rect.com Available online at www.scienceldi
Pergamon
s c r E N c. C d ) " r R E c r o \TheSocialScienceJournal40 (2003\35'7-369
The
SOCIAL SCIENCE JOURNAL
InterestgroupPAC contributionsandthe 1992 regulationof cabletelevision JeffreyE. Cohenu'*,JohnA. Hammanb " Department of Political Science,675 Faber Hall, Fordham University, 441 East Fordham Road, Bronx, NY 10458, USA b SouthernIllinois tlniversin. Carbondale. IL. USA
Abstract The literature concerning political action committee (PAC) influence over congressionalroll-call decisions offers mixed findings. Some studies find strong effects, others weak eiTects,while some detect no PAC impact on congressionalroll-call voting. These nLixedresults imply that PAC influence might vary across contexts and perhaps interact with other variables. In this paper, we suggestthat PAC influence might vary with two contexts.First, we hypothesizethat PAC influence wili vary with chamber, and in particular, PAC effects will be stronger in the llouse than in the Senate.Second,we hypothesizethat PAC effects will be strongerin policy subsystemsthan at the macropolitical level. We test thesehypotheseswith the votes on the 1992 Cable Television ReregulationAct. The fact that both chamberstwice votedupon identicallywordedbills, oncefor passage,onceto ovenidePresidentBush's veto, allows us to construct a quasi-experimentaldesign. Results of the analysisgenerally support our hypotheses.We conclude with some suggestionsfor future researchon the linkage between interest groups, PAC contributions, and congressionalpolicy making. @ 2003 Elsevier Inc. AII rishts reserved.
The conventionalwisdom of journalists,politicians,and the public is that interestgroups hold greatinfluenceover the decisionsof legislators.Political actioncommittee(PAC)contributions to legislativeelection campaignsare often touted as one sourceof tha.tinfluence (West & Loomis, 1999;Wright, 1996).However,considerablecontroversyexists among scholars over whetherPAC contributionsaffect legislative decisionmaking. Some studiesfind substantial impact, othersmodestimpact, and still othersno impact (Baumgartner&Leech,1999; Smith, 1995).1Studiesoften attributethesedisparitiesto contextualvariables,However,Smith contends,they do so without much direct evidence(1995.p.94\.2 - Correspondingauthor.Tel.:
* I -7 18-8 11 -3956. (J.E.Cohen). E-mail address.'
[email protected] 0362-33191$- seefront matter @ 2003 Elsevier Inc. All rights reserved 19(03)00035I doi:10.1016/50362-33
358
J.E. Cohen,J.A. Hamman/The Social ScienceJoumal 40 (2003) 357-369
To directly study the impact of contextualfactors requiresthat thesevariablesbe systematically incorporatedinto the researchdesign.This can be done by holding noncontextualvariables, in particular PAC contributions and the policies being decided upon, constant.Rarely havewe been ableto designsuchstudies.As a consequence, little progressha.sbeenmadein conditions under which the PAC isolating contributionsinfluLencelegislative decisionmaking (Baumgartner& Leech, 1999). However,in this paper we do use such a quasi-experimental designto isolatethe impact of two contextualfactorson passageof the 1992CableReregulationAct: legir;lativechamber and subgovernmenteffects.rWe test two hypothesessuggestedin the literature.The first is that PAC effects will be stronger in the House than in the Senate.The secondis that PAC effectswill be strongerat the subsystemthan at the macropoliticallevel. SincePresidentBush vetoed the 1992 Cable ReregulationAct and both congressionalchambersvoted to override the president'sveto,we areableto hold PAC contributionsandpolicy constant,while allowing thesetwo contextualfactors, chamberand subsvstem.to varv.
L. Chamber differences and PAC influence We expect that the differencesbetweenthe House and the Senatewill result in greaterPAC influencein the Housethan in the Senate(Baker,1989;Lee & Oppenheimer,11999). First, the more frequent election of representativesshould increasethe sensitivity of the lower chamber legislatorsto electoralfactors(Kuklinski, 1978),suchasPAC contributions.Second,thelarger size of Senaterelative to House constituenciesshould increasethe number of intereststhat senatorshave to contend with and dilute the effect of any one group comparedto their effect Larger constituenciesare likely to be more diverseand result in senators on representatives. facing a more varied set of interestsand PACs. Greaterdiversity increasesthe prospectof senatorsfacing competing interestsand more difficult decislons.It is much easierto contend with one interest(or a set of interests)on one side of an issue.When facing competitiveand opposedinterests,taking a position for one interestwill necessarilyalienatethe opposing side. Refusingto take a position may alienateboth sides.tAC contributionsalso constitute a smaller proportion of overall campaignfinancesfor sena.tors in part tlran representatives. pay heed to the preferencesof becauseSenatecampaignsare more expensive.If legislator:s campaign contributors,we expectPACs to speakwith a louclervoice in the House than in the Senate.Thus, we expect to find that PAC contributions have a greaterimpact on House than Senateroll calls.
2. Policy subsystemsand PAC influence Scholars have long held that interest groups are influentiLalwhen issuescan be contained within tightly knit policy subsystems(or subgovernments).Alternatively, when issuesrise to the macropolitical level (Redford, 1969), the influence of IACs should decline. Policy subsystems,sometimescalled iron trianglesor issuenetworks (Heclo, 1919),ale characterized by limited participation and autonomousalliances among legislative committees, agencies,
J.E. Cohen.J.A. Hamman/The Social ScienceJournal 40 (2003) 357-369
interest groups and expertswithin a specific substantivepollicy area.Interestgroupsare important participantsin subsystempolicy making. Often policy is made to please,the participating interestgroup.However,congressional floor debatesbecomemacropoliticalarenaswhenpresparties, political the mass idents, media, andthe masspublic modify or redefirnethe issueunder debateand/or assumepolicy making authority (Baumgartner& Jones,1993).Theseactorscan redefineissues,and lessenthe relativeimpact of interestgroups. disruptpolitical subsystems, Thus, we expectthat interestgroup (and/orPAC) influencewill decline as an issuemovesfrom a subsystemto the macropoliticalarena.
3. The 1,992Cable Reregulation Act: from subsystem l;o macropolity In this section, we briefly review the history and issu3s associatedwith the 1992 Cable ReregulationAct. Most importantly for our analysisis, how Bush's veto moved the issue of cable reregulationfrom a tightly structuredpolicy subsystem,in which cable television companiesand their alliescontrolledpolicy making,to a nnacropoliticalarena.This redefined the regulatorypolicy issueinto a test of the Bush presidency. Cable televisionreregulationin 1992 was a reactionto the 1984 Cable Communications Policy Act, one of the many piecesof deregulationenactedin the late 1970s'and early 1980s. The 1984 act deniedlocal municipalitiesthe power to set cable rate structuresand program services,althoughthey retainedthe authorityto grantlicensesto operators.lts a consequence, subscriptionratesfor cableservicerosesteeply.The GeneralAccountingOflice estimatedthat cableratesrose43Vofasterthan inflation from 1984to 1990(Garay,1988). Throughoutthe 1980s,the cable industry effectivelyresistedreregulation.Although conby consumer complaintsabout rising cable ratesmounted,legis.lativeinitiativess:ponsored sumeradvocatestypically diedin committee,generallywithout hearings.For instance,in 1987, television satellite dish owners protestedthat cable signal scrambling techrLiqueshad cut off their accessto cableprogramming.aCabletelevisioninterestsdefeatedthe effortsof dish owners to modify their scramblersto accesscable signals(C,cngressional Quarterly,1987).The cableindustry appearedunassailable. Only with the mobilization of the broadcastingindustry, the configura.tionof the cable policy subsystemaltered and the allies of cable finally defeated.The mobilization coincided with the inclusion of a rebroadcastingfee provision containedin legislationproposedin the SenateCommerce, Science,and TransportationCommitt,:e in 1992. The rr:broadcastingfee provision allowed cable companiesto continue to usebroadcastmedia prog;rammingfor free only if given consentby the broadcastmedia to do so. Otherwisecable had to pay for the programming. The fee provision gained the support of broadcastinterests,in particular the strongNational Associationof Broadcasters. But it also crystallizedopposrition.Hollywood program producers would not receive a portion of the rebroadcastingfe€)in the proposed legislation.Thus, they joined forceswith the cableindustry (CongressionalQuarterly,1992). In this way, the rebroadcastingfee issuemobilized new interest groups and nnovedthe issueof cable television regulation from one subsystemto another, The Senatepasseda reregulationbill, S.12,by a73-18 margin in January1992.In July, the Housepassed(340-73) its own version(HR 4850). f'he two bills differed on one major
J.E. Cohen,J.A. Hamman/The Social ScienceJournal 40 (2003) 357-369
item: rebroadcastingfees, a provision included in the Senate,but not the l{ouse bill. The provision was droppedfrom the House version when challengedby Jack Eirooks(D-TX), chair of the House Energy and Commerce Committee, who sought unsuccessfullyto attach a provision that would charge cable a flat fee to compenr;ateboth local broadcastersand Hollywood studiosfor the use of local and distantsignals(Mills, 1992).During August and September,confereesworked out a compromisebill that retainedthe rebroadcastingprovision. Both chambersapprovedthe conferencebill, theHousedoing soby a wide margin(280-128)in mid-September. PresidentBush publicly opposedcablereregulationthroughoutthe 1991-'1992legislative session,but the administrationwas divided on what action to take.Although the administration viewed the bill asjust more regulation,which it opposed,it. also faceda difficult re-election race and realized that the bill was popular. Indecisivenessand lack of action weakenedthe administration'shand in cablepolicy, but Bush's October3 veto catapultedthe issuefrom a policy subsystemto the macropoliticallevel. StrenuousWhite Houselobbyinigto preventan override of his veto reinforced the view of the cable policy debateas a test of the presidency. Furthermore,Bush tried to reframecableinto an economicissue,arguingthat the legislation would increasecable television rates, which might fuel inflation and undermine economic growth.Bush's veto was overridden.The Senateoverridevote was identicaltc,theconference vote (74-25; Republicans,24-18,Democrats,50-7), while the House vote s;howsa modest movementof Republicansto supportand Democratsto opprose the presidenton the override as comparedto the conferencevote.5
4. Data and methods 4.l. Dependentvariables The House and Senatevotes on the conferencereport and the veto overr.ideof the 1992 Cable ReregulationAct are the dependentvariablesin the ensuinganalysis.T'hesefour votes were taken on identicallegislation.Each vote is codedas a dummy variable,where a vote in supportis codedone and a vote againstis coded zero.6Eacl:tvote is regressedon independent variablesmeasuringPAC contributions,constituencyinterests,and presidentialsupport,our three setsof independentvariables. 4.2. Independentvariables 4.2.l. PAC contributions Table 1 reports descriptivestatisticson contributions front the Hollywood, ,cable,or broadcastingPACs.On average,senatorsreceivedmore contributionsthan representatives, and cable PACs,on average,contributedmore than either Hollywood or broadcastinterests.It also appearsthat the PACs targetedtheir funds strategically.Only aLbouthalf of the membersin each house received any funding, with considerablevariation in the amount legislators received. Given their issuepositions,we expectnegativesigns for cable and Hollywood and positive signsfor broadcastindustrycontributions.
361
J.E. Cohen,J.A. Hamman/The Social ScienceJoutnal 40 (2003) 357-369
Table I Descriptive statisticsof contributions receivedby House and Senatemembersof Congressfrom cable. Hollywood, and broadcastindustry PACs, 1992 lnterest type
N
Mean
S.D
Minimum
Maximunr
Not funded
House Hollywood Cable Broadcasting
435 435 435
453 1427 825
1625 36'.73 2356
0 0 0
16000 26500 25000
282 286 353
Senate Hollywood Cable Broadcasting
100 100 r00
1640 5023 1469
3514 9431 2821
0 0 0
23'750 51000 10000
56 66 62
andHollywood fromFEConlinelistingsbased of cable,broadcasting, Datacollected uponCQlist.ing Source: (Congressional p.2'7991. reregulation in cable 1992, PACsactive Quarterly, 4.2.2. Constituencl,interests The literature on congressionaldecisionmaking suggeststhe importanrceof constituent economicinterest,as well as policy preference(Arnold, i990). We empkly the percentage district/statefrom 1989 increasein monthly ratesforbasic cableservicein eachcongressional to 1992 to measureconstituent/districteconomic interests.We expectthat p;reaterpercentage pressureon increaseswill correspondto morefrequentconsumercomplaintsand/orconsl.ituent legislatorsto amendcablepolicy. A secondvariable,marketpenetration,measuresconstituent in a districtor statethat demandfor cableservices.This is deflnedasthenumbero1'households subscribeto cable televisiondivided by the numberof householdspassedby cablevendors.T We expect a positive relationshipbetweenthe level of penetrationand supportfor reregulation legislation. 4.2.3. Presidentialsupport One of our hypothesesconcernsthe impact of shifting the arenaof decision making from the policy subsystemto macropoliticallevel.Legislators,who are generallysupportiveof the president,should be expectedto side with him, especiallyonce he vetoeclthe bill. We use the CQ presidential support score to measuresupport for the president,but we recalculateit to eliminatethe contaminatinseffect of the cablereeulationvotes that were includedin the original scores. 4.2.4. Estimationissues Before turning to our findings,we needto discusstwo complicatingestimationissues:srmultaneity betweenroll-call voting andPAC contributions,and censoringof JPACcontributions (e.g.,the large number of zero contributionlevels).The possiblesimultaneitybetweenPAC contributions and roll-call voting may overstatePAC impact if one uses a rsingleequation to estimatemodelparameters.Severalstudiesrecommendtwo-stageproceduresto separatethese a first stage effects(Fleisher,7993;Grenzke,i989; Hall & Wayman,1990).Thus,we esrlimate equationof the determinantsof PAC contributions,usingpastliteratureasa guideto identify factors that seemto affectcontributionlevelsandtargets(Grier & Munger, 1986;Hall & Wayman, 1990;Langbein, 1993).From this estimation,we derive an instrumentfor eachtype of PAC
30/,
J.E. Cohen,J.A. Hamman/The SocialScienceJournal 40 (2003)357-369
contribution. The instrumentsar-ethen usedin our secondstageequationsin which we estimate the impact of the instruments,presidential support, and constituent interestson the roll-call votes. thatis,PACscontribute A secondestimationissueis thatPACcontributionsmay be censored, nothing to somelegislatorsand/orrelatively small amountsto many others.This:resultsin many zero and small valuesfor PAC contributions. Suchis the case,inour study.To compensatefor this, we will use tobit to estimatethe first stageequation(Chappell,1982; Tobin, 1958) on determinantsof PAC contributions. equationfor eachPAC (i.e.,Hollywood,cable,andbroadcasting) We first estimatea separate for eachchamber,a total of six equations.Exogenousvariablesin theseestimationsincludeour measuresfor constituencyinterestsandpresidentialvoting.The equationalsoirrcludesa series of variableslikely to predict PAC receipts,but be uncorrelatedwith our dependentvariables. These variables, such as ideology, electoral security, seniority, party and Pl'C receipts for the previous year, reflect both the attractivenessof legislatorsto PAC contributors,as well as each legislator'sdesirefor funding. The Senateequationsalso include a variableto control for state population. Finally, the first stage equations inslurCevariables for nnembershipon relevantcommitteeswith oversightover cable legislation: Co:mmerceand Banking Committee and Rules and Administration Committee in the Senate-Energy and CommerceCommittee, Rules Committee,and JudiciaryCommitteein the House.R,3sultsof the first stageequations arepresentedin Tables2 and3. As we discussbelow,our first stageresultscomportwell with findings in the literature. From the first stage equation, we extract an instrumental variable for conlributions from each PAC to be used in the secondstage.Since the votes on cable reregulationare limited, properties equations.The erxact dichotomousvariables,we uselogit to estimatethe secondstarle
Table2 First stagetobit estimatesof Housecabie,broadcast,and Hollvwood PAC contribution,b (S.E.) Variables
Cable PAC
BroadcastPAC
Holiywood PAC
CQ presidential support score (adj.) Party CQ conservativecoalition support score District Cable Market Penetration 1992 Change in monthly subscription rate 1992 Seniority Freshman Proportion of vote in previous election PAC receipts rn 1992 Ways and Means Committee Rules Committee Judiciary Committee Energy Committee Constant Ancillary parameter x? Q4) rroDabluly > xN
46 (48) 1 4 8( 1 5 5 8 ) -20 (29) -t260 (2262) -7 (12) '70(4s) 3803--(1009) 46* ,13) 1l* (2.6) -17 (1336) 6540.-(1566) 4 9 5 7 -(-1 0 5 5 ) 1 1 3 9 1 (- 9 - 51) -10629* (2164) s 1 1 4( 3 1 0 ) 204 0 431
-39 (36) 1699(1184) 27 (21) 1415(r5',70) .94(.96) 5r (35) -725 (8e'7) 27* (9) 75.. (2) 2134*'(884) 9s8 (1344) 3471**(834.5) 4558**(735) -9840.. (1640) 3972(249) 138 0 431
11(46) 7 s 3( 1 6 0 8 ) -20 (26) -3593'(2010) -16 (r2) 134--(38) -78 (1152) 1 0( 1 0 ) 13* (2) 2426* (944) (1491) 103"1 4567.-(882) 3s28.'(810) -'7198*'(1799) 3't9r (326) 121 0 431
J.E. Cohen,J.A.Hamman/TheSocialScienceJournal40 (2003)357-369
363
Table3 of Senatecab1e, First stagetobit estimates broadcast, andHollywoodP,\Ccontributions, b (S.8.) Variables
CabiePAC
CQ presidential support score (adj.) Party CQ conservativecoaiition support score District Cable Market Penetration 1992 Change in monthly subscription rate Seniority Freshman Proportion of vote in previous election Running for re-election rn 1992 PAC receipts Commerce Committee Rules Committee State population Constant Anciliary parameter x2 Q4) Probability > 1r N
360*(196.6) *101'78.(6"t14) -s9 (91.7) -1726s(12990) 9173(7466) *2'r8 (298) -'709 (440s) - 3 8 6 . -( 1 6 0 ) 22486* Q0l1) 2 . 1 0( 2 . 2 ) 6543* (2942) 6122'(3827) .0002(.0002) 8264(17275) 9497(1036) 82.86 0 97
Broadcast PAC (70) s5 42t (2334) -50 (34) -3'18 (44s'7) -1603 (2622) -73 (98) -1326 (1465) 44 (47) 8967*'(11't6) .3s(.74) 3911..(999) -491 (1420) .00003(.00008) -5232 (5471) 2962(3'79) 88.03 0 97
HoilywoodPAC 6't (61) -4401(346'7)
(46) 1.ss -479169s8) 2693(3744) 8 0( 1 3 7 ) 2237(2247) -83 (76) 8 1 5 3 -(' ,1' 6 7 5 ) -.08(1.1) -65? Rr1656r
3723.(1961) .0004-"(.0001) -4092 (8s12)
47ss(s73) 61.'7 0 97
of two-stageestimationusing tobit-logit are not well known. Mooney and Duval (1993) and Mooney (1996) argue that bootstrapping can be used to help ascertainthr: true population distributionalqualitiesof parameters, andin our particularr;ase,bolsterconfidencein the coefparametersare reported ficientsand error termsof the secondstageequations.8 Bo,otstrapped analysis. for the second-stage 4.2.5. Findings 4.2.5.1. First stage:factors influencingPAC contributions. The resultspresentedin Tables2 and 3 show the extent to which the independentvariablespredicted the attractivenessof legislatorsto cable,broadcastingand Hollywood PAC contributorsas well as eachlegislator's desirefor funding. Overall,they areconsistentwith expeclations.In the HouLse, the attractivenessof memberssitting on committeesthat had jurisdicti.onover cable legislationis shown by the statisticallyand positivecoefficientsfor the Judiciaryand EnergyCommittees.This is the casefor all three PAC types. Cable interestsalso targetedmembersof the Rules Committee. Broadcastand Hollywood singledout membersof $rays and Means.In addition,many of the seniority and electoralsecurity coefficientsare statisticallysignificantand signed as predicted-freshman in the cableequation,vote receivedin 1990 (for cable andthe broadcast industry equation),and seniorityin the Hollywood equatic,n.Finally, the indicatorfor legislator demandfor campaignfunding, total PAC receiptsin 1992,is statisticallysigniflcantand positivein all threeequations. For the Senate,we find similar results.The CommerceCommitteecoefficientis positive and significant for both the cable and broadcast industry equations and the Rules Committee coefficientis positive but significantat the .10 level for Hollywoorl PACs.Like the House, electoral security was a factor in predicting all three types of Pl\C contributions. The variablefor Senatorsrunnins for re-electiontn 1992 is positive and siLgnificant in each
364
J.E. Cohen.J.A. Hamman/The SocialScienceJournal 40 (2003) 357-369
Table 4 Logit estimates.1992 House and SenateConferenceand ovenide votes on cabie regulation, b (S.E.)
Variabies
Houseconference Houseveto report oven-ide -.05--(.006) CQ presidential support score -.04.- (.005) (adj ) District Cable Market Penetration1992 Change in monthly subscription rate 1992 Cabie PAC contributions BroadcastPAC contributions Hollywood PAC contributions Constant X1 (1) FroDaDrllty> xPseudoRr
Senateconference report
Senateveto override
-.05-- (.02)
-.05-* (.02)
*L46 (1.20)
-.72 (1.17)
3.24(4.4t)
3.24(4.41)
-.00s (.003)
-.003 (.003)
2.81(7.31)
2 . 8 1( 2 . 3 1 )
-.00006(.00004) .0003.-(.0001) -.0002--(.0001) 3.-(.67) 40'7 96 0 .19
-.00004(.00005) -.0001(.0001) . 0 0 0 1 ( . 0 0 0 1 ) .0005(.0004) -.0002--(.0001) - . 0 0 0 1 ( . 0 0 0 1 ) 3-','(.70) .156r(2.74'7) 421 96 106 36.83 0 0 .22 .34
-.0001(.0001) .0005(.0004) -.0001(.0001) .1561(2.'74'7) 96 36.83 0 .34
equation.In sum, the resultsmeet expectationsset by the P.ACcontributionliterature,particularly when viewed within the context of the politics pushing for new cable regulation in 1992. 4.2.5.2. Secondstage:factors influencingcablereregulationvotes. Theresultlsof the second stageequationsfor the conferenceand override votes in the House and Senateare reported in Table 4. We expectedthat differencesin the timing of elections,the nature of constituencres represented, andrelianceupon PAC financingbetweenthe Houseand the Senatewould result in PAC influencebeing greaterin the House.This is alsomore likely to occur on the conference votesthan the veto overrideones,becausewe have hypothe,sized that PAC influenceshould dissipate(if not evaporate)oncean issuehasrisento the macropoliticallevel.Furthermore,we are more likely to detect significant PAC effects from thoselhat were on the urinning side, in this casethe broadcastingPACs.By definition,PACson the losing sideshouldhavelesssway with legislatorsthan PACson the winning side. Our first hypothesisconcernschamberdifferences.We hypothesizedthat PAC influence should be greater in the House than the Senatebecausec,f differences in election timing and constituencysize and characteristics. This is what we find when we compareHouse and Senatevoting on the conferencebill. (As already noted, the veto pulled the issue from the policy subsystemto the macropoliticalarena,which shouldalso weakenPACIeffects.Thus, comparisonof PAC effectsacrosschambersfor the veto overr.Lde vote is not asclleara testof the chamberdifferenceshypothesis.)Both broadcastingand Hollywood PAC contributionsaffect voting pattemsin the Housewhile neitherhavea significantr:ffectin the Senate.Oddly, cable television PAC effects are not significant in either chamber.This may signal, aswe speculated, that the policy debatehad moved from a cable television dorninatedsubsysternto one where broadcastingPACs were more critical. The analysisof changesin voting probabilitiesthat correspondto changesin theoretically important independentvariables provides a more substantivesenseof their impact on the
J.E. Cohen,J.A. Hamman/The Social ScienceJountal 40 (2003) 357-36:?
roll-call votes.The other variablesare setto thefumeanvaluLewhile different valuesof the PAC variables of interest are substitutedinto the equation.Each,additional $1,000 of broadcasting PAC contributions increasessupport for the reregulation conferencebill bll .045. However, most members of the House receivedno broadcastingcontribution. Those that did tended to acquirelargesums,the highestbeing $25,000.Thus,a $10,000contributionincreasesthe odds of voting for the conferencebill by 45Va.Modestcontribution levels convertil fence-sitterinto a supporter,and somewhatgreatersumsmight actually convert an opponentinto a supporter. industry pulled This is consistentwith our explanationthat the mobilization of the broadcastirLg the reregulationpolicy out of the cable television dominaled subgovernmentand into a new one where broadcastinginterestsheld greaterinfluence. But Hollywood PAC contributionsare also consequential,althoughnot as sfong as the sulpportfor the conEach$1,000contributionfrom a Hollywood PAC decreases broadcasters. ferencebill by 3.5Vo.The Hollywood interestsfunded many more memb,ersof the House than the broadcastingindustry, but at lower levels. Takinlginto account their greater spread of money acrossmembers,their averagecontribution warsone-half that of the broadcasting industry.Their largestconffibutionsonly totaled $16,000.a full $9,000lessthan the broadcastingcontributionhighs.The combinationof lessmoneyper memberandthe corresponding the successof thebroadcasters overthe cabletelevision-Hollywood smallereffectunderscores alliance. PAC spendinghasno impact on Senatevotes,despitethe higher overall ratesof spendingby the cabletelevisionand Hollywood interests.Again we may seethe prescienceof the broadcastingstrategy.Broadcastinginterestsdid not spendas nLuchper Senatora.sRepresentative, despitespreadingits money acrossa greaterpercentageo.[ senators,Seeminglybroadcasting interestsunderstoodthe lesserimpact of PAC money on the Senatethan the House, and concentratedits money on a nufirowercohort of lower chambr:rmembersto greratereffect. Thus, thesefiguresprovide evidencein supportof our chamberdifferenceshypothesis,while also telling a story of PAC contributionstrategythat appearsplausible. level effects.Herewe expect Our secondhypothesisconcernssubgovernment versuss;ystem PAC influenceto recedeas the issueis pulled from the subgovemmentinto the macrolevel political arena.Furthermore,a subsidiary hypothesissug€ieststhat macrole.veleffects should also grow with this changein policy making arenas,in this casepresidentialeffects. We get our strongesttestof this hypothesisby concentratingon the two Filousevotes.First, whereas broadcastinginfluences are felt on the conference vote in the House, they erode to statisticallyinsignificant levels on the veto override ',/ote,which is consistentwith our hypothesis.However,the Hollywood PAC variableseemsto grow in importance,contraryto expectations.Although the logit coefficientremainsstableacrossboth votes,the probabilities suggesta doubling in impact for Hollywood PAC contributions on the override versus the conferencevote (from 3.5Vofor each $1,000 to'7.2Vofor each $1,000). Iitill, we must be cautious about over-interpretingthis effect, as it may be doing nothing more than reinforcing the presidential effects, which are pushing in the same direction. And even with the greater impact of the Hollywood contribution, it is not great enou;ih to sustainthe pr:esident'sveto. In fact, reregulationopponentslost 28 supportersfrom the conferenceto the overridevote, which gives us a senseof the pull of the macropolitical forces on membersof Congresseven in the face of such a seeminglypotent PAC contribution effect.
366
J.E. Cohen,J.A. Hamman/The Social ScienceJoumal 40 (2003)357-369
Our subsidiary hypothesis is that presidential effects should loom larger on the override vote than the conferencereport. This is what we find, although the effect is modest.First, the presidentialsupportcoefficientincreasesfrom -.04 to -.05, while its standarderror subsides. The probability effects tell a similar story. When we hold all other variables to their mean, we find that an increase of 10 percentagepoints in support of the president translatesinto aTEo grealerlikelihood of opposingreregulationforthe conferencevote. The sameshift in presidential supporttranslatesinto an 8 percentagepoint probability shift in roll-call voting. because Such a shift from the conferenceto the override vote is not statistically signrLficant, the impact of presidential support is already so strong, the rnost important of any variable in either equation.That we detect any increasein its effect from the conferenceto the override vote is telling of the potentialimpact of pulling an issueout of a policy subsystemand into the macropoliticalarena.
5. Conclusion The literatureon PAC influenceover legislativedecisionrnakingis in a theoreticalmorass, with findings acrossthe entire spectrumfrom strong to no effects.Most scholairsnow take the position that PAC effectsare not uniform, but will vary with certainvariables,contexts,and conditions.A largenumberof suchinterveningvariableshavebeenoffered,but Smith (1995) concludeshis review and critique of the literatureon interestgroup influenceby noting a need to incorporate better counterfactualsin future researchin or:derto isolate the effects of such is difficult to employ. variables.However,sucha researchstrategy,as Smith alsonLotes, In our paper,we follow Smith's leadby taking advantageof a naturalexperimentto testtwo versusmacropolity hypothesizedcontextualeffects:HouseversusSenateand suLbgovernment presidential veto that on PAC influence. We were able to take advantageof a rare event-a Congresstned to override.Thus, we were ableto hold PAC contributionsand the wording of the bill constantto test for the impact of thesetwo contextual effects. In gen,:ral,our results supportthe hypothesisthat PAC influenceswill be greaterin the House than irr the Senateand that they will be strongerat the subgovernmentthan macropolitical level. We also detectedlittle congressionalresponsiveness to constituentdemanrls,with neither constituent variable attaining statistical signiflcance in any estimation. This lack of impact correspondswell to the story of cable regulation told herr:. Despite rising cable rates and consumercomplaints to policy makers, it was not until bro:rdcastersmobilized againstcable intereststo securerebroadcasting feesthat cableinterestslost their hold on cablepolicy.While theoriesof subgovernments suggestthat constituentforces should not be verryinfluential in those arenas,theoriesof macrolevelpolicy making, in contrast,do suggestthat constituent factorswill be potentat thatpeaklevel.But they arenot in our estimation,we argue,becauseby thetime cablegot to themacrolevelwith thepresident'sveto,(lablepolicy hadbeentransformed into a test of the presidency.Instead of preferencesabout cable policy determining public attitudes,partisanshipand presidentialapprovaltook over to structurepeople's orientationson the debate. Lastly,we needto stressthat our findingsmay not generah'ze to otherpolicy areasandissues, a problem that Baumgartnerand Leech (1999) exhort is endermicto studieslikr: ours that focus
J.E. Cohen,J.A. Hamman/The Social ScienceJountal 40 (2003) 357-36t)
36'7
on one policy or issue at a time. Still, we think that designslike ours, that hold constantPAC contribution and policy variables,aliow the strongesttestspossibleof the interactionbetween PACs and other variableson lesislative decision makins.
Notes 1. For instance,Chappell (1981), Grenzke (1989), Rothenberg(.1992),Vesenka(1989), Wright (1985, 1990) find no PAC impact on congressionalvoting. However,Coughlin 'Wayman(1990), (1985), Fleisher (1993), Frendreisand Waterman(1985), Hall and Langbein (1993), Langbein and Lotwis (1990), Neustadtl (1990), QuLinnand Shapiro (1991),Saltzman(1987),Schroedel(1986),Silbermanand Durden (1976),Stratmann (1991) find supportfor PAC impact on congressionalvoting. The following studiesreport marginalor mixedresults:Chappell(1982),Durd,;n,Shogren,andSilberman(1991), Evans(1986),Feldsteinand Melnick (1984),Kau, Keenan,and Rubin (7982),Kau and Rubin (1982),McArthur and Marks (1988),Peltzman(1984),Wayman (1985),Welch
(r98D. 2. The list of conditionsthat might affect PAC influen,:eis quite long. SieeSmith (1995, pp.9a-95). 3. Some scholarsuse the phrasesubsystem.We use subgovernmentand subsysteminterchangeably. 4. Part of the reasonfor the rise in dish ownershipwas to side-stepthe increasingcable costsandrates. 5. The overridevote in the House was 308-114. From the conferenceto the override,the Republicanvote in the House went from 71-90 to 7',t-85 and the Democratswent from 208-38 to230-29. 6. Thosenot voting are treatedas missingdata. districtsand stateswere 7. Basicmonthly fee andmarketpenetrationdatafor congressional Inc. The license licensedto the authorsfor researchpurposesonly by'WarrenPublishing;, was paid for by a grant from the Office of ResearchDevelopmentand ,Administrationat SouthernIllinois University. 8. BootstrappingappliesMonte Carlo samplingto empirically estimatethe samplingdistributionofastatistic.Thisisdonebytakingalargenumberofrandomsampleswith replacementand re-estimating the equation many trimes.The resulting distribution of parametersis used to determine probable coefficients and their standarderrors. Bootstrappingthe standarderrors are efficient estimatesof the standarderrors of the second stagecoefficients. References Arnold, R. D. (1990).The logic of congressionalaction.New Haven:\'ale UniversityPress. Baker,R. K. (1989).House and Senale.New York: Norton. Baumgartner,F. R., & Jones,B. D. (1993).Agendasand instabilih, in Americanpolitics. Chicago:University of ChicaqoPress.
368
J.E. Cohen,J.A. Hamman/The Social ScienceJournal 40 (2003) 357-369
Baumgartner, F. R., & Leech, B. L. (1999). Basic interests: The importdnce of groups in politics and political science.Princeton, NJ: Princeton University Press. Chappell, H. W., Jr. (1981). Campaign contribution and voting on the cargo preferencesbill: A comparison of simultaneousmodels. P ublic Choice, 3 6, 3OI-3 12, Chappell, H. W., Jr. (1982). Campaign contributions and congressionalvoting: A simultaneousprobirtobit model. Review of Economics and Statistics, 64, 7'7-83. CongressionalQuarterly.(1987).Satellite-dishowners.In CQ Almanac (pp. 354-355).Washington,DC: CQ Press. CongressionalQuarterly. (1992). Hill enactscable TV iaw over veto. In CQ Almanac (pp. 11l-179). Washington, DC: CQ Press. Coughlin, C. G. ( 1985).Domestic contentlegislation: Housevoting and the economictheory of legulaion. Econctmtc I nquir1,,23, 43'l-448. Durden, G. C.. Shogren,J., & Silberman,J. (1991). The effectsof interestgroup pressureon coal strip-mining legislation. Soc i al Sci ence Quar ter ly, 72, 23'7-250 Evans, D. M. (1986). Contributionsand roll-call voting: Conditional power. ln A. Cigler & B. Loomis (Eds.), Interest group politics (2nd ed., pp. 114-132). Washington,DC: CQ Press. Feldstein,P. J., & Melnick, G. (19S4).Congressionalvoting behavior on hospital iegislation: An exploratory study. Journal of Health PoliticsPolicy and Law, 8. 686-701. Fleisher,R. (1993). PAC contributions and congressionaivoting on nationa.ldefense.Lcgislative StudiesQuarterly, 18.391409. Frendreis. J. P., & Waterman, R. (1985). PAC contributions and legislative behavior: Senatevoting on trucking delegulation. Social Science Quarterly, 66, 401412. Garay, R. {198U. Cable television: A referenceguide to infbrmation. New York: GreenwoodPress. Grenzke,J. (1989).Candidateattributesand PAC conffibutions.WesternI'olitical QuarterLy,42,245-264. Grier, K., & Munger. M. (1986). The irnpact of legislatorattributeson interestgroup campaigncontributions. J ottrnal of Labor Research,7, 349-361. Hall, R.. & Wayman,F. (1990). Buying time: Moneyed interestsand the mobilization of bias in congressional committees.Amer i can PoIit i cal Sci ence Revi ev,, 84,'797-820. In A. King (Ed.),The New ,AmericanpoLitical Heclo, H. (1979). lssue networksand the executiveestablishment. system(pp. 87-I24). Washington,DC: AmericanEnterpriseInstitute. Kau, J. 8., Keenan,D., & Rubin, P. H. (1982). A generalequilibrium model of conglessionalvoting. QuarterLy Jounnl of Economics, 97, 2'71-293. Determinantsof roll call vorirtg constituents,and crsntritbutions: Kau, J. B., & Rubin, P H. (1982). Congressman, in the House of Representatives.Boston: Martinus Nijhotr. and eiections:A policy anall,sis.American Political ScienceReview, Kuklinski, J. H. (1978).Representativeness 72, l6s-17'7. PublicChoice,75,254-271. Langbein.L.(1993).PACs.lobbies,andpoliticalconflict:Thecaseofguncontrol. Langbein,L.. & Lotwis, M. (1990).The poiitical efficacyof Iobbyinganclmoney: Gun control in the U.S. House. Legislative StudiesQuarterb, I 5, 413440. Lee,F.E..&Oppenheimer,B.L (1999).SizinguptheSenate:Theunequolconsequencesofequalrepresentarion. Chicago: University of Chicago Press. McArthur',J., & Marks, S. V. (1988).Constituentinterestvs. legislatorideology: The role of political opportunity cost.Economic Inquin,, 26, 46147 O. the channel.CongressionalQuarterll, Mills, M. (1992,October).Bush asksfor a sign of loyalty: Congresschang,es Weekly,October 19,1992.p.3141 . Mooney, C.Z. (1996). Bootstrapstatisticalinference:Examplesand evaluationsforpolitical science.American Journal of Political Science.40, 570-602. Mooney,C.2., & Duval, R. D. (1993).Bootstrapping:A nonparametricapproachto statisticalinference.Newbury Park, CA: Sage. Neustadtl, A. (1990). lnterest group PACmanship: An analysis of campaign contributions, issue visibility, and legislative impact. Social Forces, 69, 549-564. Peltzman. S. (1984). Constituent interest and congressional voting. Jor,trnal of Lau, and Econontics, 27, I8l-
210.
J.IE.Cohen. J.A. Hamman/The Social ScienceJoumal 40 (2003) 357-369
369
Quiun, D. P., & Shap[ro,R. Y. (1991). Businesspolitical power: The caseof taxation. American Political Science Review.85.851-874. Redford, E. S. (1969). Dernocracy in the administrative state. New York: Oxford University Press. Rothenberg, L. S. (1P92). Linking citizens to govemment: Interest group politics at common caase. New York: Cambridge Univefsity Press. Saltzman, G. M. (19$7). Congressionalvoting on labor issues:The role of PACs. Industrial and lnbor Relations Review.40. 163-179. Schrpedel,J. R. (198q).Campaigncontributions andlegislative outcorosWestern Political Quarterly,39,37l-389. Silborman, J., & Dufden, G. (1976). Determining legislative preferenceson the minimum wage: An economic approach.Journa( of Political Economy, 84, 317-329 . Smith, R. A. (1995). [nterest group influence in the U.S. congress.Legislative StudiesQuarterly,20,89-L39. Stratmann,T. (1991).What do campaigncontributionsbuy? Deciphering causaleffectsof money andvotes.Southern Ec onomic JournaN,57, 606-620. Tobin, J. (1958). EstiFnationofrelationships foriimited dependentvariables.Econometrica,26,24-36. Vesenka,M. H. (198t). Economic interestsand ideological conviction: A note onPACs and agricultural acts. Journal of Economic BehQvior and Organization, 12,259-263. Wayman, F. (1985). lArms control and sfiategic arms voting in the U.S. Senate:Patternsof change, 1967-1983. Joumal of ConfliQtResolution, 29, 225-251. Welch, W P. (1982). pampaign contributions and legislative voting: Milk money and dairy price supports.Western Political Quarte rlg,,35, 4'78495. West, D. M., & Loorifs, B. A. (1999). The sound of money: How political interestsget what they ltdnt. New York: Norton. Wright, J. R. (1985). PACs,contributions, and roll calls: An organizationalperspective.Areerican Political Sctence Review,79,4OO4l4. Wright, J. R. (1996)f Interest groups and Congress: Lobbying, contributions, and infiuence. Boston: Allyn and Bacon.
