Seniority,for example, was regularly violated by party leadership in allocating the leadership positions withincommittees.Importantly,virtuallyevery institutional change during this century that has made committee rights stronger has come at the expense of parties and centralizedleadership. This suggests a natural extension of our approach to the case of partygovernment(which includes the BritishParliamentin addition to the House of Representativesof the past). Strong partiesare characterized by control over importantresources such as entryinto the competitionforindividual seats and the positionsof power withinthe legislature(e.g., the ministerialpositions in Britain),and theywield considerable influenceover the distribution of legislative(read: electorallyuseful) benefits.Parties,like firms, can build typesof reputationsdifferent fromthose of the individualswho make them up (see, e.g., Kreps 1984). To the extent that they are able to influencethe behavior of theirmembers throughdistribution of resources,parties potentially provide an alternative means of enforcing agreements.We hope to extend our approach in the futureto yield resultsabout the institutions underpinning legislativeexchange in this context.28An importantissue of this research concerns the circumstances favoring the survivalof one mechanism over the other. One limitation of our analysisis that,while we argue thatlegislative rules mitigatecertain contractual problems, we do not explain how the rules themselvessurvive.Since majoritiesmay alterthe rules,what preventsthe breakdown of cooperation thattakes on a slightly different form?In circumstances in whichreneging,say,would occur without rules, what preventsindividuals from first voting to change the rules and then reneging? An extensive investigation of this issue is beyond the scope of this paper. However, there appear a variety of circumstancesunder which the rules will survive a breakdown whereas cooperation withoutrules would not. For example, if many different policyjurisdictions are governed by the same set of rules, then a single set of rules may link behavior in one area withthat in another. Hence incentivesto renege in one area do not automatically resultin corresponding incentivesto change rules that govern many areas.29 Since it clearlytouches on issues thathold for a large variety of organizations,this question is worthyof a separate investigation.
For an interesting beginningon this problem, see Leibowitz and Tollison (1980). As a second set of circumstances, we single out the notion of leadership explored byCalvert (1986) in his extensionof the Kreps and Wilson (1982) model to legislatures. Calvert studied circumstancesin which a particular individual is given resources by otherindividuals.Withthese resources,he then,e.g., polices the behaviorof his followers. In principle,this mechanismmightbe used to preventthe breakdownof cooperation in certain circumstancesand thereforebe valuable ex ante to members.
28 29
i6o
JOURNAL
OF POLITICAL
ECONOMY
The empiricalevidence supportsfourimplications thatfollowfrom our model of legislativeinstitutions but do not followfroma simple market exchange mechanism. First, committees are composed of "high demanders," that is, individuals with greater than average interestin the committee'spolicyjurisdiction. Second, the committee assignmentmechanismoperates as a bidding mechanismthatassigns individualsto those committeestheyvalue most highly.Third, committeemembers gain a disproportionateshare of the benefitsfrom theirpolicy area. This appears to hold across widelydiffering policy jurisdictions. Fourth, there exists importantevidence supporting a comparativestaticspredictionof the model, namely,thatas the interests representedon the committeechange, so too willpolicy,withthe interestsof non-committee members held constant. Evidence supportingthispropositionexistsin several regulatory areas; futuretests will reveal the robustnessof the results. of Congress appear remarkablysuited to In sum, the institutions legislators' reelection goals. Their specific form appears to have evolved to reduce problems that also arise in market exchange, namely,problems of measurement,moral hazard, and opportunism.
References Alchian, Armen,and Demsetz, Harold. "Production,InformationCosts, and Economic Organization." A.E.R. 62 (December 1972): 777-95. New ofInfluence. A Theory and theBureaucracy: Arnold, R. Douglas. Congress Haven, Conn.: Yale Univ. Press, 1979. New York: Basic Books, Axelrod, Robert M. The Evolutionof Cooperation. 1984. Barke, Richard P., and Riker,William H. "A PoliticalTheory of Regulation 39, no. 1 withSome Observationson RailwayAbandonments."PublicChoice (1982): 73-106. Barzel, Yoram. "Measurement Cost and the Organizationof Markets."J.Law and Econ. 25 (April 1982): 27-48. Becker, Gary S. "A Theory of Competitionamong PressureGroups forPolitical Influence." QJ.E. 98 (August 1983): 371-400. Logical Buchanan, James M., and Tullock, Gordon. The Calculus of Consent: Ann Arbor: Univ. Michigan Press, Democracy. Foundations of Constitutional 1962. and Equilibrium in an Infinite-HorizonReciCalvert, Randall. "Efficiency procityGame." Mimeographed. St. Louis: WashingtonUniv., 1985. . "The Role of Reputation and Legislative Leadership." Mimeo graphed. St. Louis: Washington Univ., 1986. Cheung, Steven N. S. "The ContractualNature of the Firm."J.Law and Econ. 26 (April 1983): 1-2 1. 4 (November 1937): Coase, Ronald H. "The Nature of the Firm."Economica 386-405. Cohen, Linda R., and Noll, Roger G. "Political-Economyof Government Sponsored R & D." Mimeographed. Stanford,Calif.: StanfordUniv., 1986
INDUSTRIAL
ORGANIZATION
i6i
Demsetz, Harold, and Lehn, Kenneth. "The Structureof Corporate Ownership: Causes and Consequences."J.P.E. 93 (December 1985): 1155-77. Denzau, Arthur T., and Munger, Michael C. "Legislators and Interest Polit.Sci. Groups: How Unorganized InterestsGet Represented." American Rev. 80 (March 1986): 89-106. Pork Faith,Roger L.; Leavens, Donald R.; and Tollison, Robert D. "Antitrust Barrel."J. Law and Econ. 25 (October 1982): 329-42. Fama, Eugene F. "Agency Problems and the Theory of the Firm."J.P.E. 88 (April 1980): 288-307. Fama, Eugene F., and Jensen, Michael C. "Separation of Ownership and Control."J. Law and Econ. 26 (June 1983): 301-25. in Committees. Boston: Little,Brown, 1973. Fenno, Richard F. Congressmen 1947Ferejohn, John A. Pork Barrel Politics:Riversand HarborsLegislation, 1968. Stanford,Calif.: Stanford Univ. Press, 1974. (a) . "Sour Notes on the Theory of Vote Trading." Social Science Working Paper. Pasadena: California Inst. Tech., 1974. (b) Context: A Case Studyof Food Stamp "Logrolling in an Institutional Jr., and PolicyChange,edited by Gerald C. Wright, Legislation." In Congress Leroy N. Rieselbach, and Lawrence C. Dodd. New York: Agathon, 1986. Fiorina, Morris P. "Congressional Control of the Bureaucracy." In Congress Reconsidered, 2d ed., edited by Lawrence C. Dodd and Bruce I. Oppenheimer. Washington: Congressional Quarterly Press, 1981. (a) in American New Haven, Conn.: . Retrospective NationalElections. Voting Yale Univ. Press, 1981. (b) Cambridge, Mass.: MIT Press, Fromm,Gary,ed. Studiesin PublicRegulation. 1981. of Resale Price Maintenance." Gilligan,Thomas W. "The CompetitionEffects RandJ. Econ. 17 (Winter 1986): 544-56. Goldberg, Victor P. "Regulation and AdministeredContracts."Bellj. Econ. 7 (Autumn 1976): 426-48. Grier,Kevin. "Congressional Preferencesand Fed Policy: A PrincipalAgent Approach." Mimeographed. St. Louis: Washington Univ., Center Study American Bus., 1984. Grossman,SanfordJ., and Hart, Oliver D. "The Costs and Benefitsof Ownership: A Theory of Vertical and Lateral Integration." J.P.E. 94 (August 1986): 691-719. Holmstrom, Bengt. "Moral Hazard and Observability."Bell J. Econ. 10 (Spring 1979): 74-91. Rev. Jensen,Michael C. "Organization Theory and Methodology."Accounting 58 (April 1983): 319-39. Jensen,Michael C., and Meckling,William H. "Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure."J.FinancialEcon. 3 (October 1976): 305-60. Jones,Charles 0. "The Role of the Congressional Subcommittee." MidwestJ. Polit. Sci. 6 (November 1962): 327-44. Joskow,Paul L. "VerticalIntegrationand Long-Term Contracts:The Case of 1 Coal-burning Electric Generating Plants."J. Law, Econ. and Organization (Fall 1985): 33-80. Kalt,Joseph P., and Zupan, Mark A. "Capture and Ideology in the Economic Theory of Politics."A.E.R. 74 (June 1984): 279-300. Ideology, and Logrollingin Kau, James B., and Rubin, Paul H. "Self-Interest, Congressional Voting."J. Law and Econ. 22 (October 1979): 365-84.
162
ECONOMY
Klein, Benjamin; Crawford, Robert G.; and Alchian, Armen A. "Vertical Integration,Appropriable Rents, and the Competitive ContractingProcess." J. Law and Econ. 21 (October 1978): 297-326. Klein, Benjamin, and Leffler, Keith. "Role of Market Forces in Assuring Contractual Performance."J.P.E. 89 (August 1981): 615-41. Koford,KennethJ. "Centralized Vote-trading." PublicChoice39, no. 2 (1982): 245-68. Kreps, David M. "Corporate Culture." Mimeographed. Stanford,Calif.: Stanford Univ., Grad. School Bus., 1984. Kreps, David M., and Wilson, Robert. "Reputation and ImperfectInformation."J. Econ. Theory 27 (August 1982): 253-79. Leibowitz,Arleen, and Tollison, Robert D. "A Theory of LegislativeOrganization: Making the Most of Your Majority."QJ.E. 94 (March 1980): 261McKelvey, Richard D. "Intransitivities in MultidimensionalVoting: Models and Some Implications for Agenda Control." J. Econ. Theory12 (June 1976): 472-82. . "General Conditions for Global Intransitivities in Formal Voting Models." Econometrica 47 (September 1979): 1085-1112. Malone, Michael T. "The PoliticalDeterminantsof the Flow of Federal Funds to Statesand the Impact of Federal Funds on Business Location Decisions." Paper presented at the LibertyFund Symposiumseries,GovernmentControl of EntrepreneurialDecisions, Indianapolis, November 1982. Mayhew, David R. PartyLoyalty among Congressmen: The Difference between Democrats and Republicans, 1947-1962. Cambridge, Mass.: Harvard Univ. Press, 1966. . Congress:The ElectoralConnection. New Haven, Conn.: Yale Univ. Press, 1974. Milgrom,Paul, and Roberts,John. "Bargaining and Influence Costs and the Organization of Economic Activity."Research Paper no. 934. Stanford, Calif.: Stanford Univ., Grad. School Bus., 1987. Miller,GaryJ., and Moe, Terry M. "The PositiveTheory of Hierarchies." In PoliticalScience:The ScienceofPolitics, edited by Herbert F. Weisberg. New York: Agathon, 1986. Moe, Terry M. "The New Economics of Organization." AmericanJ. Polit.Sci. 28 (November 1984): 739-77. . "Control and Feedback in Economic Regulation: The Case of the NLRB." American Polit. Sci. Rev. 79 (December 1985): 1094-1116. Munger, Michael C. "Neoinstitutional Response to Demand Oriented Models of Public Policy: Theory and Evidence." Mimeographed. St. Louis: Washington Univ., 1984. and Changein Economic North,Douglass C. Structure New York: NorHistory. ton, 1981. Peltzman,Sam. "Toward a More General Theory of Regulation."J. Law and Econ. 19 (August 1976): 211-40. ."Constituent Interestand Congressional Voting."J.Law and Econ. 27 (April 1984): 191-210. Polsby,Nelson W. Congress and thePresidency. 4th ed. Englewood Cliffs, NJ.: Prentice-Hall,1984. Roberts,John. "Battles for Market Share." Mimeographed. Stanford,Calif.: Stanford Univ., Grad. School Bus., 1986. Rubin, Paul H. "The Theory of the Firm and the Structureof the Franchise Contract."J. Law and Econ. 21 (April 1978): 223-33.
77.
INDUSTRIAL
ORGANIZATION
163
Rundquist, Barry. "Congressional Influence on the Distributionof Prime Military Contracts."Ph.D. dissertation,Stanford Univ., 1973. Theorem." Public Choice37, Schwartz,Thomas. "The Universal-Instability no. 3 (1981): 487-501. Shepsle, Kenneth A. "Congressional CommitteeAssignments:An OptimizaPublicChoice22 (Summer 1975): Constraints." tion Model withInstitutional 55-78. in the Modern Committee Assignments JigsawPuzzle: Democratic . The Giant House. Chicago: Univ. Chicago Press, 1978. Shepsle, Kenneth A., and Weingast, Barry R. "Structure-inducedEquilibrium and Legislative Choice." Public Choice37, no. 3 (1981): 503-19. 46 (February 1984): ."When Do Rules of Procedure Matter?"J.Politics 206-21. Polit. . "The Institutional Foundations of CommitteePower." American Sci. Rev. 81 (March 1987): 85-104. Telser, Lester G. "A Theory of Self-enforcing Agreements."J.Bus. 53 (January 1980): 27-44. Tirole, Jean. "Hierarchies and Bureaucracies: On the Role of Collusion in 2, no. 2 (1986): 181-214. Organizations."J. Law, Econ. and Organization ofPolitics.Ann Arbor: Univ. MichiTullock, Gordon. Towardsa Mathematics gan Press, 1967. Public Choice37, no. 2 (1981): 189-202. . "Why So Much Stability?" Weingast,Barry R. "Regulation, Reregulation,and Deregulation: The Political Foundations of Agency Clientele Relations." Law and Contemporary 44 (Winter 1981): 147-77. Problems . "The Congressional-Bureaucratic System: A Principal-AgentPerspective(withApplication to the SEC)." PublicChoice44, no. 1 (1984): 147Weingast,Barry R., and Moran, Mark J. "Bureaucratic Discretionor Legislaby the Federal Trade Commission." tiveControl? RegulatoryPolicymaking J.P.E. 91 (October 1983): 765-800. and Antitrust ImplicaAnalysis Williamson,Oliver E. Marketsand Hierarchies: tions.New York: Free Press, 1975. Relational Institutions Contracting. Firms, Markets, Economic ofCapitalism: New York: Free Press, 1985. Wilson, Robert. "An Axiomatic Model of Logrolling." A.E.R. 59 (June 1969): 331-41.
91.

Industrial Organization Congress

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Industrial Organization Congress

Uploaded by

Copyright:

Available Formats

The Industrial Organization of Congress; or, Why Legislatures, Like Firms, Are Not Organized as Markets Author(s): Barry

Preference .368 .306 .254 .308 .144 .166 .243

TOTAL FIRS T PREFERENC E ASSIGNMENT SUCCESS

Less than 1 94.4 5.6

Committee Mean 76.8 50.2 46.5

59.1 61.7 37.5

37.9 50.4 46.7

55.5 60.0 58.3

9.57 3.33 2.08*

Water and Sewage Grant Selection (1970)

Subcommitteeof Committee Banking and Currency Committee Neither committee Total

.313 .281 .176

Model Cities Project Selection

Subcommitteeof Committee Banking and Currency Committee Neither committee Total

.86 .62 .29

10.81 and significance level is .01.

CASE WHEN A SENATOR'S

Change in Probability .005 .013 .060

and Moran (1983).

You might also like