Are Term Limits Undemocratic

Are Term Limits Undemocratic?
Author(s): Einer Elhauge

Source: The University of Chicago Law Review, Vol. 64, No. 1 (Winter, 1997), pp. 83-201
Published by: The University of Chicago Law Review
Stable URL: http://www.jstor.org/stable/1600199 .
Accessed: 07/10/2013 05:51
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
The University of Chicago Law Review is collaborating with JSTOR to digitize, preserve and extend access to
The University of Chicago Law Review.
http://www.jstor.org
This content downloaded from 153.90.89.164 on Mon, 7 Oct 2013 05:51:49 AM

All use subject to JSTOR Terms and Conditions
ARTICLES
Are Term Limits Undemocratic?

EinerElhauget
INTRODUCTION:THE FALSE PARADOXOF TERM LIMITS
The Supreme Court's decision in US Term Limits, Inc v

Thornton'resolvedthe law regardingfederaltermlimitsbut not
theirdesirability.
The decisionmakes thismuchclear: the QualificationsClauses2 prohibitboth Congress and the states from
limitingthe termsof federallegislators.Constitutionalamendmentis the onlymeans leftopen.3But the questionthat remains
largelyunilluminatedis whethertermlimitsare sound as a matterofpoliticaleconomyor democraticprinciple.
That question remains practicallyimportantto more than
just scholars.The QualificationsClauses do not apply to state
legislatorsand governors,but many lawsuits claim that a fundamentalconflictwith democraticprinciplesmakes termlimits
on state officialsunconstitutional
under otherclauses.4Leaving
ofLaw, HarvardLaw School.I am gratefulforhelpfulcommentson prior
t Professor
draftsor outlinesof this workfromLucian Bebchuk,Jim Blumstein,RichardFallon,
BarryFriedman,HowellJackson,Beth Garrett,Paul Gewirtz,Louis Kaplow,Ken Klee,
JohnLott,RichardManning,Dan Meltzer,EugeneMills,MarthaMinow,Paul Mishkin,
HenryMonaghan,Richard Parker,Nelson Polsby,Bob Rasmussen,RichardRevesz,
Steve Shavell,Anne-MarieSlaughter,David Strauss, Peter Strauss, Steve Sugarman,
Alex Tabarrok,LaurenceTribe,MartinWattenberg,
David Wilkins,and otherparticipants at workshopsat the HarvardLaw School Colloquium,the Harvard Law & EconomicsWorkshop,
and theVanderbiltLaw SchoolWorkshop.
115 S Ct 1842(1995).
2 US Const,
ArtI, ? 2, cl 2; US Const,ArtI, ? 3, cl 3.
US Term Limits, 115 S Ct at 1845,1871.
See, forexample,Nevada JudgesAssn v Lau, 112 Nev 51, 910 P2d 898, 900 (1996)
(Equal Protection
and Due Process);Bates v Jones,904 F Supp 1080,1091 (N D Cal 1995)
83

84
ofChicagoLaw Review
The University
[64:83
aside any legal issue,we need to knowwhetheras a policymatter states shouldcontinueto press fortermlimitsor rescindthe
forexecutiveand
ones theyhave and whetherthe answerdiffers
legislativetermlimits.Andothernationsremainconstitutionally
Shouldtheyadopttermlimitsand does the answer
unrestricted.
differdependingon the votingsystemand on whetherthe limit
or to federato nationalgovernment
applies to representatives
Noris constitutional
tionsliketheEuropeanCommunity?
change
beyondquestionfortheUnitedStates. Shouldwe pressfora constitutionalamendmentallowingcongressionaltermlimitsor for
one repealingthelimiton presidentialterms?ShouldPresidents
make approvalof termlimitsa litmustest forfutureSupreme
And thereare questionsofinterestto scholCourtnominations?
ars even if nothingpracticalturnson them.We want to know
whetherUS TermLimitswas correctly
decided,and whetheran
ideal politicalprocesswouldincludetermlimits,howeverimpossibletheiradoptionmaybe.
To say that the Courtleftthe meritsof termlimitslargely
unilluminatedis not to say it did not adopt a strongpremise
about them.It did: termlimitsstruckthe Courtas patentlyundemocratic.As Part I discusses,the Court invokedtraditional
tools oftext,framers'intent,history,and precedent,but it conin thosematerials-an ambiguityit resolvedat
ceded ambiguity
everyturnwiththe assertionthattermlimitsare undemocratic.
This assertionborrowedpartlyfromstatementsby our framers
objectingthata legislativepowerto restrictwhocouldrunforoffice could produce elitist exclusions and legislative selfperpetuation.These are convincingobjectionsto requirements
or membersof
thatcandidatesbe property
owners,professionals,
themajorityreligionor party,whichwerethekindofrestrictions
the framersmainlyconsidered.But the last thingone can say
abouttermlimitsis thattheyhelpan eliteor legislativemajority
to perpetuateitselfin office.No, what reallyseemedto motivate
the Courtwas a broaderpremise:thatdemocraticprinciplesgive
votersan unfettered
rightto voteforwhomevertheyplease.
This premisewas hardlyunique to the Court.The conventional wisdomamongmost academics and leading newspapers
was that, whateverthe constitutionalambiguity,term limits
wereclearlya bad idea.5Doubtlessthispartlyreflected
theelite's
(First Amendmentand Equal Protection).
5 See, for example, Nelson W. Polsby, Constitutional Mischief: What's Wrong With
Term Limitations,Am Prospect 40, 41 (Summer 1991); Gary S. Becker, ReformingCongress: WhyLimiting Terms Won'tWork,Bus Wk 18 (Aug 6, 1990); Kathleen M. Sullivan,
Dueling Sovereignties:U.S. Term Limits, Inc. v. Thornton,109 Harv L Rev 78, 109 (1995);

1997]
AreTermLimitsUndemocratic?
85
distasteforwhat it perceivedas a populistknow-nothing

movementbased on the ridiculouspremisethat experienceand expertise worsenedlegislativejudgment.But the more fundamental
and largelyunchallengedobjectionwas that term limits were
undemocratic
because theypreventedvotersfromfreelyexercisingtheirjudgmentto retainexperiencedrepresentatives
ifvoters
wished. Term limits furtherseemed to have no redeeming
prodemocratic
virtue;if a majorityof voterswanted to replace
experiencedincumbentswithnewcomers,theycould do so without termlimits.Justvotethebumsout.
Indeed,no accountoftermlimitscan be satisfactory
unless
it explains an apparent paradox: why do the same voterswho
fortermlimitsalso vote overwhelmingly
vote overwhelmingly
to
returntheirseniorincumbentsto office?At the same time that
termlimitswerebeingpassed in twenty-four
states (with70 percentsupportin some states),6voterswere generallyreturning90
percentof state incumbentsand up to 98 percentof federalincumbentswho ran forreelection,the bulk ofwhomexceededthe
termlimitsthosevotershad just approved.7To many,the simultaneous votes forseniorincumbentsand termlimitsthat would
oust themare inconsistent,
suggestingthat votesfortermlimits
reflectedlittlemorethan an irrationaltempertantrumat odds
notonlywithdemocraticprinciplesbut withvoters'ownrevealed
preferences.
The paradoxdisappears,however,oncewe realize that there
is no inconsistency
betweenthe twovotes.As Part II shows,voters whowishto oust incumbentsfacea collectiveactionproblem.
Incumbentsby definitionhave more senioritythan challengers,
and this senioritygivesthemmorelegislativeclout.Anyindividual districtthat ousts its incumbentis thus penalized by a
smaller share of legislative power and governmentalbenefits
unless the otherdistrictsalso oust theirincumbents.This penalty growswith the seniorityof the incumbentand can coerce
BruceE. Cain, The VaryingImpactofLegislativeTermLimits,in BernardGrofman,
ed,
LegislativeTermLimits:Public ChoicePerspectives
21, 21 (Kiuwer1996) (Most political
scientistsopposetermlimits.);David J. Olson,TermLimitsFail in Washington:
The 1991
in GeraldBenjaminand MichaelJ.Malbin,eds,LimitingLegislativeTerms
Battleground,
statenewspapersopposedtermlimits.).
65, 79, 89 (CQ 1992)(Virtuallyall Washington
6 The twenty-four
includethetwenty-three
listedin ScaringMississippiVoters,Wall
St J A18 (Oct 30, 1995),and New Hampshire,see NH Rev Stat Ann ? 653:3-4(Michie
1996).
7 See US TermLimits,115 S Ct at 1912 (Thomasdissenting);
GeraldBenjaminand
MichaelJ. Malbin,TermLimitsforLawmakers:How to StartThinkingAbouta Proposal
in Process,in Benjaminand Malbin,eds, LimitingLegislativeTermstable B-2 at 293
(citedin note5); AndrewR. Dick and JohnR. Lott,Jr.,Reconcilingvoters'behaviorwith
legislativetermlimits,50 J Pub Econ 1, 2 & n 3 (1993).

86
The University
ofChicagoLaw Review
[64:83
whoseideologicalviewsthe indidistrictsto voteforincumbents

vidual districtsdisfavorand who deliverporkthat on balance
harms all districtsas a group.Each districtmightthus prefer
oustingits seniorlegislatorifotherswoulddo the same because
and porklevels without
thatwouldreduceideologicaldivergence
a smallershareofgovernmental
powerand
anydistrictsuffering
benefits.But because each individualdistrict'svote on whether
to reelectan incumbentdoes not affectwhetherotherdistricts
reelecttheirs,each districtrationallyreelectsits own seniorlegpowerand benefits.
islatorto geta greatershareofgovernmental
The districtswillthuscontinueto reelectseniorincumbentseven
iftheywouldbe betteroffcollectively
oustingthemall. By oustlevel and reducingthe
ing all legislatorsovera certainseniority
penaltyforreplacingincumbentsbelow that level, termlimits
can bothdecreaseideologicalslack and lessen overallpork.The
them.
but to further
preferences
resultis notto fetterdemocratic
Withtermlimits,each districtgets bothideologicalrepresentationcloserto its viewsand a loweroverallamountofporkit prefers.
In addition,evenifhappywiththeideologyofits representative,each districtmay desireto get rid ofthe seniorrepresentatives fromotherdistricts.Representativesfromother districts
cannotbe reachedthroughordinaryvotingbut can be reached
throughtermlimits.If the ideologicaland otherpoliticalbenefit
is less than
each districtderivesfromits seniorrepresentative
the politicalharmit suffersfromotherdistricts'seniorrepresentatives,each districtcan be made betteroffby agreeingto give
up its seniorrepresentativeif the otherdistrictswill give up
theirs.This goal also presentsa collectiveactionproblem:no districthas incentivesto individually
oust its seniorrepresentative
unless the othersoust theirs,but each districtcan be made better offif all will oust theirseniorrepresentativescollectively.
thoughwith a signifiLikewise,the goal seems prodemocratic,
cant difference:
each district'sindividual representativemay
conform
but the agless to the district'sdemocratic
preferences,
comescloserto each district's
gregatelegislativerepresentation
preferences.
a majorityof the
However,sometimesdistrictsconstituting
politicaljurisdictionmaywishto oust the seniorrepresentatives
ofotherdistrictseven thoughthoseotherdistrictsare not made
betteroffby the bargain.This presentsless a collectiveaction
problemthan a problemofthe majorityimposingits will on the
such majoritypreferminority.
But, I will argue thatfurthering
ences throughtermlimitsis also prodemocratic,
especiallysince

1997]
87
the underlyingmotivationis generallycorrectingthe externalities the minority's

representatives
imposeon the majority.8
The modeldevelopedin Part II does morethanjustifylegsthe conditionssupportingthese
lative termlimits.By specifying
justifications,it offerspredictionsand explanationsabout when
and wheretermlimitsare likelyto be popular.For example,as
the foregoingsuggests,districtsdiffer.Even thoughsometimes
all districtsmay favor term limits, they will not always be
unanimous.The modelexplainswhichdistrictsare likelyto favor
and oppose termlimits.It furtherexplains in whichpoliticaljurisdictionsand historicalperiodstermlimitsare mostlikelyto be
popular.It explainswhytermlimitsare morepopularin legislathan in thosewithproportional
tureswithdistrictrepresentation
representation.It predictsthe formsterm limits are likely to
take when imposedby states on theirfederalofficials.And it illuminatesa seemingconundrum-whydespitethe seemingfacial
neutralityof legislativeterm limits, conservativeshave been
in favorof them and liberal leaders have been
more uniformly
spliton theirdesirability.
But the analysisin Part II holds onlyforpoliticalrepresentatives elected to a collectivebody throughdistrict-by-district
voting.What explainsthe moreestablishedpracticeofimposing
term limits for positionslike President,governor,or mayor,9
wherethe entirepoliticaljurisdictionvotes on each seat? Collective actionproblemscannotjustifysuch termlimits.Instead, as
Part III shows,thejustificationis that votersmay not be seeing
the alternativestheypreferon the ballot because ofentrybarriers. Long-termincumbentshave brandname advantages that
createhighbarriersto politicalentry.By reducingthesebarriers,
in legislativeand executive
termlimitscan increasecompetition
races. The additionalentrymay presentnew candidatesthe districtspreferto incumbentsbut did not see on the ballotwhileentrybarrierswere high.More subtly,even if votersdo not prefer
the new candidatesto the incumbents,greatercompetitioncan
8
See Part IV.A.2.
The United States Constitutionimposes term limitson the President,see US

Const,AmendXXII,? 1, and manyothercountriesdo the same fortheirchiefexecutives.
See James D. Adams and LawrenceW. Kenny,OptimalTenureof ElectedPublic Officials, 29 J L & Econ 303,314 & n 21 (1986). Most statesimposetermlimitson theirgovernors.See Beth Garrett,TermLimitationsand theMythof the Citizen-Legislator,
81
CornellL Rev623,630 n 15 (1996) (forty
states);F. Paul Calamita,Comment,
Solvingthe
Voters'Dilemma:The Case forLegislativeTerm-Limitation,
8 J L & Pol 559, 589 n 156
(1992) (collectingcitationsforthirtystates).And 35 percentofmajorcitiesimposeterm
limitson theirofficials.Mark P. Petracca and Kareen Moore O'Brien,The Experience
withMunicipalTermLimitsin OrangeCounty,California,in Grofman,
ed, Legislative
TermLimits289,289 (citedin note5); Calamita,Comment,8 J L & Pol at 561.

88
ofChicagoLaw Review
The University
[64:83
broadenand hone the issues debated and move incumbentsto

positionscloser to those held by their electorates.Again, this
analysis helps us understandnot onlythe motivationforterm
limitsbut whenand forwhichofficestheyare likelyto seem attractive.
Nonetheless,the alleged paradox does dispose of many
popularargumentsfortermlimits,includingclaimsthatcitizenlegislatorsare betterthan career politicians.Many argue that
longlegislativestintsmakelegislatorscorrupt,arrogant,cynical,
to special interest
resistantto reform,
sympathetic
unprincipled,
with theirelectorate.'0This would be
groups,and out-of-touch
were like Cincinnatus,leaving their
avoided if representatives
in the legislatureand then replows to serve onlytemporarily
turningto theirfarms.Muchinkhas been spiltdebatingwhether
citizenterm limits would produce such noble, fresh-minded,
easily manipulatedamateurs."
legislatorsor ratheruninformed,
The ultimate question, stated in many differentguises, is
legislativeexperienceon balance
whether,at somepoint,further
That seems impossibleto anbeginsto worsendecisionmaking.
in the abstract.And some have persuasively
swer satisfactorily
argued that term limits might not even produce citizenlegislatorsbut rather politicalcareeristswho move fromone
But the deeper problemis
term-limited
positionto another.12
if
that,even the proponentsof citizenlegislatorsare absolutely
10These werethe arguments
debate aboutrotation,
used duringthe constitutional
see US TermLimits,115S Ct at 1859& n 23; PartI.A.3,and bymostcurrentsupporters
US TermLimits(Nos 93oftermlimits.See, forexample,Briefforthe State Petitioner,
1456,93-1828),1994WL 444683at *4 (quotingArkConst,Amend73, preamble);George
F. Will, Restoration: Congress, Term Limits and the Recoveryof Deliberative Democracy
40-41(Free Press 1992);MarkP. Petracca,Rotation in Office:The Historyof an Idea, in
Benjaminand Malbin,eds,LimitingLegislativeTerms19,20-28(citedin note5); Garrett,

prove
81 CornellL Rev623 (citedin note9) (collecting
sources).Becausethesearguments
thatboththeconstito be unpersuasive
justifications
fortermlimits,it is notsurprising
termlimits.
tutionalframers
and thecurrentSupremeCourthad littledifficulty
rejecting
" Compare,forexample,Polsby,AmProspectat 40-42(citedin note5) (Termlimits
will produceuninformed
amateurswho willbe disproportionately
old and richpersons
needingno career.),withWill,Restoration
at 9-10(citedin note10) (The careerismassociatedwithlonglegislativetenuresproducesbad legislativedecisions.).
12 See Garrett,
81 CornellL Rev at 656-57(citedin note9); GeraldBenjaminand Michael J. Malbin, Term Limits forLawmakers: How to Start Thinkingabout a Proposal in
Process, in Benjamin and Malbin, eds, Limiting Legislative Terms 3, 14-15 (cited in note
5). Noris theregoodevidencethatthebackgrounds

oflegislatorswouldchangewithhigh
turnover.See Alexander Tabarrok, A Survey, Critique,and New Defense of Term Limits,
14 Cato J 333,335-36(1994) (notingthatpercentageoflawyersin Congresswas just as

remain
highor higherwhenturnover
was higherand thatmostlegislatorcharacteristics
stabledespitevaryingtenurelevels);Linda L. Fowler,A Comment
on Competition
and
Careers, in Benjamin and Malbin, eds, LimitingLegislative Terms 181, 182 (cited in note
5) (same).

1997]
89
rightthat legislativeexperienceis bad and reducedby termlimits, that still does not explain why we need termlimits.If the
onlyproblemwiththe currentsystemwere that long legislative
stints make incumbentsunattractiverepresentatives,voters
couldjust vote against them.To explain and justifytermlimits,
we need rationalesthatcannot be accomplishedby ordinaryvoting.'3
The collectiveactionand entrybarrierproblemsprovideboth

that explanationand tellingreasons to findtermlimits prodemocratic.Part IV shows that counterpoints
exist,but thereare
certainlyrationalgroundsforvotersto thinkthat termlimitson
balance enhance democraticrepresentation.Given the theoretical uncertainty,
democraticprinciplesseem to mandate respect
forthejudgmentofa votermajoritychoosingto adopt termlimits. Indeed,since onlytheentrybarrierproblemapplies to executives and both problemsapply to legislators,'4the argumentfor
termlimitsis strongerforCongressthan forthe President-an
ironysince the Constitutionnow prohibitstermlimitsforCongressand mandatesthemforPresidents.
This Articlewill proceed on the definitionalpremise that
more accuratelyreflectingthe preferencesof an electorateis
prodemocratic.
Some mightquarrelwiththis definitionor argue
accurate preferenceregistrationis
that, even if prodemocratic,
not necessarilydesirable.In particular,a Burkean mightargue
that ideological divergencebetween representativesand their
electoratesis more desirable or reflectiveof representativedemocracy.This is decidedlynotthe articulatedpremiseofthe Supreme Court or politicallyactive opponentsof term limits. Indeed, it is the supportersoftermlimitswho argue that theyare
It is not,however,surprising
to see such explanationsoffered
fortermlimits.The
flawsthat people see in incumbentlegislatorsare, afterall, what would motivatesupportersoftermlimitsto wantto replaceall incumbentsevenunderthe publicchoiceexplanationsI describe.Perhaps termlimitssupportersalso intuitedthe public choice
problemsarticulatedhere.The statedpurposeoftheCaliforniatermlimitsinitiativewas,
afterall, to "restore. . . competitive
to "encouragequalifiedcandidates"to run,
elections,"
and to limitthe "unfairincumbentadvantages"that resultedin the "extremely
high
numberofincumbents
whoare reelected."Cal Const,Art4, ? 1.5. Or perhapstheyknew
onlythatoustingnoxiouslegislatorsthroughordinaryvotingsomehowdid notworkand
decidedto trysomething
else. Whatevertheirsubjectivemotivation,
thejustification
for
its expressionthroughtermlimitsmustrest on groundsthat could not be furthered
by
ordinaryvoting.That is, eveniftheclaimedempiricalparadoxofthe simultaneousvotes
forseniorlegislatorsand termlimitscouldbe otherwiseexplained,we wouldstillhave to
facetheclaimed justificatory
paradoxthattermlimitscannotbe justifiedunlesstheycan
accomplishsomething
ordinary
votingcannot.
'4 Consistent
withthis,the rate of incumbentreelectionis generallyfarhigherfor
legislatorsthanforexecutivesin bothfederaland stategovernments.

The University
ofChicagoLaw Review
90
[64:83
morelikelyto produceBurkeanrepresentatives'5
and the opponents who complainthat termlimits interferewith electorate
I thus defersuch alternativedefinitions
preferences.
of democracyuntilPart IV, whichconcludesthateven a Burkeanconceptionofdemocracy
offersno sound groundsforcondemning
term
limitsas undemocratic.
But the questionsofwhat democracyis
and whetherit is desirableare perhapstoolargeand distracting
forfruitful
resolutionhere. I thus preferto focuson the more
concreteconclusionthat termlimitscan enhancethe accuracy
withwhichthe electorate'sviews are represented.That conclusion seemsofinterestno matterhow one definesdemocracyand
the socialgood.Andin termsofthe constitutional
question,more
voterpreferences
accuratelyregistering
surelyseemsa permissible goal fortheelectorateto further
byenactingtermlimits.
I.
THE CONSTITUTIONAL QUESTION
in theLegal Materials
A. The Ambiguity
The legal questionin US TermLimits was fairlystraightforward.Do the Constitution's
QualificationsClauses, whichrequirethatall federallegislatorsbe ofa minimumage, citizensfor
some years,and state residents,implicitlyprohibitstates from
imposingany additionalqualificationson runningforCongress?
The evidencefromthe text,framers'intent,history,and precedentwas surprisingly
ambiguous.
1. Constitutional
text.
The text forRepresentativessays simplythat "No Person
whoshall nothave attainedto theAge
shall be a Representative
of twentyfiveYears, and been seven Years a Citizen of the
UnitedStates,and whoshall not,whenelected,be an Inhabitant
ofthatState in whichhe shall be chosen."'6The textforSenators
is thesame exceptthattheminimumage is thirtyand theperiod
of citizenshipis nine years.'7The text does not say that states
may imposeadditionalqualifications.It does not say theymay
not.It is simplysilenton thesubject.
Nonetheless,the majoritycited the canon of construction
that expressiounius exclusioalteriusest (the expressionof one
thingexcludesothers),reasoningthatthe listingofsomequalifi-
See Part IV.A.3.

US Const, Art I, ? 2, cl 2.
17 US Const, Art
I, ? 3, cl 3.
5
16

1997]
91
cations excludes any others.18However,this canon, like many

Indeed, two
canons,has an equal and oppositecounter-canon."9
counter-canons:
(1) thatcourtsshouldinterpretambiguityto furtherthe purposeembodiedin the includedlanguage,whichoften
involvesextendingit by analogyto unincludedsituations;and (2)
that where lawmakers have shown theyknow how to exclude
theirfailureto do so impliesan intentto
otherthingsexplicitly,
is that the failureto
allow inclusion.The underlyingdifficulty
list otherthings may reflectsimple inadvertence,a failureto
considerthose otherthings,or an inabilityto reach a consensus
in favor of them rather than a consideredconsensus against
them.That is particularlytrue here,wherea failureto reach a
consensuson otherminimalqualificationsthat mustbe required
forofflcehardlysuggestsa consensusthatno otherqualifications
mightreasonablybe foundor added.
But there were more particularproblemsin the case at
hand. Even if one applied the expressiounius canon exclusively,
the canon couldbe interpretedto mean that the inclusionofcertain qualificationsexcludedthe possibilitythat the Constitution
itselfimposes additional qualificationsbut not the possibility
thatCongress,let alone the states,maydo so. The text,afterall,
couldhave statedthat any personmeetingthe above threequalificationscould be elected;instead it stated onlythat any person
not meetingthe three qualificationscould not be elected. One
mightaccordinglyread the language's purpose as merelypreventingCongress or the states fromallowing the election of
someonewithlowerqualifications,such as a twenty-one-year-old
If the clause sets
naturalizedalien withone year ofcitizenship.20
a
not
a
then
the
unius
canon means
expressio
only floor,
ceiling,
onlythat the clause exhaustivelylists all qualificationsmaking
up thatfloor.
Worse,even by itselfthe expressiounius canon cuts both
ways,forthe Constitutionexplicitlyprohibitsthe impositionofa
This impliesthat under
religiousqualificationforfederalofflce.21
thecanonthereare no otherqualificationsCongressor the states
are forbiddento add. But the majorityreasoned instead that
variousotherprovisionspolicingstate involvementin aspects of
US Term Limits, 115 S Ct at 1850 n 9.
See Karl N. Llewellyn, Remarks on the TheoryofAppellate Decision and the Rules
or Canons about How Statutes are to be Construed, 3 Vand L Rev 395, 401-06 (1950).
' The example is not fanciful.Age and citizenship qualifications forstate officeat the
time were in fact lower than the constitutional minimum,oftenallowing one to become a
state legislator with as little as 21 years of age and one year of citizenship. See US Term
Limits, 115 S Ct at 1902 (Thomas dissenting).
21 US Const, Art VI, cl 3.
18
19

92
The University
ofChicagoLaw Review
[64:83
federalelectionsotherthan the qualificationsof candidatesevidenced a generalpurposeof restrictingstate interference

that
should be extendedby analogy.22
That is, the Court applied a
classic counter-canon
to the expressiounius canon. Had the expressio unius canon been applied here it would have produced
preciselythe oppositeconclusion:that the listingofthe various
provisionsrestricting
state involvement
in federalelectionsimno
pliedthattherewere otherrestrictions
on such stateinvolveon state powerto set qualifications
for
ment,such as restrictions
canonsin
federalcandidates.But the ironyof using conflicting
the same opinionwas apparentlymissed. Ultimately,the truth
remainsthatall thetextprovideson the topicat hand is silence,
fromsilence.
and it is hardto deriveanything
meaningful
2. Textualeditingdecisions.
The majorityfoundit significantthat the Constitutional
Convention
rejectedtwoproposalsto give Congressthe powerto
add qualifications.23
Further,the Conventionrejecteda formof
termlimits,a "rotation"provisionrequiringincumbentlegislatorsto relinquishtheirseats fora termbeforerunningagain.24
But it is always dangerousto reason fromunadoptededits.
Like unincludedtext,theymayindicatemerelya failureto reach
consensus in favor of the unadopted item, not a consensus
against it. To be sure,the proposalof the edit means the issue
was flagged,so thatits nonadoption
was notinadvertent.
But the
flaggingoftheissue cutsbothways,forifthe lawmakersdid not
includelanguage foreclosing
the unadoptedproposal,one could
arguablyinfera lack of consensusagainst it or maybeeven an
intentnotto foreclose
it.
Even if one views nonadoptionas an intendedban on the
proposal,that ban hardlydeterminesthe issue at hand. One
mightwell objectto givingfederallegislatorsan unlimitedpower
to impose propertyor otherqualificationswithoutopposinga
state'spowerto imposetermlimitson its federalrepresentatives.
The concernsthatqualifications
mightbe used foreliteor legism The Court noted clauses: (1) requiring voter qualifications to be the same forfederal and state office,(2) givingCongress the power to alter state time, place, and manner
regulations of federal elections,(3) having congressional salaries set by federal law, and
(4) allowing Congress to be the judge of the qualifications of its members. US Term
Limits, 115 S Ct at 1857-59.
v Id at 1849. Under one proposal, Congress could have added a propertyowner qualification;under the other,any qualificationit chose. Id.
21 See id at 1859 n 22, citingMax Farrand, ed, 1 The Records of the Federal Conventionof 1787 20, 217 (Yale 2d ed 1937).

1997]
93
lative self-perpetuation
are arguably less serious when states
are the sourceofthe qualifiratherthan the federalgovernment
cations.And those concernscertainlyare less serious when the
qualificationtakes the formof a bar on seniorincumbentsrunningforreelection.
Likewise,therefusalto enshrinerotationin the Constitution
need onlyreflectan oppositionto requiringa particularversion
of it forall time and all states. The particularversiondebated
was, afterall, quite draconian.It would have made all incumbentsineligibleforreelection,thus maximizinglast-periodproblems.25One mightwell oppose such a one-consecutive-term
limit
withoutopposingmoderntermlimitsproposals.
Or perhapsthe framerssimplywantedto allow futureelectoratesor individualstates someflexibility
on whetherto impose
rotationand whatpreciseformto choose.Maintainingsuch flexibilityseems like a wise decision,fortermlimitsare farmoreattractivenow than at the time of the constitutionalframing.
Turnoverrates were higherthen,26and of course the tenure of
initial candidatesforthe new Congresswas zero. Further,distributablegovernmental
benefitsper districtwerefarlowersince
government,
especiallythe federalgovernment,
spent and regulated farless. As subsequentanalysis shows,such lowerlevels of
prevailingseniorityand governmentalbenefitsdirectlylower
collectiveactionpressuresto vote forideologicallydivergentrepresentatives.27
Since such pressuresare largelywhat make legislativetermlimitsattractive,it is hardlysurprisingto findless
enthusiasmfortermlimitsthenthannow.
Moreover,the dissenthad an editingdecisionofits own.An
earlydraftofthe House ofRepresentativesQualificationsClause
specifiedthat"anypersonpossessingthese qualificationsmaybe
elected,"but this language was later deleted.28The deletionof
languagethat would have made the qualificationsexclusive,the
US Term Limits dissent argued, made clear the clause was
meantto set onlya floornot a ceiling.29
The majoritycountered
that this edit was equally consistentwiththe view that the deleted language was superfluous.30
True, but one could say the
same about all the omissions,rejections,and nonadoptionsthe
5 See Part IV.C.
' See Part II.C.2.
27
See id and Appendix.
US Term Limits, 115 S Ct at 1895 (Thomas dissenting), citing Farrand, ed, 2 Records at 139 (cited in note 24).
US Term Limits, 115 S Ct at 1895 (Thomas dissenting).
2
30 Id at 1860-61 n 27 (majorityopinion).
2

94
ofChicagoLaw Review
The University
[64:83
are equallyconsistentwith almajorityfoundsignificant-they

ternativeinterpretations.
is that the framers
The mostlikelyexplanationthroughout
wantedto avoidlanguagethatmightexposeor provokea conflict
enterprise.This tenthat could doom the whole constitutional
dencywill be recognizedby anyonewho has ever served on a
and it is quite helpfulin gettinglanguage approved
committee,
it offersno help on how to reby diversebodies.Unfortunately,
whenit eventuallyarises.
solvethesubmergedconflict
3. Framers'statements.
In rejectingthe proposalsto add a powerto add qualifications,someframersobjectedthatsuch powerwouldallow a currentlegislativemajorityto limitfutureelectionsto personslike
As JamesMadisonput it: "A Republicmaybe conthemselves.3"
vertedinto an aristocracyor oligarchyas well by limitingthe
numbercapable of being elected,as the numberauthorisedto
debatesMadisonrebuttedthe
Further,in theratification
elect."32
favoredthewealthybystatingthat,
chargethatthe Constitution
other than the qualificationslisted in the Constitution,"No
ofwealth,ofbirth,ofreligiousfaith,or ofcivilproqualification
fessionis permittedto fetterthejudgmentor disappointthe inclinationof the people."33Similarly,AlexanderHamilton reofthepersonswhomaychooseor be
sponded,"Thequalifications
and are unchosen... are definedand fixedin the Constitution,
alterablebythelegislature."34
But the contextforthis debate was defendingagainst the
imposedor allowedthe federal imchargethatthe Constitution
on who could run foroffice.The
positionof unfairrestrictions
goal, afterall, was gettingratificationfromstates leery that
to a powerfulnew federalentitymightprocedingsovereignty
duce bad consequences.By theirown termsthe quotes betray
this context,for Madison equated limits on who could be
"elected"withthoseon who could"elect,"and Hamiltonasserted
that the qualificationsof who "maychooseor be chosen"were
bothequally"fixed."Since states clearlycould set qualifications
Id at 1849-51,citingFarrand, ed, 2 Records at 250 (cited in note 24); US Term Limits, 115 S Ct at 1856-57, citingFederalist 52 (Madison), in Clinton Rossiter, ed, The Federalist Papers 325, 325 (Mentor 1961); US TermLimits, 115 S Ct at 1862-63.
32 Farrand, 2 Records at 250 (cited in note 24).
3 Federalist 57 (Madison), in Rossiter, ed, Federalist Papers 350, 351 (cited in note
31).
3' Federalist 60 (Hamilton), in Rossiter,ed, Federalist Papers at 371 (cited in note 31).

1997]
95
the
for federalvoters under explicitconstitutionalauthority,35
quotes seem to concedean equal role forstates to add qualifications for candidates. The statementscould thus be read as
meaningonly that the Constitutionand federallegislaturedid
not fetterthe judgmentof the people, not that the people of
states could not fettertheirown judgment.Both Madison and
Hamiltonwere,afterall, focusedon allayingfearsthat the new
wouldestablisha tyrannyby restricting
who
federalgovernment
could run foroffice,not on allayingfearsthat states mightmisuse a powerto restrictwhocouldrunforofficethatthe statesalreadypossessed overboththeirown legislatorsand theirrepresentativesto the Federal Congressunderthe ArticlesofConfederation.
Further,as the Courtnoted repeatedly,the framers'major
concernwas that an unlimitedpowerto add qualificationswould
allow a currentelite or legislativemajorityto entrenchitselfand
forma new "aristocracy
or oligarchy."36
That is, the concernwas
that limits on who could run forofficewould diminishvoters'
"capacityto displace incumbentsin favorof the representatives
of a recentlycoalesced majority."37
But this capacityis not diminishedby termlimits-it is enhancedwith a vengeance.The
objectionto termlimitsis not that theypreventvotersfromdisplacingincumbentsbut that theypreventvotersfromretaining
them,and thereis no evidencethe framersshared the latterconcern. To the contrary,James Madison, whose quotes the Court
made ExhibitA forits interpretation,
favoredtermlimitson the
merits; the Virginia Plan he forwardedto the Constitutional
Conventionwouldhave requiredeach House memberto leave officeafterone term.38
Moreover,the dissentcould cite its own FoundingFatherThomas Jefferson.
Unlike Madison and Hamilton,Jefferson
did
directlyaddress the questionwhetherstates could add qualificationsforfederalcandidatesand concludedtheycould.39
Jefferson
3 The Constitution sets the qualifications of voters for federal officeequal to whatever qualifications states impose on voters for the most numerous branch of the state
legislature. US Const, Art I, ? 2, cl 1. See also US Term Limits, 115 S Ct at 1856-57.
US Term Limits, 115 S Ct at 1848-50 & n 10, 1869, citing Farrand, ed, 2 Records at
3
250 (cited in note 24) (statement ofMadison).
" Laurence H. Tribe, American ConstitutionalLaw ? 13-18 at 1097 (Foundation 2d ed
1988). See also Parts II.A.3, II.B.1, and II.B.2.
38 See Stephen C. Erickson, James Madison: Term Limits Radical the Father of Constitution'sSolution to Congressional Career-ism,4 Term Limits Outlook Series No 3 (Oct
1995).
3
US Term Limits, 115 S Ct at 1860 n 24; id at 1888-89 (Thomas dissenting), citing
Letter of December 20, 1787, fromThomas Jeffersonto James Madison, in Andrew Lipscomb, ed, 6 The Writingsof Thomas Jefferson385, 389 (JeffersonMemorial Assn 1905).

96
The University
ofChicagoLaw Review
[64:83
concededthiswas "oneofthedoubtfulquestionson whichhonest

judicious
butthismerelyhighlights
Jefferson's
menmaydiffer,"40
forany fair-minded
personwouldfindthe matter
temperament,
debatable.4'In anyevent,whenone side has Madisonand Hamiltonsupportiveon a relatedpointand theotherside has Jefferson
supportiveon the exact pointbut concedingits ambiguity,perhaps it is timeto call a draw in the battle of FoundingFather
quotes.
Finally,themajorityreliedon an absenceofstatements.The
failureto imposerotarecordeddebate about the Constitution's
no
statement
that
the
states
themselvescouldimcontained
tion
Relyingon
pose rotationon theirown federalrepresentatives.42
an absence of recordedstatementsis always perilous.The absence may reflectonly the record'sincompleteness,especially
natureof the recordat issue.43And any
giventhe fragmentary
absence of statementsseems equally likely to indicate that
the states couldadoptrotaeveryoneassumedit was understood
the
After
as
to
indicate
tion
all, therecordis also devoid
opposite.
of statementsthat the states could not adopt rotationfortheir
ownfederalrepresentatives."
True,one mighthave thoughtthat
ratification
supporters,respondingto complaintsabout the lack
of constitutional
rotation,would have tried to blunt the complaintby pointingout that states could imposerotationif they
thoughtstatescould.But one wouldalso have thoughtthatrotaprovition proponentswould have arguedthat a constitutional
sionwas theonlywayto getrotationiftheythoughtstates could
not imposeit. Perhapsneitherside wantedto concedethe point
' US TermLimits, 115 S Ct at 1860 n 24 (majorityopinion); id at 1889 n 14 (Thomas
dissenting),citingLetter ofJanuary 31, 1814, fromThomas Jeffersonto Joseph Cabell, in
Lipscomb,ed, 14 Jefferson
at 82, 83 (cited in note 39).
views came in a letter
41 Since the statement that honest men may come to differing
to someone who held the opposingview, it may also have indicated simple politeness. See
Letter of January 31, 1814, fromJeffersonto Cabell, in Lipscomb, ed, 14 Jeffersonat 8385 (cited in note 39).
42 See US Term Limits, 115 S Ct at 1860. The Court also relied on statements by
Hamilton and Robert Livingstonagainst a rotationrequirement.Id. However, their context of defendingthe Constitution'sfailure to impose rotation deprives the statements of
any clear intentto deny states' abilityto tryrotationor other formsof term limits. In addition,the particular term limits proposal they were opposing would have required U.S.
Senators alone to leave offlceaftereach six-yearterm,which not only targeted the very
chamber of Congress meant to providethe most stabilitybut also meant that all Senators
would always be in theirfinal terms.See Michael J. Malbin, Federalists v. Antifederalists:
The Term-LimitationDebate at the Founding, in Benjamin and Malbin, eds, Limiting
Legislative Terms 53, 55, 57-61 (cited in note 5). See also Part IV.C (discussing final period problem).
4.3 US TermLimits, 115 S Ct at 1901 (Thomas dissenting).
4Id.

1997]
97
norwantedto bluntits own positionon ratifion state authority,

cation by announcingits secretlypreferredposition on state
authority,
leavingthematterforanotherday.
Whateverthe general wisdom of relyingon an absence of
statements,publicchoicetheoryprovidesan especiallypowerful
reason not to rely on it here. Collectiveaction problemsmade
states unlikelyto adoptrotationindividuallyhowevermuchthey
forany states thatimposedromightdesirerotationcollectively,
tationwouldhave less influencein the national legislaturethan
other states with long-timeuninterruptedrepresentation.45
It
was thus completely
rationalforall debatersto ignorethe unrealistic possibilityof individualstates adoptingrotationfortheir
federalofficialsevenwhiletheywere hotlydebatingwhetherthe
Constitution
shouldcollectively
adoptrotationforall states.
For the same reason, it is unsurprisingthat, despite the
widespreadpopularityof rotationat the time, no state in the
earlypost-Constitution
periodindependently
imposedrotationon
its federalrepresentatives.46
The Courtthus erredin concluding
that the states' failureto tryto rotate their representativesto
Congressindicatedthatthe statesmusthave thoughtthe Constitutionprohibitedrotationaltermlimits.47
More likelyit indicated
the collectiveaction problemswith such individualstate adoption. Consistentwith this explanation,the one state that still
had federaltermlimitsin 1788 rescindedthemin 1790,just in
timeto avoid losinglegislativesenioritywhen no otherstate followed suit withtheirown termlimits.48
Furthermore,
states did
impose various otherqualificationson their federalrepresentatives,as thenextsectionshows.
4 See Part II. For an explanationofhowtherecenttermlimitmovementendeavored

to overcomea similarcollectiveactionproblemwithstate-imposed
termlimitson federal
legislators,see Part II.D. Because rotationoustsall incumbentsat the end ofthecurrent
term,it wouldhave been muchmoredifficult
to organizesuch concertedbehavioramong
states.The motivation
was also likelyweakerat thetimegivenlowerlevelsoflegislative
tenureand distributable
benefits.See PartsII.A.2; II.C.2.
governmental
4 See US TermLimits,115 S Ct at 1865-66.Various states did imposerotationon
theirrepresentatives
undertheArticlesofConfederation,
but theyfacedno collectiveaction problembecause the Articlesimposedrotationon all congressmen.See Edmund
Cody Burnett, The Continental Congress 250 (MacMillan 1941).
4
US Term Limits, 115S Ct at 1866.
ComparePa Constof1776,? 11,in FrancisNewtonThorpe,ed, 5 The Federaland

StateConstitutions
3085 (Hein 1993),withPa Constof1790,in id at 3092-3103.See Brief
forState Petitionerin US TermLimits,1994 WL 444683 at *14 & n 14; BriefforPetitionerin US TermLimits,1994WL 444704 at *26 & n 32. See also Denboerand Jenner,
eds, 1 DocumentaryHistoryof the First Federal Election, 1788-1790 229-30 (1984).

98
The University
ofChicagoLaw Review
[64:83
4. Practicearoundthetimeoftheframing.
The practiceand law beforethe adoptionofthe Constitution
what the
can providebackgroundnormshelpfulin interpreting
framersmayhave meantorbeen thinkingabout.Practiceimmediatelyafterthe Constitution'sadoptionhelps us understand
partiesunderstoodits termsat the time.
how contemporaneous
neitherare veryhelpfulhere.
Unfortunately,
a) Priorpractice.The Court stressedthat the Constitution
was enactedjust afterthe resolutionofa relatedBritishcontroversy.The BritishParliamenthad foryearsrefusedto seat John
a
Wilkes,whohad been convictedofseditiouslibelforcriticizing
peace treatywith France. Afterseveral reelectionsand exclusions,Parliamentfinallyrelented,allowingthe votersto choose
But such exclusionsobviouslyraise
theirown representative.49
the concern of a currentlegislative majoritymanipulating
chargesto excludepoliticalopponentsand thusperpetuateitself
framerswished
in office.
Thus,evenassumingthe constitutional
to embodythis concern,it has no bearingon termlimits,a rule
excludes
adopted by the votersthemselvesthat prospectively
onlylegislatorswithlongincumbencies.In any event,the English rule barredonlyexclusionsformisconductnot coveredby
existingqualifications,not the promulgationof new, prospecbyParliament.50
qualifications
tively-applied
rules and
Moreover,if one were to look to pre-Constitution
practices,the morerelevantplace to look wouldbe in America.
Andheretherewas no doubtthatstates couldadd qualifications
notonlyforstate legislatorsbut forthe state'sfederalrepresenStill,it
tativesto CongressundertheArticlesofConfederation.5'
asit
the
states
shows
cuts.
Perhaps
way
this
is notclearwhich
sumed the same practicewould continue.But one mightalso
thinkthatit means states wouldhave realizedthe Constitution
failedto givesuch authorityand wouldhave raised objectionsif
theyhad any.
unpractice.Was the Constitution
b) Earlypost-Constitution
derstoodto changestates'priorpracticeand prohibittheaddition
The answeris unclear.Several states did imof qualifications?
pose additionalqualificationsin theirfirstlaws governingelections to the new Congress.The old PennsylvaniaConstitution
representacontinuedto imposetermlimitson its congressional
4 See US TermLimits, 115 S Ct at 1848.
5 Id at 1894 n 18 (Thomas dissenting); Powell u McCormack, 395 IJS 486, 528-29
(1969).
" US TermLimits, 115 S Ct at 1894 n 18 (Thomas dissenting).

1997]
Are TermLimitsUndemocratic?
99
Virginiarequiredits congressmento be propertyowners,

tives,52
five states required federal representativesto be residents of
theirdistricts,and threestates requiredfederalrepresentatives
to have been residentsforat least a year.53The Qualifications
Clauses not onlyimposedno propertyrequirementbut also required residencyonlyin the state "whenelected"and thus imApposed no districtresidencyor durationrequirementeither.54
parentlythese states thoughtaddingqualificationswas constitutional, and the recordsuggests state legislatorsfaced and reobjections.55
jected constitutional
But the Courtstressedthata largernumberofstates did not
includingmanystates that
imposeany additionalqualifications,
retainedpropertyownershiprequirementsforstate legislators.56
Moreover,the Court noted that state practice was unreliable
since states are predictablybiased in favorof state power.57It
also dismissed later state laws prohibitingpersons like felons
fromrunningforoffice,findingthemtoo removedin time from
the constitutional
framing.58
Congress'searly practicewas likewise ultimatelyinconclusive. In 1807, Congress consideredwhetherto seat William
McCreery,who allegedlydid not residein his district,as a state
statute required.59The relevant House committeereport concludedthatstate legislatureslacked authorityto add any qualifiBut the committee's
sets forth.60
cationsto thosethe Constitution
bothinside the
contested
was
constitutional
interpretation hotly
committeeand on the House floor.So contestedthat, afterfour
text.
See note48 and accompanying
See US TermLimits,115 S Ct at 1864-65;id at 1903-04(Thomasdissenting).
S See US Const,ArtI, ? 2, cl 2; US Const,ArtI, ? 3, cl 3.
s See US TermLimits,115 S Ct at 1904-05& nn 32-34(Thomasdissenting).
opinion).Ofcourse,thesestatesmayhave
Id at 1864-65& n 34, 1866n 41 (majority
forfederalcandidateswere less necessaryor
qualifications
concludedthat:(1) property
desirablethan forcandidatesto the state legislature,whereit was moreplausiblepauon federalcandidates
perswouldrun;or (2) any benefitsofimposingsuch qualifications
challengeeven ifthe states did not agree with
werenotworththe riskofconstitutional
thegroundsforsucha challenge.See id at 1907-08n 37 (Thomasdissenting).
opinion).
57 Id at 1864,1866n 41 (majority
(notingstatelaws disquali58 Id at 1866n 41. See also id at 1909(Thomasdissenting)
convictedofvotefraud,
mentalincompetents,
fyingfederalcandidateswhoare prisoners,
orunqualifiedtovote).
" Id at 1861 (majorityopinion);id at 1908-09(Thomasdissenting).The factthat no
such controversy
arose in the firsttwentyyears afterthe Constitutionsuggestspreviqualificaand property-ownership
ouslywidespreadacquiescencein thedistrict-residency
tionsimposedby somestates.Of course,such acquiescencemaynothave indicatedconof challengingthe requirementand
stitutionalagreementbut ratherthe impracticality
winningfroman outsidedistrictor withoutproperty.
' Id at 1908(Thomasdissenting),
citing17AnnalsofCongress871 (Gales and Seaton
1852).
52

100
The University
ofChicagoLaw Review
[64:83
days of debate,the House voted to recommitthe issue to the

whichthereuponexcisedall constitutional
committee,
interpretationsfromits reportand focusedsolelyon whetherMcCreeryin
factsatisfiedtheresidencyrequirement.
The House thenvotedto
seat McCreery,but deliberatelyrefusedto take any positionon
the constitutional
botha resoluissue, rejectingoverwhelmingly
tion that McCreeryhad the qualificationsimposedby his state
and a resolutionthathe neededonlythe ones listedin the QualificationsClause.6' If all this indicatesanything,it is that the
earlyCongressfoundthematterquiteuncertain,as it was. Other
congressionalexampleswere less equivocalbut came long after
the Constitution
was adopted,62
and werethus too late to reflect
oftheConstitution.
theoriginalunderstanding
In short,congressional
practiceis notdispositiveforreasons
parallelto thosethatthe Courtused to dismissstate practice.It
is equivocal,comesfroma sourceone wouldexpectto have a biased (here,relativelynarrow)view ofstate authority,
and much
ofit cametoolate tobear on theoriginalunderstanding.
5. Precedent.
The majorityreliedheavilyon Powell v McCormack,63
which
invalidateda House decisionto exclude a representative
allegedly guiltyof divertingHouse funds.The Powell decisionreasonedthatCongresshad no authority
to add to the qualifications
set forthin theConstitution.64
The dissentconcededthat Powell correctly
barredcongressionallyenactedtermlimits.65
Instead the dissent stressed,as
the majorityacknowledged,66
that (notwithstanding
some over6' US TermLimits, 115 S Ct at 1908 (Thomas dissenting),citing 17 Annals at 927-47
(cited in note 60). Moreover,some of the supportforstrikingdown the state statute came
from those who thought a state legislature could not add qualifications because that
power rested withthe people ofthe state acting by convention.See RoderickM. Hills, Jr.,
A Defense ofState ConstitutionalLimits on Federal Congressional Terms,53 U Pitt L Rev
97, 125-27 (1991). This position would not justify strikingdown qualifications added to
the state constitutionby popular initiative,as almost all modern legislative term limits
were.
62 See US TermLimits, 115 S Ct at 1862 (majorityopinion).
395 US 486 (1969).
Id at 522, 532, 550.
5 US TermLimits, 115 S Ct at 1889 (Thomas dissenting).
Id at 1890; id at 1852 (majorityopinion). Indeed, arguably Powell can stand only for
the propositionthat a single house of Congress cannot impose additional qualifications.
Referenceto Congress being unable to do the same are technicallydicta, and an Act of
Congress prospectivelyadding disqualifications raises less concern about opportunistic
exclusions. See also text accompanying note 50 (noting that Wilkes case indicated only
that Parliament would no longer exclude elected persons for conduct judged to be improper afterthe fact but not covered by a prospective statutorydisqualification);Powell,

1997]
101
broad dicta) Powell did not mean the states lacked authorityto
add qualifications.Nonetheless,ifthe policygroundsdrivingthe
Powell decisionare correct-thatallowingCongressto add qualificationscould lead to elitist exclusions and entrenchcurrent
representatives-theyalso suggest the same conclusion for
states.The dissent'sreadingofPowell thus leftit witha federalstate distinctionthat was unpersuasivein termsof democratic
policybecause its basis lay notin such policybut in relativelylegalisticand metaphysicalclaims about the meaningofstate sovereigntyin ourfederalsystem.67
Perhaps a betterresponsewouldhave been to challengethe
premisethat Powell applied to term limits.Afterall, Powell's
sweeping statementsthat Congress lacked power to add any
qualificationswere dicta. The actual holdingwas only that the
House couldnotexcludea particularmemberforhis alleged misconduct. Such exclusions raise the concern of trumped-up
chargesbeingused to targetpoliticalopponentsand thus implicate Powell's concerns(and those of the framers)regardingthe
ofa currentlegislativemajority.Powell couldthus
entrenchment
reasonablybe read to prohibitCongressfromadding onlythose
qualificationsthatraise the legislativeself-perpetuation
concern.
On this reading,term limits would be constitutionalwhether
adoptedby states or the federalgovernment
because, whatever
their other evils, term limits cannot furtherlegislative selfperpetuation.The betterdistinction
was notbetweenfederaland
state imposedqualifications,but betweentermlimitsand other
qualifications.
B. PolicyGroundsforResolvingthe Legal Ambiguity
1. Democraticpolicy.
In the end, the Court concededthat the constitutionaldebates were "inconclusive."68
In resolvingthe ambiguity,it relied
repeatedlyon "the fundamentalprincipleof our representative
democracy... thatthe peopleshouldchoosewhomtheyplease to
governthem."69
Indeed,the Courtitselfidentifiedthisdemocratic
395 US at 527-31 (focusing on Wilkes case). But the dominant concerns of entrenchment
would remain raised by a prospective statute, and much of the Powell opinion's logic applied equally to congressional statutes. See Powell, 395 US at 532-41, 547-48.
67
68
See Part I.B.2.

US TermLimits,115 S Ct at 1848.
69 Id at 1845, 1850-51, 1862 (internal quotation marks omitted). In addition to repeating this exact phrase no less than fivetimes, the Court continually stressed different
versions of this principle throughout its opinion. Id at 1848-51, 1857-58, 1860, 1862-64.
See also id at 1866 (summing up that its conclusion is based on "the available historical

102
The University
ofChicagoLaw Review
[64:83
principleas the "mostimportan[t]"

factorin its decision,more
importantthan "thetextand structureofthe Constitution"
and
"therelevanthistoricalmaterials."70
The policyargumentwas straightforward.
Restrictionson
whom voters could elect not only threatenedlegislativeselfperpetuationand elitistexclusionsbut violatedthe core democraticvalue thatvotersare the bestjudges ofwho shouldrepresent them. This normallydoes seem a sufflcientargument
against imposingadditionalqualificationson candidates.If the
votersdo not like Communists,
or Catholics,or renters,or persons refusingto pledgeto supporta flattax, theycan simplyrefuseto voteforthem.Theyhave no need to committhemselvesto
a legal bar on such personsrunningforoffice.But the Court
failedto see thatthisreasoningdid notapplyto termlimits.Voters couldnot simplyvotefornewcomersoverseniorincumbents
because,as Parts II and III show,collectiveactionproblemspenalized such a decisionand entrybarrierspreventedviable opponentsfromentering
thepicture.
It appearsthattheCourtdid notevensee thesepublicchoice
issues, forit showedno hesitationwhateverin brandishingits
democraticprincipleto strikedownwhat was, afterall, a democraticallyadopted rule on narrowingthe candidates.Had the
Courtseen the meritsas at all debatable,the ambiguityof the
constitutional
legal materialswouldsurelyhave made the Court
hesitantto interfere
withthedemocratic
wishesofelectoratesfavoringtermlimits.Its simplisticcondemnationof termlimits
seems particularlystrikinggiven that the same Constitution
providestermlimitsforthe President,showingthat even deliberativesupermajorities
can findtermlimitsdesirable.
If takenseriously,thereare broaderimplicationsto the majority'sclaim that fundamentaldemocraticprinciplesrequired
the invalidationof congressionaltermlimitsdespiteambiguous
constitutionallegal materials. The claim suggests the Court
shouldbe amenableto the argumentthat even termlimitson
staterepresentatives
are unconstitutional,
perhapsas a violation
oftheFirstor Fourteenth
Amendments.
Andindeedseveralsuch
challengesto state termlimitshave been raised in the wake of
US TermLimits.7"No doubtthe Court would recoilfromthat
and textual evidence, read in light of the basic principles of democracy").To be sure, at
the end ofits opinion the Court gave lip service to not consideringthe meritsoftermlimits, id at 1871, but this was belied by its repeated prior reliance on democraticpolicyand
principleto justifyits reading ofthe otherwiseambiguous textual and historicalevidence.
70
Id at 1856.
71 See, for
example, Nevada Judges Assn v Lau, 112 Nev 51, 910 P2d 898 (1996);

1997]
103
conclusion,not onlybecause of precedentto the contrary,72

but
also because it wouldbringintosharpreliefa conflict-theCourt
wouldbe sayingstates could not imposetermlimitson theirexecutiveseven thoughthe federalgovernment
does on its executive. However,the failureto extendits conclusionto the states
bringsintoreliefanotherincoherence:if it is fundamentally
undemocraticto imposetermlimitson federallegislators,whynot
on statelegislatorsas well?The Courtgave no answer.
2. Federalism,metaphysicsand defaultrules.
The dissentdid not avoid similar policyincoherence,forit
also adopted a federal/state
distinctionunrelatedto democratic
policyor principle:that additionalqualificationsforfederallegislatorscould be imposedby states, but not by the federalgovernment.But if the impositionof additional qualificationsby
Congresswas undemocraticand unsound,whywouldit not also
be undemocratic
whenimposedby states?The dissentbarelyaddressedthe issue, forits federal/state
distinctionwas grounded
not in policybut in a claim regardingthe logical entailmentof
state sovereignty
in our federalsystem.73
Nonetheless,two policy
distinctionswere implicitin its analysis,bothbased not on the
substantivemeritsof the qualificationat issue but on processbased claims about the identityofthe bodyimposingthe qualification.
First,the dissentarguedthat qualificationsadoptedby Congressinsteadofstate legislaturesposed a greaterdangerofelite
or legislativeself-perpetuation
because federallegislatorswould
be restricting
theirown opponentsand thus had a greaterconflictof interest.74
This does affordsome groundsfordistinction.
But the groundsare not verystrong.Afterall, elites or parties
thatcontrolCongresstendto controlstate legislaturesas well. If
the Democratsduringthe New Deal had decidedthat all future
candidatesshouldbe Democrats(or have some qualificationthat
serves as a usefulproxyforthat),theyprobablycould have imposedsucha qualificationas easily throughthe states as through
Congress.75
Moreover,even iffederally-imposed
qualificationsare
Bates v Jones,904 F Supp 1080 (N D Cal 1995).
72 See Gregoryv
Ashcroft,501 US 452, 462-63 (1991) (People of states have authority
to set qualificationsforstate officials.).
7;3US Term Limits, 115 S Ct at 1875-84 (Thomas dissenting).
74
See id at 1890.
75 See Rafael Gely and Pablo T. Spiller, The Political Economy of Supreme Court Constitutional Decisions: The Case of Roosevelt's Court-Packing Plan, 12 Intl Rev L & Econ
45, 60-64 (1992) (By 1936, the Democrats not only had two-thirdsof both houses of Con-

104
The University
ofChicagoLaw Review
[64:83
moredangerous,thatdoes not mean state-imposed

categorically
qualificationsare attractiveor remotelydemocratic.A state legislaturecomposedof millionairesmightconcludethat havinga
milliondollars shouldbe a qualificationforfederaloffice,or a
state legislaturecomposedof Republicansmightfindmembership in their partya requisitequalification.The potentialfor
of a governingclass or party reabuse and self-perpetuation
mains.
Second,the dissenthintedthatit was importantto its conwas imposedbyvoterinitiative,not
clusionthatthequalification
But the dissentneverdefendedthe proposilegislativestatute.76
of a state law could turn on
tion that the constitutionality
Indeed the
whetherit was adoptedlegislativelyor by voters.77
even in
the
of
term
limits
dissent supported
constitutionality
In anyevent,whilevoter
statesthatimposedthemlegislatively.78
adoption does lower the risk of elite or legislative selfit does not eliminateit, and democraticprinciples
perpetuation,
Voterswho are mostlyhomeowners
wouldremaincompromised.
or who belongto the currentmajoritypartymightwish to preor otherpartiesfromtakingofficein the
clude nonhomeowners
future.Just because such qualificationsmightbe imposed by
majorityvote does not make them democratic,for democracy
to selfmeans not a tyrannyofthe majoritybut a commitment
that leaves open the possibilitythat the majority
determination
fromthemajorityoftoday.79
To be sure,
oftomorrow
mightdiffer
ifthe majorityofvoterswish to electonlyhomeownersor membersoftheirparty,theycan alreadydo so. But addingqualifications to that effectundemocratically
imposes the views of the
currentelectorateon futureelectorates."Democracyenvisions
If one ofthosetemporuleby successivetemporary
majorities."80
it
to entrenchitselfpermanently,
rarymajoritieswerepermitted
to a newlycoalescedmajority
wouldnotonlydenyrepresentation
less accountableby undermining
but also make representatives
themechanismbywhichvoterscan registertheirdissatisfaction.
virtueone couldposto see whatprodemocratic
Andit is difficult
no matterwhoadopted
siblysee in suchadditionalqualifications,
them.
states.).
offorty-eight
gress but also controlledboth houses in thirty-three
US TermLimits, 115 S Ct at 1893-94, 1896, 1900 n 22 (Thomas dissenting).
77 No priorcase had ever so held. See id at 1858 n 19 (majorityopinion).
Id at 1893 (Thomas dissenting).
See Einer R. Elhauge, Does InterestGroup TheoryJustifyMore Intrusive Judicial
Review?, 101 Yale L J 31, 103-04 (1991).
Tribe,American ConstitutionalLaw ? 13-18 at 1097 (cited in note 37).
76
7
7

1997]
105
The dissent,however,never fullycame to gripswith these

in our
issues, forit focusedon the meaningof state sovereignty
federalsystem.To the dissent,the constitutionalstructureand
language ofthe TenthAmendmentmean that our primaryidentityis as citizensofstates thathave partiallyceded onlydefined
to the federalgovernmentand reportionsof theirsovereignty
served the rest.8'Thus, whereverthe Constitutiondoes not expresslyor implicitlyaddress an issue, the proper"defaultrule"
(i.e., the rule that operatesby defaultabsent contrarylanguage)
is that the states have poweroverit and the federalgovernment
does not.82Since the Constitutionis silenton termlimits,by this
logic states have the power to impose them thoughthe federal
does not.
government
The Courtwas drawnintothis discussiononlyto the extent
of respondingto the dissent's argument.It concludedthat because the rightto vote forfederalrepresentativeswas a "new"
rightcreatedby the Constitution,the states could "reserve"no
power regardingit.83Further,because federallegislatorswere
ofthe People of
notdelegatesfromthe states but representatives
federal
the UnitedStates,theirqualificationswere an inherently
Thus, on this issue, the properdefaultrule is that unmatter.84
less the Constitutiondelegates the matterto the states, they
the Courtdid not see
have no poweroverit.85But fundamentally
the questionas one ofstate versusfederalpower.Rather,its emphasis was that neitherthe states nor the federalgovernment
had any powerto add qualificationsto those set by the Constitution.And the democraticprinciplesit invokedapplied equally to
bothlevelsofgovernment.
JusticeKennedywas more enthralledby federalistmetaconclusionthan
physics,but his concurrencereached a different
the dissent.To him,the "Framerssplit the atom of sovereignty.
It was the genius oftheiridea that our citizenswouldhave two
political capacities, one state and one federal,each protected
such a theoryofUnfortunately,
fromincursionby the other."86
fersno explanationofhow to determinewhichside ofthe soverreeigntyatomwe are on-all the policyquestionsare implicitly
solvedby the largelyunexplainedlocationof the boundaryline.
Thus, Kennedyconcludedstate termlimitswere an "interference
8' US Term Limits, 115 S Ct at 1875-84 (Thomas dissenting).
82
Id at 1876.
Id at 1854-55 (majorityopinion).
4 See id at 1863-64, 1871.
83
85
See id at 1854.
' Id at 1872 (Kennedy concurring).

106
The University
ofChicagoLaw Review
[64:83
withthe federalrightto vote . . . in a congressionalelection,"87

but one could have equally concludedthat the Court'sdecision
withthe staterightofvotersto screen
was a federalinterference
and selectcandidatesforfederaloffice.Kennedy'sconclusionis
completelydeterminedby his premisethat the issue is on the
federalside ofthe atom,a premisethat neverreceivesa policy
justification.
One couldofcoursesay the same ofthe dissentand of the
majority'sforaysinto federalistmetaphysics.Each was grand
and evocative,but each turnedon definitional
premises:whether
the powerwas "new"or "reserved,"
whetherour primarystatus
is as federalor state citizens,whetherthe selectionofrepresentativesto Congressfroma givenstate is a federalor state matter,and whetherthe "people"referredto in the TenthAmendment are the people of the United States or of the individual
states. And each offeredlittle policy explicationof why we
should, given the inevitableambiguities,adopt one definition
over the other.Instead, we were treatedto a conflictbetween
largelyconclusoryintuitionsand inferencesthat do nothingto
advance the debate and indeed hinderit by obscuringpolicy
premises.
The impulseto operateat a highlevel ofabstractionis understandable.It allowed the majorityto avoid explainingjust
how termlimitscould possiblyfurtherelite or legislativeselfAnd it allowedthe dissentto avoidexplainingwhy
perpetuation.
the Qualifications
Clauses shouldleave statesfreeto adoptnoxious qualificationscategoricallybarred fromfederaladoption.
But the impulseis hardlyjustifiable,norrequiredby the methodologyof defaultrules. The defaultrule concept,comingfrom
contractand corporatelaw, is a quite practicalone. It holdsthat,
wherea contractor corporatecharteroffers
no clear answeron a
question,themattershouldbe resolvednotby strainedefforts
to
divinean intentof the contracting
parties,but ratherby reference to rulesunderstoodto applyby "default"unless the parties
contractout ofthem.The applicationofthisconceptto constitutional and statutoryinterpretation
is quite promising,albeit
largelyundeveloped,88
but it is plain that the sourceof default
rulesis notmetaphysical
butpolicydriven.
'The contractand corporateliteratureprovidestwo basic
sourcesfordefaultrules.The first,moredominant,sourceis the
Id at 1875.
" It is the topic of a separate article on which I am currentlyworking,tentativelyentitledStatutoryDefault Rules.
87

1997]
107
court'sbest guess about what rationalpartiesin the positionof

the contractingparties would have contractedto had they
thoughtabout the subject.In contractand corporatelaw, this is
generallypresumedto be the rule that maximizesthe aggregate
wealthofthe contracting
parties,on the groundsthat any effort
to impose distributionalshiftswould be offsetby price adjustments. In constitutionallaw, one mightlook more broadlyto
what rule wouldseem mostdesirableto partieswithouta vested
interestin the matterat hand.
The second source is what are knownas penaltydefaults.
Here the notionis that the defaultrule should be not the most
desirablerule but rathera rule that penalizes one ofthe parties
to encourageher to contractmoreexplicitlyabout the matter.89
The theoryis that explicitcontracting
will make the finalset of
than theywould be if relativelyimprecontractsmore efficient
cise courtstried to guess about the most efflcientrule forthe
parties.Penaltydefaultsare best limitedto areas whereexplicit
contractingseems likelyto occurbecause, if the penaltydefault
sticks,uncontractedout of,it worsenssocial welfarefromwhere
it wouldbe ifthe mostdesirabledefaultwereused instead.
The dissent'sdefaultrule seems to be a penaltydefault.The
opinionofferedno argumentthat state poweroverthe qualificationsoffederalofficialsreflectedthe mostdesirablerule. Rather,
the dissent stressedthe need forstate consentto the Constitufromwhichone mightinferthat
tion the framershad drafted90
the defaultrule of interpretation
should be designed to force
those framersto make any limitationson state power explicit.
But the problemwith this penaltydefaultlogic is the constituofamendment,an undesirable
tionalcontext.Giventhe difficulty
in
is
to
stick
default likely
perpetuity.
Moreover,since almostall
our constitutionalcontractingis behind us, the adoption of a
penaltydefaulthas little prospectiveeffecton futureconstitutional drafting.Finally,whilewe mighthave some hope ofrelaout ofundesirablepenaltydefaultsin
tivelycompletecontracting
the contractualand corporatesetting,the range ofissues in the
constitutional
settingis so vast thatwe cannothope foranything
close to completecontracting.
Thus, ironically,defaultrule logic
suggestscourtsshouldfeelfreerto importnotionsofgeneralsocial desirabilityto resolveconstitutional
ambiguitiesthan to resolve ambiguitiesin statutes,let alone those in contractsand
89 See, for example, Ian Ayres and Robert Gertner, Filling Gaps in Incomplete Contracts:An Economic TheoryofDefault Rules, 99 Yale L J 87, 97-100, 128 (1989).
9 See US Term Limits, 115 S Ct at 1875-77 (Thomas dissenting).

108
ofChicagoLaw Review
The University
[64:83
corporatecharters.Of course,oftenit willbe clear thatthe most

partiesor societywouldbe that
desirableruleforthecontracting
I
judgmentsofdesirability.
courtsnotmake ongoingcase-by-case
next.
turnto thisproblem
competence.
3. The slipperyslopeand institutional
defaultschool,one mightconclude
In line withthe efficient
that,giventheambiguouslegal materials,courtsshouldconclude
that
Clauses bar onlythosequalifications
thatthe Qualifications
exclusionsand self-perpetuation.
raise the risk of undemocratic
True,the Courtwouldlack authorityto imposesuch a rule abbasis. But the ambiguousmaterialsprosentsomeconstitutional
vide enoughbasis forjudicial authority,especiallygiventhat,
one thingwas clear:theframerswereconamidstthe ambiguity,
mightbe adoptedto perqualifications
cernedthatundemocratic
eliteorlegislature.9'
petuatea governing
an argumentthe
This conclusionmust,however,confront
Court did not make the slipperyslope argument.Suppose a
that only undemocratic
courtacceptedboth the interpretation
and the conclusionofthisArqualificationsare unconstitutional
it might
Nonetheless,
ticlethattermlimitsare notundemocratic.
reasonablybe concernedthat if it startsdownthe slope,it will
whoseclassification
as pro-or antidemofaceotherqualifications
to excraticwouldbe moreslippery.Wouldit be prodemocratic
partiesfromrunningfor
cludecandidatesoffascistor communist
election,at least ifthosepartiesmade clear theirintentto overifin offilce?
Undemocratic
to excludecandidates
throwdemocracy
or withtoo muchof it? To excludethosewith
withoutproperty
legal trainingor withoutfamilies?Or thosewho had not taken
civics?
college-level
on closer
Even seeminglyeasy cases mightbe moredifflcult
examination.An exclusionoffelonsmight,forexample,seem to
But theoldJohnWilkes
presentan easy case ofconstitutionality.
case itselfshowedthat felonychargescan be politicallymotiand
vated,92and the currentpracticeof seekinginvestigations
prosecutionsof politicalopponentshardlyprovidesconvincing
evidencethat the days ofsuch concernslie behindus. Fromthe
oppositeextreme,a blatantexclusionof,say, Republicansseems
But one mightreasonablyfearthatlegclearlyunconstitutional.
benignqualiislatureswillpose hard cases by adoptingformally
ficationsthatnonethelessserveas effective
proxiesformember" Id at 1849-51, 1856-57, 1862-63 (majorityopinion). See also Part I.A.3.
See Part I.A.4(a).
92

1997]
109
ship in the opposingparty.Once the camel's nose is in the tent,

keepingout therestofthe camel can be impossible.
The slipperyslope argumentreallyencompassestwo subarguments.First is the "pure" slipperyslope argument,that the
principleagainst undemocraticqualificationsis so slipperythat
no one could maintain a sharp distinctionbetweenthe instant
case (term limits) and the bottomof the slope (more noxious
Second is the argumentagainst giving added
qualifications).93
authorityto a particular entity.Here, the argumentis that
courtsshouldnothave a powerto decidewhichqualificationsare
democraticbecause theyeitherare likelyto misuse it or lack the
institutional
competenceto exerciseit.94Even in the case ofterm
limits,forexample,one maydoubtthe capacityofcourtsto reach
resolutionofthe complexissues identifiedin Parts
any definitive
II and III.
If the slope were too slipperyto undertake,the Court had
onlytwooptions:it couldcondemnall additionalqualificationsor
allow all of them. The formermeant the invalidationof seeminglybenignlaws, such as those disqualifyingcandidates who
or convictedofvotefraud.95
wereprisoners,mentalincompetents,
But presumablytherewas littleriskofvotersvotingforsuch folk
anyway.Given the self-policing
by votersof unqualifiedcandidates, the extremeof prohibitingall additional qualifications
seems preferableto the extremeof allowingall of them,especiallywhen one considersthe framers'clear concernsabout legislativeself-perpetuation.96
Such argumentsmightstrikeme as persuasivebut forthe
factthat the Court has long been engagedin reviewingthe le9 See FrederickSchauer, SlipperySlopes, 99 Harv L Rev 361, 370-71 (1985).
Schauer'sdelightful
piece argues in partthat such slipperyslope argumentsdependon
claimsaboutlinguisticimprecision
or limitedcomprehensibility,
suggestingthatsuch argumentsoftenlack plausibilitybecause morepreciselanguageis possible.Id at 370-76.
But the slope'sslipperinessmay be determined
by the principlesthat wouldbe deemed
ofthe legal materials.For example,one surelycouldlinguistiacceptableinterpretations
the QualificationsClauses to emcallygivethe slope excellenttractionby interpreting
bodythefollowing
principle:termlimitsare constitutional
but all otheradditionalqualificationsare unconstitutional.
However,the legal materialsprovideno basis fordrawing
thatline.Courtsare restricted
to the principlesthatmightplausiblybe generalizedfrom
the legal materials,and ifthoseare all eitherextreme(all additionalqualifications
valid
or invalid)or slippery(invalidatingonly "bad" or "undemocratic"
qualifications),then
courtshave a slipperyslopeproblemno matterhowlinguistically
possiblea high-traction
solutionmightbe.
See id at 367-68.
US TermLimits,115 S Ct at 1909(Thomasdissenting).
See Part I.A.3. While such concernswere focusedon federally-imposed
qualificaand state-imposed
tions,thereis littlepolicygroundto distinguishfederallyqualifications.See PartI.B.2.
9

The University
ofChicagoLaw Review
110
[64:83
gitimacyof the qualificationsstates impose on candidates for

state officeand on votersforbothfederaland state office.The
Courtnormallydoes so underthe Firstand FourteenthAmendThe result
ments,sincethe Qualifications
Clauses do notapply.97
has apparentlynotbeen a pell-mellslide downan unmanageable
slope.Indeed,giventhatthe state powerto set qualificationsfor
far
federalvoters98
is notonlyno less slipperyto reviewbut offers
morepotentialforabuse than a powerto qualifyfederalcandion thelatteroffers
littleadditional
dates,an absoluteprohibition
In any event,the doctrinealreadybuiltup underthe
protection.
First and FourteenthAmendment
createsguide rails forcourts
Whensubjectedto such reto keep theirfootingin the future.99
view,termlimitson state officialshave survivedwithoutdifficulty."0
Moreover,the Court's opinionwas susceptibleto its own
slipperyslope problem,forit invalidatednotjust absolutebans
on incumbents
runningbutbans on theirgettingballotaccess on
the groundsthat laws withthe purposeand effectofhandicappinga class ofcandidatesviolatedthe QualificationsClauses."0
had such a purposeand effect.But,
Clearlytheballotrestriction
as the dissentpointedout,ifone is willingto invalidateany law
thena host oflaws mightbe unwithsuch a purposeor effect,
constitutional:
campaignfinancelaws on the theorythat they
handicapchallengersto help incumbents(or vice versa); or the
drawingofdistrictlines (includingbothnormalgerrymandering
to help a partyand majority-minority
districtsto help a race) because it makesit harderforsomecandidatesto winelection.102
9 See US TermLimits,115 S Ct at 1913 (Thomasdissenting)

(notingthat those
clauses have been used to reviewlaws thathandicapparticularclasses of candidates).
The availabilityof such Firstand Fourteenth
Amendment
reviewmightbe thoughtto
forit wouldalso serveas a backstoppolicingnoxiousqualificasave thedissent'sposition,
tionsevenifthe Qualifications
Clauses wereinapplicable.ComparePart I.B.2. Withregardto states,then,the dissent'spositionmayeffectively
boildownto the same as that
advocatedhere,withtheonlydifference
beingthatthedissentwouldlocatetheauthority
forconstitutional
reviewin a different
clause. But thedissentwouldretaina distinction
betweenstateand federally
addedqualifications,
foronlythelatterwouldbe categorically
banned,and thatdistinction
cannotbe justifiedbyanystrongpolicydistinction.
US Term
Limits,115S Ct at 1913(Thomasdissenting).
98 See US Const,
ArtI, ? 2, cl 1 (givingstatespowerto qualifyfederalelectors).
See generallyTribe,AmericanConstitutional
Law ?? 13-10to 13-21at 1084-1112
(citedin note37).
' US TermLimits,115 S Ct at 1913(Thomasdissenting)
(collecting
cases). See also
Law ? 13-19at 1098(citedin note37) (notingthatcandiTribe,AmericanConstitutional
dateeligibility
requirements
generally
receiverationalbasis scrutiny).
US TermLimits, 115 S Ct at 1868.
102
at 1913-14(Thomasdissenting).

1997]
111
4. The possibilityofconstitutional
amendment.
Still, one mightfindthe issue of qualificationsforfederal
legislatorsdistinguishable,perhaps not because it is more slipperybut because its greaterimportancemakes similarslipperiness intolerable.If so, only a constitutionalamendmentcould
draw an acceptablynonslipperyline between term limits and
other additional qualifications.As a practical matter,that is
whereUS TermLimitsleaves us.103
But even iffederaltermlimitsare immenselydesirable,this
makes theiradoptionratherhopeless,does it not?The Constitutionprovidesforamendment,but onlyif proposedby two-thirds
ofbothhouses ofCongressor by a conventionapplied forby twothirds of the state legislatures.104It must also be ratifiedby
ofthe state legislatures.105
three-fourths
And if anythingis clear
about termlimits,it is that incumbentlegislatorsare loath to
enact them.Onlytwo ofthe twenty-four
state termlimitswere
passed by state legislatures;the rest were all by popularinitiative.106
CongressionalDemocratshave consistentlyopposedterm
limitswhen theywere in the majority.Having trumpetedterm
limitswhile in the opposition,Republicanshave feltobligedto
continueto voice supportforthem aftertheir 1994 takeoverof
Congress.But theyhave doneso withall the enthusiasmofa boy
asked to dance withhis sister,being carefulto fall shortof the
two-thirdsvote needed to actuallypropose an amendment.107
If
incumbentlegislatorshave a veto on termlimits,the likelihood
'Term limitsproponents
mightalso attemptconstitutional
changevia judicial appointment-that
is, appointing
a newjusticecommitted
to overturning
the5-4decisionin
US TermLimits.But thiswouldnotavoidslipperyslope problems,and it wouldcause a
wrenching
conflict
withstaredecisisprinciplesthatare generallywell worthsustaining.
Thereare also seriouspracticalproblemswithsuch a strategy.Litmustestsare hard to
implementand enforce,
and it is perilousto predicthow newjusticeswillvote.Furthermore,amongthejustices in the US TermLimitsmajority,onlyJusticeStevensis old
enoughto make his retirement
and replacementa realisticstrategyover the nextten
years.Even ifhe weresuccessfully
replacedby a newjustice who wouldvotedifferently
on termlimits,staredecisismightwell driveone ofthe originalfourdissentersto switch
sides ifthe same issue wereraised again. In any event,the morefundamental
problem
remainsthat,likeconstitutional
amendment,
thejudicialappointment
strategystillgives
thelegislaturean effective
vetosincetheSenate mustconfirm
anynominee.
Const, Art V.
05Ratificationmayalso be by state convention,
id, but Congresshas exclusivepower
104US
to decidewhatratification
procedureto use. See Tribe,AmericanConstitutional
Law ? 36 at 64-65n 9 (citedin note 37). It can thus requireratification
by state legislaturesif
that is morelikelyto blocktermlimits.To date, onlyone of the twenty-seven
amendmentshas been ratifiedbystateconventions.
Id.
106
Sullivan,109 Harv L Revat 78 n 1 (citedin note5). See also note6.
'07See Benjamin Sheffner,Term Limits Plan: Constitutional Convention, Roll Call 6
45 votesin Sen(Dec 18, 1995) (227 votesin House,63 shyof290 neededfortwo-thirds;

ate, 22 shyof67 needed).

112
The University
ofChicagoLaw Review
[64:83
proofgettingthemadoptedseems as highas gettinguniversity

fessorstovoluntarily
giveup tenure.
Grantedstate legislatorshave no directincentiveto oppose
legislatorcan
termlimitsforfederalofficials.
But anyincumbent
be expectedto be relativelybiased against argumentsthat incumbencyis bad.108
Moreover,state legislaturescan circumvent
the need forcongressionalapprovalonlyby applyingfora conventionthat mightnot produce the desired amendmentand
mightproduceotheramendmentsthat are notdesired.This unhas provena big enoughdisincentive
that no constitucertainty
tionalamendmenthas ever been producedby constitutional
convention.109
Finally,ifthe partyin controlofCongresscontrolsat
least one state legislativechamberin one-fourth
of the states,
whichit alwayshas, it can blockratification.
The practicaleffect
veto against
is to give any congressionalmajorityan effective
to its interests.Indeed,evenwhen
constitutional
changecontrary
of the states apply fora constitutional
two-thirds
convention,
Congressmustvoteto call forit."0
All this seems particularlyperversein this contextsince
conMadisonexplainedthat the rootcause ofhis entrenchment
cernaboutgivinglegislaturestheunfettered
rightto add qualificationswas that"therepresentatives
[ ] have a personalinterest
distinctfromthat oftheirConstituents.""11
Yet the Court'sdecision effectively
givesthat same legislaturea veto overa qualificationschangedesignedto reduceentrenchment
thatthosesame
legislatureshave a personalinterestin opposing.Shouldthe will
ofthepeoplebe thwartedin thisfashion?
Perhapsnot,if a noveltheoryby Yale Law ProfessorAkhil
Amaris recognized.Amar argues thatArticleV sets the exclu"8One mightthink that state legislators would like to impose term limits on federal
legislators so the state legislators could move up to the federal position,but state legislators are probablymore worriedabout term-limitedfederal officialscoming afterthe state
legislators'jobs. And ifthey do hope to move up to the federallevel, they are likely to disfavorlaws that limitthe amount oftime they can serve once theyget there.
"Tribe, American ConstitutionalLaw ? 3-6 at 64-65 n 9 (cited in note 37).
...US Const, Art V. This might be read to give Congress not just an effectivebut an
absolute veto on constitutionalchange. However, the betterreading (given the historical
evidence) is that Congress has an essentially ministerialobligation to call for a convention upon application by two-thirdsof the states. See American Bar Association, Special
ConstitutionalConventionStudy Committee,Amendmentof the Constitutionby the ConventionMethod Under Article V 12-18 (ABA 1974). Still, even a ministerial role involves
significantpower to determinewhether two-thirdsof the states have made the requisite
application in a sufficiently
contemporaneous manner and power to delay and influence
the shape and compositionof a convention.See id at 17-20; Patrick J. Fett and Daniel E.
Ponder, Congressional Term Limits, State Legislative Term Limits and Congressional
Turnover:A Theoryof Change, 26 Political Sci & Politics 211, 213-14 (1993).
"' Farrand, 2 Records at 250 (cited in note 24).

1997]
113
sive means by whichgovernment

officialscan amendthe Constitution,but does not bar thePeople fromamendingthe ConstituThe historicaland
tiondirectlythrougha simplemajorityvote."12
textual argumentsare fascinating,but would take us too far
afield.113
However,the basic policyargumentforthe distinction
will readilybe graspedby studentsof public choicetheory.The
is essentiallya contractlimitingthe authorityofthe
Constitution
People's agents,who make up the government,
and one does not
let agentsrewritetheirown agencycontractsfreely.The grounds
foroversightand onerousprocedureare less powerfulwhen the
principal, the People themselves, has amended the agent's
authority.
This theoryhas considerableproblems,nottheleast ofwhich
is determining
exactlywhen a majorityof the People have officially amended the Constitution.Moreover,how could we ever
without
get 250 millionpeople to decide anythingmeaningfully
goingthroughprocedures(like conventionsor elections)run or
set up by agentsjust as suspect as the government
agents subjected to the onerous ArticleV procedure?There is also the
problemof how a Constitutioncan protectagainst majoritarian
tyrannywhen the majorityis freeto alter the Constitutionat
will, a matterthat should concernall the People since everyone
is in the minorityon some issue and mightbecome part of a
broader political minorityin the future. These problemsuncertainty,
instability,irreducibleagencycosts,and majoritarian exploitation-suggestthe People wouldrationallydeclinethe
at least
independentamendmentpowerAmarwouldgivethem,114
absent a situationwhere governmentofficialswould be biased
againstthe amendment.
But forpresentpurposesit sufficesto notethat,ifit applies
Amar'stheoryshouldapplyto termlimits.For,unlike
anywhere,
with most amendments,thereare clear reasons whylegislators
mighthave biases leading them to block the preferencesof the
votersfortermlimits.And with term limitstherehas been a
procedureby whichthe will of the People has been registeredballot initiatives.If each state holds such a vote,
state-by-state
and a majorityofthosevotingapprovecongressionaltermlimits,
12
See Akhil Reed Amar, The Consent of the Governed: Constitutional Amendment
Outside ArticleV, 94 Colum L Rev 457, 458-59 (1994).
"3For a biting critique of Amar's textual and historical arguments, see Henry Paul
Monaghan, We the People[s], Original Understanding, and ConstitutionalAmendment,96
Colum L Rev 121, 126-73 (1996).
114See Charles Fried, The Supreme Court, 1994 Term: Foreword: Revolutions?, 109
Harv L Rev 13, 31-32 (1995) (The People have consented to a regime of more regularityin
the expression oftheir own will.).

114
The Universityof Chicago Law Review
[64:83
we wouldhave as gooda testcase oftheAmartheoryofconstitutional amendmentas one could imagine.The ironyis that the
Amartheoryrestson a premise-thepreeminence
ofthe People's
status as federalratherthan state citizens-directlycontraryto
thepremiseofThomas'sdissent.
II.
THREE COLLECTIVEACTIONPROBLEMS
A. The Basic Model

1. Lesseningideologicalslack.
It is perhapsbest to beginwitha concreteexample.Suppose
and receives
everydistricthas similarlyseniorrepresentatives
and pays fora similarshare of governmental
benefits.One districtis representedby Sally Conservative.Sally's views on ideologicalissues have becomeincreasingly
moreconservativethan
those of the votersin her district.Those voterswould prefer
someonewithmoreliberalviews.However,iftheyreplaceSally
witha newcomerwhoseideologicalviewsmorecloselymatchthe
district'selectorate,the districtwill lose seniority.With less
the districtwill get a smaller share of distributable
seniority,
governmental
benefits,thoughit will stillbear an equal share of
the costs.The district'svotersvote to retainSally because they
value her greaterabilityto delivergovernmental
benefitsmore
than her divergenceon ideologicalissues. Otherdistrictsretain
incumbentswho are more liberal than theirelectorateforthe
same reason.
Whyshouldthis be a problem?Afterall, no candidatecan
reflecteveryone'sviews.In choosingcandidateswe
everperfectly
mustchoosebetweenbundlesofpositionsand abilitiesand trade
offfavorableviewsand abilitieson someissues withunfavorable
views and abilities on other issues. That votersreelect Sally
must mean theypreferthe advantageof her abilityto deliver
benefitsto any disadvantagefromdivergenceon
governmental
ideologicalissues. They cannot overall prefera newcomerto
Sally,or else theywouldnotvoteforSally.
However,this is true only because of a collectiveaction
If all the districtscoulddo so collectively,
problem."5
theywould
".Although I originallydeveloped this analysis independentlyin 1990, and published
a brief outline of the argument in Einer Elhauge, Term Limits: Voters Aren't Schizophrenic,Wall St J A14 (Mar 14, 1995), the honor ofbeing the firstpublicationto point out
the connectionbetween termlimits and a collectiveaction problemgoes to Dick and Lott,
50 J Pub Econ at 8-9 (cited in note 7). See also Calamita, Comment,8 J L & Pol at 579-81
(cited in note 9) (noting the penalty on ousting powerfulincumbentswithoutlinking it to
the collective action problem); Hills, 53 U Pitt L Rev at 144-45 (cited in note 61) (same);

1997]
115
preferto replace theirseniorlegislatorswithnewcomershaving

closer ideologicalviews for,if everydistrictdoes so, no district
loses any advantagein gettinga share of governmental
benefits
and each districtgains moreaccuraterepresentation
on ideological issues. But, actingindividually,each districtwill retain its
senior legislatorno matterwhat it assumes the otherdistricts
will do. If it assumes the otherdistrictswill retaintheirlegislaEugene Mills, TermLimits and the Prisoner'sDilemma, 10 Pub AffairsQ 143 (1996)
(subsequentlypointingout the same collectiveactionproblem);JamesM. Buchananand
bylegislators,79 Pub
dilemmaand rentextraction
RogerD. Congleton,The incumbency
Choice47, 48-49(1994) (describinga similarcollectiveactionexplanation).See generally
Action:Public Goodsand theTheoryofGroups
MancurOlson,Jr.,TheLogic ofCollective
(Schockenrev ed 1971) (explainingthe collectiveactionproblemmoregenerally).Mills
thesame problembecause theycalledit a free
arguesthatDick and Lottdid notidentify
whichcan differ
fromthecollectiveaction/prisoner's
dilemmaproblem.See
riderproblem,
Q at 151-52n 15. But thisis an uncharitablereadingsinceDick and
Mills,10 Pub Affairs
featureofsucha problem:thatall partiescan be madebetteroff
Lottdo notethedefining
by collectively
agreeingto refrainfromactivitythat is individuallyattractive.See Dick
and Lott,50 J Pub Econ at 1, 4, 8. See generallyRussellHardin,CollectiveAction25-28
(JohnsHopkins1982) (showingthat collectiveactionand prisoner'sdilemmaproblems
are essentiallythe same). Mills also distinguishesDick and Lott's account as too
themotivesofactual voters,Mills,10 Pub Affairs
to describerealistically
Q at
"technical"
toois unpersuasive.WhileDick and Lott'sparticularmodel
151 n 15,butthisdistinction
freerider,collective
is probablytoo mathematicalforthe averagevoter,the underlying
action,and prisoner'sdilemmaconceptsare equally easy (or hard) to understandand
equally amendableto plain Englishor mathematics.In any event,it is unlikelyactual
votershad anysuch academicconceptsin mind,but thishardlymeanstheydid nothave
intuitivenotionsamountingto the same or thatit mattersto theinquirywhattheirsubjectivestate ofmindwas. See note 13. Dick and Lottmustbe giventhe credittheyare
due.
Dick and Lott's workis, however,morelimitedthan this Articlein various ways.
are rationalegoistswhosesole motiveis to
First,it assumesthatall politicalparticipants
transfer
maximizegovernmental
payments.Dick and Lott,50 J Pub Econ at 5-6.See also
Buchananand Congleton,79 Pub Choiceat 47-48(assuminglegislativeprocessis a distributivegame). This Articlegeneralizesthe problemto includemorerealisticassumpand behavior.Second,
tions that participantshave and expressideologicalpreferences
who
thisArticleshowshowcollectiveactionproblemscan produceseniorrepresentatives
notonlyin the allocationoftransferpayments
do notaccuratelyreflectvoterpreferences
but on any politicalissue. See Part II.A.1. Third,priorwork(includingmine) has assumed that collectiveaction problemsresultedsolelybecause clout over distributable
This Articleshows that collectiveactionprobbenefitsvariedwithlegislatorseniority.
cloutoverideologicalissues only.See id. Fourth,
lemscan resulteven ifseniorityaffects
and representative
this Articlegoes beyondthe divergencebetweendistrictpreferences
behaviorto describehow termlimitscan also alleviatetwo otherseparatecollectiveacthemutionproblems:(1) reducingoverallporklevels,see Part II.A.2; and (2) facilitating
likedbytheirown districtsbut considerednoxious
tual ousterofseniorrepresentatives
betweenthe
see Part II.A.3. Fifth,thisArticleexplainsthe connection
byotherdistricts,
natureofthe collectiveactionproblemand variationsin supportfortermlimitsamong
different
politicalparties,districts,eras, and jurisdictions.See Parts II.B.3 and II.C.
Sixth,thisArticleexplainswhystateswouldeverpass termlimitson theircongressional
delegationsgivensimilarcollectiveactionproblems,see Part II.D. ComparePart II.D
thisinexplicable)and Tabarrok,14 Cato
withMills,10 Pub AffairsQ at 152 n 17 (finding
it inexplicableundercollectiveactiontheory).
J at 344 (citedin note12) (likewisefinding

116
ofChicagoLaw Review
The University
[64:83
tors,it will retain its legislatorbecause oustingher would pebenefits.If

nalize the districtwitha lowershareofgovernmental
it thinksthe otherdistrictswill ousttheirlegislators,it will still
retainits legislatorbecause she will be able to deliveran even
benefitsand theystillvalue those
biggershare ofgovernmental
on ideologicalissues.
benefitsmorethan accuraterepresentation
Since each individualdistrictmakes the same calculation,none
oustsits seniorlegislator.
As a result,all districtscan be made betteroffbycollectively
agreeingto ousttheirseniorlegislatorswithtermlimits.So far,I
action
have said nothingaboutthe overalllevelofgovernmental
benefitsforparticulardistricts.Seniorityneed onlybe
producing
assumedto enable districtsto get a greatershare ofany governmentalbenefitsthereare to distribute.Nor have I made any asbenefitsare desirsumptionabout whetherthosegovernmental
able or not.116
The firstbenefitoftermlimitsis reducedideologibenecal slack,whetheror notthecurrentlevel ofgovernmental
fitsis desirableor affected
bytermlimits.
As I use it, the term"ideological"includesany view on govofthat
ernmentalactionthatdoes notdependon the distribution
actionacrossdistricts.For example,abortionis an ideologicalissue because one'sviewsaboutabortionare unlikelyto be affected
acrossdistrictsofits benefitsand costs.But,
by the distribution
"ideological"extendsfarmorebroadlythan
givenmydefinition,
is
the term ordinarily
used. If a certaindistrict'spositionon tariffsis drivensolelyby views on freetrade that apply equally
impactslocal producers,
whetheror not the tariffdifferentially
then the issue is "ideological"in that district.A districtmight
also have ideologicalviewson whethera programis enactedbut
viewson howanybenefitsofthatprogramare disnonideological
tributedacross districts.The term "ideological"also includes
mattersthat mightordinarilybe called questionsof character,
116In an earlier piece, I asserted that ideological slack resulted fromthe link between
seniorityand the distributionofundesirable pork. See Elhauge, Term Limits, Wall St J at
A14 (cited in note 115); see also Mills, 10 Pub AffairsQ at 145-46 (cited in note 115)
(similar assumption that seniorityclout over porkcreated the collectiveaction problemin
question). Upon reflection,this limitation to pork was mistaken. As long as some governmental action confersnet benefitson some districtsand more senior legislators have
greater power to affectthe distributionof those net benefits across districts,then the
collectiveaction problemwill result even ifall governmentalaction desirablyhas national
benefitsthat exceed national costs. (Shortly before publication, I came upon an article
making effectivelythis same point within a differentmodel. See Buchanan and Congleton, 79 Pub Choice at 47-48 (cited in note 115).) In addition, forreasons noted below, the
link between seniorityand legislative clout can create collective action problems even if
all issues are ideological. See Part II.A.3.

1997]
117
such as whetherthe legislatoris involvedin a sex scandal or

bouncedchecksat his congressionalbank.
Now we need to add anotherpoint.Increasedsenioritygives
governmental
a legislatormoreclout not only over distributing
benefitsbut also over many ideologicalissues. As long as the
legislator'sideologicalpositionscoincidemorewiththe district's
than theydiverge,the increasedclout on ideologicalissues that
comeswithseniorityconfersa net ideologicalbenefiton the district."7Thus,a legislatorwithsome ideologicalslack may still do
the ideologicalpreferences
ofher
a betteroveralljob offulfilling
ifshe has morepowerto getideologicallycompatible
constituents
legislationthrough.But if the legislatorbecomesmoreideologically divergentthan compatible,the effectof her seniorityon
unattractive.
ideologicalissues becomesafflrmatively
Withinthat limit,however,the effectof seniorityon ideological issues createsa collectiveaction problem.Indeed, it is a
problemthat would exist even withoutany distributablegovernmentalbenefits.Each districthas incentivesto vote for a
senior incumbentover a newcomerwith less ideologicaldivergence because the incumbenthas more clout on the ideological
issues where his positionscoincidewith the district's.But because total legislativepoweris a fixedsum, the districtscannot
collectivelyincrease their power by having senior legislators.
Theycan onlytryto increasetheirrelativepower.And sinceeach
otherdistrictdoes the same,notevenrelativepoweris increased.
Theywouldall thusbe betteroffwithless seniorlegislators,who
collectivelywouldhave no less legislativepowerbut who would
betterreflecttheideologicalviewsofeach district.
Nothingin the foregoinganalysis relies on the traditional
law and economicsassumptionthat voters and other political
participantsare rationalegoists,concernedonlywiththeirmaterial well-beingand maximizing governmentaltransfer payI merelyassume that whateveraims the participants
ments.1"8
pursue,theydo so rationally.Indeed, one major purposeof this
Articleis to demonstratethat termlimitscan be attractiveeven
to thosewhothinkthe politicalprocessmorefrequently
involves
the pursuitofideologicalaims that do not confermaterialbenefitson thosepursuingthem.The essential collectiveactionproblem is that-no matterwhat legislativepoweris used for-each
districthas incentivesto tryto increase its relativelegislative
"7 See Appendix.
118 Past law-and-economics

modelsofthe collectiveactionproblemaddressedby term
limitshave reliedon such assumptionsand thus been morelimitedin theirappeal and
abilityto explainbehavior.See note115.

118
ofChicagoLaw Review
The University
[64:83
no overall inpowerby reelectingincumbents,but collectively

crease is possiblesince the total amountof legislativepoweris
fixed.
2. Lesseningporklevels.
benefitsinto
Now let us divide distributablegovernmental
Oftenporkis definedin a
twocategories:"pork"and "nonpork."
narrowway thatputs it at $10 billion-an annoyance,but at 0.5
percentoffederalspending,perhapstoosmalltojustifydramatic
of "pork"is
changesin electoralprocedure."9But my definition
far morecapacious. It includesany governmental
actionwhose
totalcostsacrossall districtsexceedits benefitsbut whichsome
subset of districtsnonethelessfavorsbecause the benefitsare
in those districtsthan the costs. The classic
moreconcentrated
case mightbe an unnecessarydam put in a districtto createjobs
thereand fundedby taxes collectedfromall districts.But pork
thatdiscriminate
can includeregulationsor tariffs
againstsugar
and are favoredbycertaindistrictsbecause theybeneimporters
fitlocal sugarproducersand spreadtheircostsacrossa nationof
like Sosugar consumers.And it can even includeentitlements
cial Securityifsomedistrictshave moreelderlythan othersand
thus disproportionately
benefit.Anyotherdistributablegovernmentalbenefit,
whetherdesirableor not,is "nonpork."
Lest one thinkI am imposingan efflciency
normon a political systemthatmayjustifiably
be seekingotherends,I hastento
A
add thatbenefitsand costsneed notbe measuredmonetarily.
utilitarianmightpreferto measurethemin utiles,an egalitarian
in termsofwhethertheymoveus closeror further
froman equal
distribution
ofwealth,anotherin termsofmoralpluses and minuses. The problemthat remainsunderany normativebaseline
is that the pluses
that admits the need for some tradeoffs120
mightoutweighthe minusesin some districtsbut not in all the
districtsas a whole.And this mightcause individualdistrictsto
votequite differently
than theywouldif theyconsideredthe effecton all districts.This is the problemof porkbroadlyunderstood.Underany normativebaseline,somegovernmental
action
wouldbe classifiedas undesirablepork.12'
The followingmodel
"9 See Daniel Shaviro, The Political EconomyofBudget Deficits376 (1994 drafton file
withU Chi L Rev).
1 Since legislation occupies the realm leftto politicaljudgment rather than the realm
of absolute rights enforceableagainst the government,legislative issues presumably involve such tradeoffs.
121 I initially assume that all districts share the same normative baseline for determiningwhat counts as pork. I relax that assumption in Part II.C. 1.

1997]
119
can thus be broughtto bear even by thosewho whollyrejecteconomicefficiency

as a social goal.122
Suppose seniorityaffectsnot only the distributionof pork
but also its overalllevel. That is, a legislaturecomposedoflongservinglegislatorswill tendto create and pass moreporkthan a
legislaturecomposedofnewcomers.Then thereis a secondcollectiveactionreason whydistrictswill tendto voteforseniorrepresentativesbut favortermlimits.Votersfavorporkin theirdistrictsonly because they enjoy all the benefitsand spread the
costs overall districts.But theysufferfromthe costs of funding
pork forall the otherdistricts.Since, by definition,pork costs
morethanit is worth,each districtwouldenjoya net gain ifpork
in all districtswereloweredequally.
If each districtwants less pork,whydon'ttheyjust vote for
legislatorswho generateless of it? Because the benefitsof reduced pork levels must be shared with all the other districts
whetheror not the otherdistrictscontributeby pushingforless
pork.In contrast,each districtthat electeda low-porklegislator
would bear all the costs ofvirtuouslyrefrainingfromtakingits
share ofthe totalporkdistributed.Thus, each individualdistrict
rationallydeclinesto penalize an incumbentwho votes forpork
even ifthe districtwouldpreferlowerporklevels and the collective ousterof all representativeswho vote forpork.'23Ergo, the
districtsvote fortermlimitsas a means of overcomingthis collectiveactionproblemand achievingthe loweroverallamountof
porkeach districtprefers.
3. Gettingridofotherdistricts'representatives.
In lesseningpork levels, each districteffectively
trades its
senior representativeto eliminate senior representativesfrom
otherdistricts.This benefitseach districtbecause the costsofthe
pork avoided in otherdistrictsexceed the benefitsof the pork
"Again, past models have been more limited, assuming economic efficiencyas the
measure of social desert. See note 115.
'23The fact that a legislator suffersno penalty for approving pork distinguishes this
problem from ideological slack. As long as other districts are approving and splitting
pork, each districtrationally wants a representative who engages in pork barrel politics.
See Part II.B.3. A representative who approves pork is thus not ideologically diverging
fromhis electorate. Of course, ifthe representative's votes on pork levels did diverge from
the district'swishes, there would still be the problem that removing him would penalize
the districtwith a lower share of pork and other distributable governmental benefits. But
that is the same problem with any ideological divergence, and a representative who diverged from his district's wishes on pork levels would be able to diverge less on some
other issues.

120
The University
ofChicagoLaw Review
[64:83
forgonein theirdistricts.Could similartradingoccuroverideologicalissues?Yes.

Imagine a NorthCarolina voterwho on ideologicalissues
loves Jesse Helms but whohates Ted Kennedyeven more.Or a
Massachusettsvoterwhohates Helms morethan she loves Kennedy.Each voterwouldbenefiton purelyideologicalgroundsby
in exchangeforthe
givingup her favoredseniorrepresentative
term
limits
wouldeffectively
hers.
National
othervotergivingup
allow themto consummatesuch a trade.Moreover,such a situado
tioncreatesa rationalefortermlimitsevenifrepresentatives
divergefromtheirowndistrictsand evenifterm
notideologically
limitswouldnotreduceoverallporklevels.
To be clear,I am notclaimingany systematicreasonwhya
district'selectoratewould be more offendedby the contrary
ideologicalviews of other districts'representativesthan it is
ideologicalviewsofits ownrepresenpleased by the conforming
tative.I claim onlythat it is plausiblethis mightsometimesbe
the case, at least forsome significantshare of votersfavoring
termlimits.
Whereit is the case, thevotersagain facea collectiveaction
problemin ordinaryelections.Althougheach districtmightbe
willingto oust its ideologicallyfavoredrepresentativein exchangeforbeingrid of the ideologicallynoxiousrepresentatives
fromotherdistricts,no districthas incentivesto do so in ordinaryelections.For any such unilateralousterwoulddo nothing
to assure thatotherdistrictsfollowsuit.
But why should such votershave any confidencethat the
juniorlegislatorswho replace the departingseniorones will be
noxious?Afterall, the viewsof thejunior
any less ideologically
legislatorsmaybe no morenoxiousif,as positedforpurposesof
thissubsection,thedistrictfavorstheideologicalviewsofthe deBut the departingsenior legislatorwould
partinglegislator.'24
thoseideologicalviews,and would
have had morecloutto further
on thosein otherdistrictswith
thushave a greateradverseeffect
ideologicalviews.If the districtssuffera net ideological
contrary
loss fromeach districthavingits favoredideologicalrepresentaof
tive,the magnitudeofthatloss can be reducedifthe seniority
is likewisereduced.
thoserepresentatives
12 Given the greaterbrandname advantages described in Part IV, the departingsenior

legislator may also have been able to sustain a more extreme ideologythan the entering
junior legislator could. But I defer discussion of such brandname advantages until that
Part.

1997]
121
4. Conclusion.
As the above analysis suggests,thereis notjust one collectiveactionproblemat issue but threedistinctones. Anyofthem,
eithersinglyor in combination,
providesample explanationwhy
voterscouldjustifiablyfavortermlimitseven thoughtheyvote
for incumbents.What is individuallydesirable (keeping one's
ownincumbent)need notbe collectively
desirable(keepingall incumbents).
None ofthe above theoriesimpliesthe districtsmust dislike
theirown incumbents.They like them,whichis why they continueto reelectthem.Whateverdefectsthe incumbentshave are
offset(giventheirseniorityadvantage)bytheirgreaterabilityto
providethepork,the distributablebenefits,and/orthe ideological
gains each districtsincerelyvalues.'25What Americansdo not
like is the seniorityadvantageitselfand the collectiverepresentationtheyreceivein Congress.Typical polls at the time term
limitswere firstbeing enacted showedthat while morethan 85
percentof Americansexpressinga view approvedof their individual congressman,"6 78 percentof them disapprovedof Congress.'27This apparentgulfcan readilybe explainedby the various collectiveaction problemsdescribedabove. And the connection to termlimitsis that such limitswould help districtsproduce a Congressand a set ofincumbentsthattheywouldlike far
more.Thus termlimitswouldimprove,notworsen,the accuracy
withwhichthe legislaturereflectsthe democraticpreferencesof
the electorate.
B. Justifying
the Model'sAssumptions
1. The effectsofseniority.
I assume greaterseniorityhelps legislatorsdelivera larger
share of governmentalbenefitsto a district.I also assume
(separately)that senioritygives a legislatormoreclout on ideologicalissues. This is easy to see underour currentsystem,forit
gives senior legislatorsmore powerfulcommitteeassignments,
biggerstaffs,and a greaterchance at leadershippositions.One
mightobject,however,thatthiseffectofseniorityis an artifactof
"5Evenif an incumbentdisplaysideologicalslack fromthe district,the incumbent
would,underthismodel,be the best the districtcoulddo underthe currentsystemand
thusbe popular.That wouldnot,however,mean the districtcouldnotdo even betterby
changingthesystem.
' See WarrenE. Miller and Santa Traugott,AmericanNational ElectionStudies
Data Sourcebook,
1952-1986341, 371 (Harvard1989).
8 J L & Pol at 560 (citedin note9).
127Calamita,Comment,

122
The University
ofChicagoLaw Review
[64:83
thatcurrentsystem.Ifso, thecollectiveactionproblemI describe

couldbe curedmoredirectlyby abolishingthatsenioritysystem.
Rather than adopt termlimits,perhaps Congressshouldjust
cease allocatingcommitteeassignmentsand staffsby seniority.
Thereis something
to this,and it is notsurprisingthata Republican House relativelymoresympathetic
to termlimitshas also
beenless weddedto allocatingcommittee
assignmentsby seniority.But it cannotserveas a substitutefortermlimits.
First,no matterwhatthe internalsystemin the legislature,
moreexperienced
legislatorswillhave greaterabilityto workthe
system.'28
Experiencein officeincreasesa legislator'sfamiliarity
withlegislativerules and proceduresand the processby which
benefitsare distributed,
governmental
porkis passed, and statutes withideologicalimpactare enactedor blocked.Senior incumbentswill also have developedthe contacts,reputationsfor
honoring
deals, and politicalfavorsnecessaryforlogrolling.
They
willalso be moreskilledat recognizing
problemsoflegislativecyand at manipulatingthe agenda to
clingand path-dependence
take advantageofthem.
This last pointmay requiresome explanation.A legislative
majoritycan cyclebetweenpolicyoptionseven thougheach individuallegislatorrationallyranksthemin an intransitive
noncyFor example,supposeone-third
ofthe legislators
clingfashion.'29
A to B to C, one-third
B to C to A, and onepreferred
preferred
C toA to B. Then,by a voteof2-1,A wouldbeat
thirdpreferred
B, C wouldbeat A, and B wouldbeat C. Unless a previouslydefeatedoptioncannotbe raisedagain,thisproducesan infinitecycle. To avoidsuchcycles,legislativeproceduregenerallydoes not
allowthepreviously
defeatedoptionto be raised again. But then
theoutcomedependssolelyon theorderin whichthe optionsare
votedon.'30More seniorlegislatorsare morelikelyto have the
necessaryknowledgeabout otherlegislators'prioritiesto spot
such situationsand determinewhich order of votingis most
likelyto producetheirpreferred
result.Theyare also likelyto be
betterat manipulating
thatorder,as
theagenda thatdetermines
"See John R. Hibbing, Congressional Careers: Contours of Life in the U.S. House of
Representatives163 (North Carolina 1991) (findingthat holdingformalpositions has little
impact on measures of legislative effectivenessonce one controls for tenure, suggesting
that tenure, not position,is the cause of effectiveness);John R. Hibbing, Contours of the
Modern CongressionalCareer, 85 Am Pol Sci Rev 405, 424 (1991) (same).
"See Elhauge, 101 Yale L J at 101-02 (cited in note 79) (summarizing and collecting
sources).
'301If
the firstvote is betweenA and B, then C wins. If betweenA and C, then B wins.
If between B and C, then A wins. As this simple example suggests, it is generally advantageous to go last in the agenda.

1997]
123
well as to have positionsthat give them directagenda-setting

power.
Second,it wouldbe undesirableto whollyeliminatea legislative senioritysystembecause it has virtuesunrelatedto the distributionof governmentalbenefitsand ideological clout. Presumablywe do not want a newcomerheadinga foreignrelations
committeeher firstday on the job. Expertise and experience
about both subjectmatterand legislativeprocedureare helpful,
whichis whymanytermlimitsproponentsopposemoreextreme
formsthat would limit legislatorsto six years in officerather
than twelve.131
(Note,however,thatnothingin termlimitsmeans
thenew candidateswill lack substantiveexpertise.The freshman
Senator may have servedtwelveyears on a House foreignrelationscommittee;the freshmanRepresentativetwentyyears as a
public policy analyst.132)Seniorityrules also offera neutral
means of allocatingcommitteeassignmentsthat minimizespoliticalinfighting
and preventspossiblebias or otherabuses from
infecting
such selections.133
Congress'ssenioritysystemposes a large
Third,reforming
practical problem. It would require action by a sitting Conin pregress,134
which (absent term limits) has a self-interest
servingthe senioritysystemthat helps assure its reelection.Sigeven freshmanlegislatorshave an incentiveto favor
nificantly,
internalrules that conferseniorityadvantages (withinreason)
because such advantages make their own reelection more
The incentivesare even greaterformoreseniorlegislalikely.135
"'A House majorityapproveda twelve-year
termlimitbut rejecteda six-yearlimit.
ofHouse JointResolution73, the TermLimitsConSee Providingforthe Consideration
HR Rep 104-82,104thCong, 1st Sess 3-4 (1995) (settingforth
stitutionalAmendment,
bothamendmentproposals).See Garrett,81 CornellL Rev at 628 n 11 (citedin note9).
Unless termlimitsmakeraces morecompetitive
(i.e.,lowerthereelectionrate),a twelveyear termlimit'sreductionin mediancongressionaltenureis predictedto be relatively
small:fromnineto sevenyears.See W. RobertReed and D. Eric Schansberg,The House
underTermLimits:WhatWouldIt LookLike?,76 Soc Sci Q 699, 715 (1995).
'32InCosta Rica, whichhas had the longest(forty-year)
historyof legislativeterm
limits(a draconianversionallowingjust one four-year
term),only11 percentoflegislabeforebeingelected.JohnCarey,Parties,Incentives,
and Term
torsheldno politicaloffice
Limitsin CostaRica, in Grofman,
ed, LegislativeTermLimits321, 324-25(citedin note
5).
'3Minority
groupslike the CongressionalBlack Caucus, forexample,strongly
favor
rules.See CongressionalQuarterly,
seniority
Inc, CongressA to Z: CQ's ReadyReference
Encyclopedia35-36(CQ 1988).
' Indeed,the Constitution
giveseach House the powerto determineits own rules,
see US Const,ArtI, ? 5, cl 2, a powerthatperhapseven a federalstatutecouldnotrestrictwithoutviolatingthisconstitutional
provision.
'3Consistent withthis,even in the seniorityreformsthat followedin the wake of
large freshmanclasses in 1974 and 1994,see note 251, the ultimatechangewas quite
modest.The 1974 Congressallowedonlya caucus vetoofchairswho tookofficeby sen-

124
The University
ofChicagoLaw Review
[64:83
torswhogain notonlya reelectionadvantagebut also a present

increase in power.In contrast,beforeUS TermLimits,voters
had somehopethatstate-enacted
termlimitscouldcircumvent
a
self-interested
congressional
veto.136
Indeed,iftheyhad notbeen
invalidated,the enactmentof term limitsby states holdinga
majorityofcongressional
seats mighthave been the best way to
sufflciently
changetheincentivesofa majorityoffederallegislatorsto getthemto adoptinternalchangesin seniority
rules.'37
As the above suggests,a legislator'scloutturnsnotjust on
his ownseniority
buton the seniority
ofotherlegislatorsas well.
A highly experiencedlegislator with a powerfulcommittee
chairmanshiphas a larger advantage over inexperiencedcolleagues withno chairthan overthosewithmiddlingexperience
and positions.Votersthuscare aboutrelativeseniority
as well as
absoluteseniority.
This raises a questionabouttermlimits,forwhiletheylimit
the absolute tenureof legislators,theydo not eradicatevariationsin seniority.For example,in a twelve-year
termlimitregime,a districtwithan incumbent
havingtenyearsofexperience
still has incentivesto voteforthat incumbentovera newcomer
who muststartfromscratch.Termlimitsthus cannoteliminate
the coercivepressurethat relative seniorityhas in a system
where legislatorsrepresentseparate districts.However,term
limitscan lessen the pressureto vote forseniorincumbentsby
limitingthe possible size of the senioritydisadvantage.This
should set an upperboundon the ideologicaldistancebetween
electoratesand theirrepresentatives.
Assumingthe leveloftotal
porkdependson the average tenureof legislators,termlimits
shouldalso lessen thataveragetenurein a way thatlessens the
totalamountofporkthelegislatureproduces.
Politicalpartiescomplicatethispicture.For example,a seniorincumbent
mightbelongto a partyexpectedto be outofpower
in the nextlegislature.In that case, a districtmightbe able to
increaseits relativelegislativeclout by votingfora newcomer
fromthepartyexpectedto be in power.This pointis true,though
iority,and this veto was exercised only three times in 1974 and zero times the next Congress. See Garrett, 81 Cornell L Rev at 662 (cited in note 9). In the 1994 Congress,
Speaker Newt Gingrichdeclined to followseniorityin selecting committeechairs in only
three instances. Id at 145. Consistentwith the propositionthat such seniorityreformsdo
not address the underlyingcause of the seniorityadvantage, incumbentreelectionrates
increased, rather than decreased, after 1974. See Benjamin and Malbin, eds, Limiting
Legislative Terms Table B-2 at 293 (cited in note 5).
"Now the Court's decision requires term limits to be adopted by a constitutional
amendment procedurethat gives Congress an effectiveveto. See Part I.B.4.
117
See PartII.D.

1997]
125
limitedin the followingways. Even forlegislatorsbelongingto

leader ofa legthe minority
party,senioritymatters.A minority
islature probablyhas more clout than a freshmanbelongingto
has blockingpower.
the majorityparty,especiallyifthe minority
for
Consider, example,the stronginfluenceof Ted Kennedyin
the 1996 RepublicanSenate. Further,given uncertaintyabout
whichpartywill win,a seniorlegislatorfromthe ultimatelylosing partycan have moreexpectedlegislativecloutat electiontime
than a juniorlegislatorfromthe winningpartyeven if those belongingto thelosingpartyhave zero cloutin the end.138
More important,there are limitsto the ideologicalslack a districtwill
endure.The effectsdescribedhere are less likelyto make a districtelectsomeonefroman ideologicallyincompatiblepartythan
to elect someonewith divergentviews withinthe same party.
And as to ideologicaldivergencewithina party,the complication
raised in thisparagraphdoes notapply.
Nonetheless,the introductionof politicalparties does raise
some interestingpoints.First,termlimitsmay be less likelyto
shiftwhichpartya districtvotes forin generalelectionsthan to
shiftwho it votesforin partyprimaries.Second,the seniorityof
incumbentsfromthe minoritypartyis less advantageous,and I
wouldthus expectmoreturnoverin that party'slegislators.Consistentwiththis,the reelectionrates of incumbentRepublicans
in Congressfrom1970 to 1990 were lowerthan those of incumbent Democrats.139
Third,forthe partyin power,the effectscan
be mutuallyreinforcing.
Once a partygains legislativepower,the
coerciveeffect
ofseniorityhelps createan expectationthat all the
otherdistrictswillkeep the partyin power.Andthat expectation
furtherreinforces
the pressurein each districtto vote to retain
the incumbentfromthe majorityparty.This helps explain why
periodsof partydominanceover legislativehouses (such as the
Democraticdominanceofthe House fromthe 1950s to 1994) last
longerthan partydominanceof presidentialand gubernatorial
offices.
Finally, one might object that seniorityin a particular
branchofthe legislatureis notwhat matters,but ratherpolitical
experienceand expertisegenerally.If so, termlimits mightbe
"'Suppose, forexample,PartyX has a 60 percentchance oftakingthe House and
PartyY has a 40 percentchance.Candidatex fromPartyX willhave a cloutof 1.0 ifhis
partywins;the moreseniorcandidatey will have a cloutof2.0 ifher partywins.(Both
have a cloutof0.0 iftheirpartyloses.)The expectedcloutofcandidatey is thushigher(.8
to .6) thantheexpectedcloutofcandidatex eventhoughPartyX is expectedto win.
'39JohnB. Gilmour and Paul Rothstein, Term Limitation in a Dynamic Model of Partisan Balance, in Grofman,ed, Legislative Term Limits 145, 160 (cited in note 5).

126
ofChicagoLaw Review
The University
[64:83
perverse since they do not normallypreventa person who

reachesher termlimitin one house fromrunningforelectionin
The flawwiththis objectionis
the otherlegislativechamber.'40
that in factthe above explanationsforthe seniorityadvantage
are almostentirelyspecificto the branchoflegislaturein which
the incumbentserves.This goes most obviouslyforthe formal
by legislativerules. Spendingtwelveyears
advantagesconferred
mayhelpme gain a
as a memberoftheHouse ofRepresentatives
lotofpoliticalexpertise,butwhenI moveto the Senate I am still
a freshmanwithoutseniorityon committees.But even if the
thosemovingto a new legislativechamber
rules werereformed,
would still not knowthe ins and outs of that chamber'sprocedures,have establishedcontactsthere,or possess expertiseon
how new colleagues are likelyto prioritizeoptionsand which
agenda will thus likelybe favorable.FormerRepresentatives
do not seem any morepowerwhoswitchto the Senate therefore
fulthan otherfreshmanSenators.This explanationof the senconsistentwith term limits
iorityadvantage is thus perfectly
thatare similarlybranch-specific.'4'
2. The linkbetweenideologicaldistanceand individual
seniority.
For myfirstclaimedbenefitfromtermlimitsto be true,the
seniorityof individuallegislatorsmust correlatepositivelywith
ideologicaldistancefromtheirelectorates.Whyshould this be
so? One possibilityis that,as termlimitsproponentsoftenasof ideocorrupting
sert,time in the legislatureis affirmatively
intuitionis
logicalvalues.'42But manydoubtthis,and a contrary
that over time legislatorsshould become betterand betterat
matchingthe views oftheirelectorate.Indeed,one can imagine
an evolutionary
processwherethosewhofailto mirrortheirelectoratesare weeded out, meaningthat those who surviveto become seniorlegislatorsare those who mirrortheir electorates
best.4
81 CornellL Rev at 656-57(citedin note9) (stressingthattermlimits
1 See Garrett,
ofambitiouspoliticianwhomovesfromone term-limited
do notdeterthe progressively
ficeto another).
But even
advantagesdiscussedin Part III are less branch-specific.
14 The brandname
theyare somewhatspecificin thata largepartofthatbrandnameadvantageis therepuStill,the
its preferences.
tationforservingthe districtwell and accuratelyreflecting
is less,whichmayexplainwhytermlimitsjustifiedonlybybrandnameadvanspecificity
or President,thatwould
such as governor
tages have tendedto be appliedto positions,
thefinalcapstoneon a careerratherthana likelysteppingstone.
generallybe considered
42 See textaccompanying
note10.See also PartIV.B.
' See John R. Lott, Jr. and W. RobertReed, Shirking and sortingin a political mar-

1997]
127
I thus do not relyon any claim that legislativeseniorityaffirmatively

causes legislatorsto want to divergemore fromthe
views of theirelectorates.Rather I assume onlythat there are
naturaldriftsin theviews ofbothelectoratesand legislators,and
that moreseniorlegislatorshave morefreedomto express that
driftfully.Overtime,the politicalpreferences
ofthevotersor the
representativemay change with intellectualcurrents,or the
compositionofthe districtmay changewithdemographicshifts.
For whateverreason,an ideologicalgap develops.This is truefor
bothseniorand juniorlegislators.But because the seniorlegislatorhas morelegislativeclout,voterswill acceptgreaterideological slack withoutdenyinghim reelection.'"Assumingall legislators take advantage of the degreeof ideologicalslack presented
to them,the senior legislatorwill exhibitmore ideologicaldistance fromhis electorateeven if his true views do not vary any
morefromthe electorate'sthan the views of a freshmanlegislator.Moreover,because seniorlegislatorsand theirdistrictshave
had longerto driftfromthe ideologicalcorrespondence
that first
won the legislatorthe offilce,
I wouldexpecta greaterdistancein
theirtrueviewsas well.
We can all thinkofanecdotalexampleson the topic.For example, Ted Kennedyand StromThurmondseem to have more
extremelyliberal and conservativeideologiesthan less senior
Senatorsfromthe same or similarstates. But is thereany more
systematicevidence?Yes. JosephKalt and Mark Zupan findthat
the longera legislator'stenure,the less closelyhis votingbehavior reflectsthe ideologicalpreferencesofhis constituents.145
Steven Levittfindsthat first-term
Senatorsare morethan twiceas
responsiveto voterpreferencesas later-termSenators.146
Similarly,morepowerfulcommitteeassignmentscorrelatepositively
withideologicaldivergencein voting.147
In addition,many empiricalstudies have proven that incumbencyconfersadvantagesthatgiveincumbentsmoreroomto
ket withfinite-livedpoliticians, 61 Pub Choice 75, 76 (1989).
' Sometimesevena veryseniorlegislatorwillbe vulnerablewhenthe ideologicaldistancebetweenhim and his constituents

becomeslarge enoughto offsetthe value ofhis
experienceat providing
governmental
benefits.This mayexplainthe remarkableouster
ofHouse SpeakerTomFoley.But themoreimportant
lessonis just howremarkablesuch
an ousterwas. Unless a seniorlegislatormistakenly
overstepshis bounds,he shouldbe
able to retainhis office
despiteconsiderableideologicaldistancefromhis electorate.
'Joseph P. Kalt and Mark A. Zupan, The Apparent Ideological Behavior of Legislators: Testing for Principal-AgentSlack in Political Institutions,33 J L & Econ 103, 118-
20, 126(1990).
146
Steven D. Levitt, How Do Senators Vote? Disentangling the Role of Voter Preferences,Party Affiliation,and Senator Ideology, 86 Am Econ Rev 425, 436 (1996).
47Kalt and Zupan,33 J L
& Econ at 121-22,126 (citedin note145).

128
The University
ofChicagoLaw Review
[64:83
maneuver.Studies showthatnew legislatorspredictably

experience a "sophomoresurge,"a large statisticalvote gain between
theirelectionas challengerand theirfirstelectionas an incumbent,eventhoughtheyremainthesame person."4Further,when
an incumbentannounceshis retirement,
his partyexperiencesa
"retirement
slump,"a large statisticalvote loss betweenthe retiree'slast electionand the party'scandidatein the next election.'49Indeed,parties are fivetimesmorelikelyto lose a seat
theyheld the priortermifit is openthaniftheyare runningan
incumbent.'50
One recentstudyestimatesthe overallincumbency
advantage,theincreasedvotepercentagean incumbentpartyreceiveswhenit is runningan incumbentlegislatorratherthan a
newcomer,
at around10 to 12 percentin 1988.1'lIf anything,
this
the incumbency
studyunderestimates
advantagebynotincluding
uncontestedelections,wherethe incumbent's
advantagemay be
In 1990,
so greatthatit scares away all potentialchallengers.'52
19.5 percentofHouse incumbentsrunningfacedno majorparty
challengeat all.'53
Some disputethis evidence,arguingthat empiricalstudies
showlittleevidenceofideologicalshirkingby legislatorsand instead supportthe theorythat evolutionarysortingresults in
seniorlegislatorswho share the same ideologyas theirconstituents.'54But this argumentis based mainlyon evidencethat re"8AndrewGelman and Gary King, Estimating IncumbencyAdvantage withoutBias,
sources).
34 AmJ Pol Sci 1142,1149-50(1990) (collecting
149 Id.
'"Alexander Tabarrok, Term Limits and Political Conflict,in Grofman,ed, Legislative
TermLimits237,237-38(citedin note5) (25.7percentofopenHouse seats lost,5 percent

ofincumbent
seats).Likewise,42.7 percentofopenSenateseats are lostwhileonly15.5
seats are lost.Id.
percentofincumbent
and King,34 Am J Pol Sci at 1148,1157-58(citedin note 148). This vote
'51Gelman
doesnotappearto varymuchoverthecourseofa legislativecareer,see John
percentage
in Congress,in LawrenceC. Doddand BruceI. Oppenheimer,
eds,
R. Hibbing,Careerism
We
Reconsidered
Congress
67, 72 & fig3-2(CQ 5thed 1993),butthisis hardlysurprising.
morefromthe
wouldexpecta candidatetotakeadvantageofincreasedslackbydiverging
electorate,
meaningthata seniorlegislatormightwellhave and exercisefarmoreideobyanyhighermargin.
logicalslackbutnotwinreelections
152Gelmanand King,34 AmJ Pol Sci at 1158-60(citedin note148).
'S3
text accompanyingnote 259.
ofthisviewis JohnLott.See BruceBenderand JohnR. Lott,

"The mainproponent
Jr.,Legislator votingand shirking:A critical reviewof the literature,87 Pub Choice 67,
68,72-79(1996);JohnR. Lott,Jr.and StephenG. Bronars,Timeseriesevidenceon shirk-
ing in the U.S. House ofRepresentatives,76 Pub Choice125,126 (1993);JohnR. Lott,Jr.

and Michael L. Davis, A critical review and an extensionof the political shirking litera-
ture,74 Pub Choice461, 479 (1992); Lottand Reed,61 Pub Choice75, 87 (citedin note
143); JohnR. Lott,Jr.,PoliticalCheating,52 Pub Choice169, 183 (1987). This is surprisingsinceLottis one oftheauthorsofa piecethatarguedtermlimitsreduceideological slackbyreplacingseniorlegislatorswithjuniorones.Dickand Lott,50 J Pub Econ at
conflict-ifthereis no
11 (citedin note7). These positionsstrikeme as in unresolvable

1997]
129
change theirvotingbehavtiringlegislatorsdo not significantly

ior, at least not any more than legislatorsplanningto run for
But the bigThat evidenceis somewhatdisputed.'56
reelection.'55
ger problemis that such evidence of an absence of final-term
changes in legislatorbehaviordoes not contradictthe evidence
that ideologicaldivergenceincreases with legislator tenure.157
There are at least two explanationsthat can reconcilebothfindings.
First,retiringlegislatorsmaysimplyfeelobligedto carryout
the promisestheymade when theyran in theirlast election.If
so, what mattersis not that theyface no reelectionbut the degree of slack in the last electionthey ran. The slack operates
Since thereis littlerearatherthan retroactively.
prospectively
son to thinkthe degreesof slack betweenthe last electionand
there is littlereason to
the penultimateone are verydifferent,
howcan termlimitsreduceit?For an attemptto finessetheconflict,
ideologicalshirking,
seeidat8&n 11.
'5 See sourcescitedin note 154; JamesR. VanBeek,Does theDecisionto RetireIncrease theAmountofPoliticalShirking?,19 Pub Fin Q 444 (1991). Typicalis Lott and
Bronars,76 Pub Choice 125 (citedin note 154). It findsthat fromone termto the next,
changetheirideologyfrom7 to 15 pointson various100-pointscales, id at
congressmen
terms
betweena congressman'svotingrecordin different
129, and that the correlation
declinesthemoreseparatedbytimethosetermsare,id at 131-32.This seemslike a large
changeovertime,but Lottand Bronarsconcludefromthis evidencethat"congressional
stableovertime"becausethechangesare no different
votingpatternsremainremarkably
legislators.Id at 133.
forretiringand continuing
findsthat the magnitudeof ideologicaldivergenceincreasesbetweenthe
"56Zupan
firstand secondyearsofthe finaltermofa retiringlegislator.See MarkA. Zupan,The
notsubjectto a reelection
representatives
lastperiodproblemin politics:Do congressional
65 Pub Choice167, 172-76& 178 nn 17, 19 (1990).
altertheirvotingbehavior?,
constraint
ofretiringlegislatorsin theirfinalterm
But evenZupanfindsthattheoveralldivergence
legislatorsrunningforreelection.Id at
higherthanthatofnonretiring
is notsignificantly
176.
157 Some directtechnical
objectionsto one paper makingsuch a findingare raised in
Lott& Davis, 74 Pub Choiceat 462-63(citedin note 154),whichalso arguesthateven if
betweentenureand ideologicaldivergence,
relationship
significant
thereis a statistically
it is notverylarge.See also Benderand Lott,87 Pub Choiceat 76, 79 (citedin note 154)
(same); Lottand Bronars,76 Pub Choiceat 139-40(citedin note 154) (same). This is a
betweentenure
ofthe correlation
claimaboutthesize ratherthanstatisticalsignificance
the size of
For reasonsto thinkthe measuresunderestimate
and ideologicaldivergence.
notes166-68.Notealso thata failureto finda stasee textaccompanying
thecorrelation,
it simis notthe same as provingthereis no correlation;
correlation
tisticallysignificant
(usually90 percentor more)thatany
plymeansthatwe cannotsay withgreatconfidence
thatwas 89
is notaccidental.Thus,a studythatshoweda strongcorrelation
correlation
beinsignificant
percentlikelyto be morethanan accidentcouldcomeout as statistically
that a correlationexistedcould not be statedwith90
proposition
cause the affirmative
percentconfidence.But the negativepropositionthat no correlationexisted could be
In any event,Lott'spositionhere is again
statedwitheven less (11 percent)confidence.
between
hardto squarewithhis positionon termlimits:ifthereis littleifanycorrelation
how couldlimitingtenurepossiblyreduceideological
tenureand ideologicaldivergence,
divergence?

130
The University
ofChicagoLaw Review
[64:83
thinkthe retiringlegislatorshould change his behaviorvery

much,at least not any more than any legislatorrunningfor
A legislatorshould
reelection,
whichis whattheevidencefinds.'58
enjoysomewhatincreasedideologicalslack everyterm,but this
applies equally to legislatorstowardthe end and beginningof
the evidencethatideologitheircareers.None ofthiscontradicts
cal slack increases with tenure or suggestslegislatorsshould
increasetheirideologicalslack once theyannounce
dramatically
retirement.
Publicchoicescholarshave missedthisexplanationbecause
theyhave made "thevirtuallyuniversalassumptionthat politicians keep promisesto theirconstituents
solelybecause of the
threatof reelection."'59
But this assumptionis dubious. Even
withintheassumptionthatlegislatorssolelymaximizetheirpersonal utility,they have many reasons to stick to campaign
promises,includingretainingthe respectoftheirfamily,friends,
or historyin general.'60
And we cannotdismissthe
supporters,
factthat people generallyfeel honor-bound
to live up to their
promisesevenifit does notbringthempersonalgain.'6'Afterall,
we have all lefttips at restaurantseven ifwe neverplan to returnto thembecause we implicitlypromisedto do so when we
sat down.
Second, even if we adopt more cynical assumptions,we
mightsee no changein retiringlegislators'finaltermsprecisely
because theirseniorityhas giventhemso muchideologicalslack
that theydon'thave any furtherto go when the reelectionconstraintis removedentirely.'62
Afterall, thelegislatorsold enough
to have reachedretirement
age are generallyalso seniorenough
to enjoygreat ideologicalslack because of the collectiveaction
problemsdiscussed above and the entrybarrierproblemsdetailedin Part III. Andto getelectedin the firstplace,politicians
withtheirelecmustbeginwithsomeideologicalcorrespondence
"See sources cited in notes 155-57. This theoryhas the added benefitof explaining
why politicians ever announce their retirementbeforethey run forreelection.Voters do
not refrainfromelecting such politicians because they expect them to largely live up to
theircampaign promises.
'59Lottand Reed, 61 Pub Choice at 87 (cited in note 143).
160Lottadjusts for the possibilitythat post-legislativeemploymentmight discipline
how final-termvotes are cast and generally (though not always) finds no significantcorrelation. See Lott, 52 Pub Choice at 176-79 (cited in note 154).
6' Such honor is not inconsistent with the premise that senior legislators exercise
ideological slack fromtheir constituentsbecause such slack is exercised to advance what
the legislatorsbelieve to be right.
]62 Compare Levitt, 86 Am Econ Rev at 436 & n 22 (cited in note 146) (findingsecondterm Senators half as responsive to electorate preferencesas firsttermers but no more
responsivethan higher-termSenators).

1997]
131
amountofslack can thus allow themto vote

torate.A significant
theirown preferences
fullyby theirpenultimatetermand leave
themwithno furtherdesireto changevoteswhentheyno longer
facereelection.163
JohnLott dismisses this second possibilitybecause of evidence that,whiletheydo notchangetheirvotes,retiringlegislaHe thus concludesthat theymust
torsdo voteless frequently."6
be enjoyingmoreslack but choosingnot to exerciseit to change
theirvotes because theyderivedpersonalutilityfromthe votes
theycast in the past. But decreasedattendanceofretiringlegislatorsmaybe due to lowerenergyor worsehealthratherthan to
increased slack. True, Lott findsthat age and reportedillness
explain some but not all the decrease in attendance.165
Still,unreportedillnesses and a more nebulous dropoffin energyare
commoncauses ofretirement
that could easily explain the rest.
In anyevent,greaterchangesin whethervotesare cast thanhow
theyare cast are perfectly
consistentwith the firstexplanation
in
notedabove. Incumbents theirfinalelectionare likelyto have
made explicitpromiseson theirpoliticalstances that theygenerto keep. But fewcandidatesmake promises
ally feelhonor-bound
abouttheirfutureattendancerecords.
If anything,the empiricalstudies underestimateideological
slack for various reasons. First, they measure only recorded
votes.'66But muchideologicalslackingis likelyto take the form
of failingto push forpositionsthe electoratefavorsby making
speeches,offering
trades,twistingarms, and the like.167
Making
'"Again, this explanationis consistentwithpoliticianannouncements
ofretirement
beforeupcomingelections.Because votersalreadyvalue theirlegislator'sseniorityclout
morethan theydisvaluethe fullexerciseof the politician'sideologicalslack, such announcements
do notdissuadevotersfromelectingtheretiring
politician.
M See Benderand Lott,87 Pub Choiceat 75-76(citedin note 154); JohnR. Lott,Jr.,
AttendanceRates,PoliticalShirking,and theEffectofPost-ElectiveOfficeEmployment,
28 Econ Inquiry133,135 (1990);Lott,52 Pub Choiceat 181-83(citedin note154).
'6Lott,28 Econ Inquiryat 141 (citedin note164).
Thus, failureto attendcongressionalvotes would not countas ideologicaldivergenceundertheirdefinition
even thoughit wouldundermine.See Part II.A.1. Another
oddityis that the studiesrelyalmostexclusivelyon data fromcongressionalraces in
1976-78.See, forexample,Lott,28 Econ Inquiryat 136 (citedin note 164); Kalt and Zupan, 33 J L & Econ at 110 (citedin note145). Whetherthe same findingsholdup in differentperiodsor legislaturesremainslargelyuntested.
167 Although
mostofthesecannotbe directly
measured,thereis evidencethat,in additionto votingless, congressmen
in theirlast termmake fewerspeeches,travelless, and
offerfeweramendmentsand bills.See RebekahHerrick,MichaelK. Moore,and JohnR.
the ElectoralConnection:The Behaviorof U.S. Representatives
Hibbing,Unfastening
WhenReelectionIs No Longera Factor,56 J Pol 214, 219-20(1994). Unfortunately,
the
evidencedoes notenable one to determine
whetherthisreflectsa generalloweringofeffortor a selectiveloweringofeffort
forunfavoredcauses, thoughthe rangeofissues on
whichbillsare introduced
in thefinalterm.Id at 221.
is narrower

132
The University
ofChicagoLaw Review
[64:83
recordedvotes contraryto the electorate'sviews (or one's past

positions)is a less attractivemeansofexploitingslack sinceit is
far moresalient and easier forothersto notice.Unfortunately,
push theirelectorates'views
the degreeto whichrepresentatives
if not impossibleto measure. Second,the empirical
is diff'icult
in a way thatlumpswhatI call ideostudiesdefine"ideological"
benefits.
ofgovernmental
logicalwithwhatI call the distribution
Thus,a legislatorwho betterrepresentshis districton the comfactorsmight
bination of ideologicaland benefit-distribution
undermy
even
though
underthestudies
scorea lowerdivergency
his "ideological"
distancehas increasedas his provision
definition
ofbenefitshas improved.Third,manystudiesdefineideological
withvariousspecialinterestindexesof
divergence
byconsistency
voting,such as the AmericansforDemocraticAccongressional
Ideologicaldivertion or the AmericanConservativeUnion.168
genceon mattersthatare not ofinterestto thosegroupsis thus
leftoutofthepicture.
Whateverwe thinkof the empiricalevidence,our normal
means of resolvingempiricalconflictsabout the state of the
worldwhen makingchangesto legislativerules is throughthe
politicalprocess.Andthereis certainlyenoughevidenceto make
it rationalforvotersto believe ideologicalslack increaseswith
legislativetenure. This sufficesto answer the constitutional
in a way thatallows
question:whetherwe shouldbe constituted
ofvotersto set termlimitsiftheywish.
a majority
Moreover,even if the empiricalevidence did show that
thatwouldonly
ideologicalslackdoes notincreasewithseniority,
in Part II.A.1. The
rebutthe collectiveactionproblemidentified
in Parts II.A.2 and II.A.3collectiveactionproblemsidentified
but notcollectively
individually
legislativeporkthatis preferred
fromother
noxious
eliminate
legislators
collective
to
trading
and
districts-wouldremainand be addressedbytermlimits.
levels.
3. The linkbetweenoverallporkand seniority
My secondclaimedbenefitof termlimitsis that theywill
Whyshouldthis be so?
loweroveralllevels oflegislativepork."69
"See, forexample, Lott and Bronars, 76 Pub Choice at 128 (cited in note 154); Lott
and Davis, 74 Pub Choice at 467-72 (cited in note 154); Lott, 52 Pub Choice at 171 (cited
in note 154). Other studies define "ideological' divergence as the residual of whatever
cannot be explained by various constituencyinterestvariables. See Bender and Lott, 87
Pub Choice at 73, 79-80 (cited in note 154) (collectingsources). This creates the considerable problemthat the ideological residual may just reflectconstituencyinterestvariables
that were leftout. Id at 73.
'69Mills also assumes it is at least prima facie plausible that term limits will lower the

1997]
133
True, withina legislatureof varyingseniority,the more senior

legislatorsshouldbe moreenthusiasticabout porkthan thejunior ones since the seniorones get a largershare.170
But one cannot necessarilydeduce fromthis that decreasing the average
tenureofa legislaturewill decreaseporksince the average pork
share could remainthe same. For that matter,whyis porkever
enactedifits costsexceedits benefits?
There are two standardexplanationsof the pork phenomenon. The firstis that the concentrated
benefitsof porkin a particulardistrictare moresalient and affectan easier-to-organize
groupthan the costsofpork,whichare diffusely
spread throughout other districts."7'For example, a district'svoters know
whetherits main employerhas gone out of business,but voters
fromotherdistrictscannotmonitoreach itemin the governmenoveralllevelofporkrelativeto general-interest
legislation.See Mills,10 Pub Affairs
Q at
147 (citedin note115). But unlikethisArticle,thatis the onlycollectiveactionbenefitof
termlimitsthathe cites.ComparewithParts II.A.1; II.A.3. Further,his reasonsforassumingtermlimitswillreduceporkare unpersuasive.First,he equates undesirablepork
withthe distributable
governmental
benefits.See Mills, 10 Pub AffairsQ at 145, 147
(citedin note115): "at least somesuchlocal economicbenefitsare pork,in the sense that
theyservelocal interestswhilethwarting
broadernationalones" and "it oversimplifies
onlyslightlyto say thatunderourcurrentsystem,each legislatorgets as muchas possible forher own districtwithoutregardto the nationalcosts, so that voters'interests
wouldprobablybe betterservedif[they]foreswore
theporkthatlegislatorswithoutterm
limitsprovide."However,distributable
governmental
benefitsthat are nationallydesirable can also drivethecollectiveactionproblem,
see note116 and accompanying
text,and
thereis no compellingreasonto believethat a reductionin distributable
governmental
benefitsgenerallywillbenefitvoters.Second,Mills assumesforpurposesofhis primafacie case thatlegislatorsseekingreelectionundertermlimitswill have less incentiveto
enactpork.Id at 147. But thatis refutedby the analysisin the textfollowing
thisfootnote.Third,his main argumentis thattermlimitswill producemorelegislatorsin their
last termwho,notneedingto be reelected,willenactless pork.Id. But he offersno persuasive reasonto thinkthat,primafacie,the benefitsofunaccountability
on porkwould
be outweighedby the disadvantageofunaccountability
on all otherpoliticalissues. See
Part IV.C. Moreover,
theactualvalidity(as opposedto primafacieplausibility)
ofthisargumentis undermined
by evidencethatvotingbehaviordoes not changein finalterms.
See Part II.B.2. Iflegislatorsdidlessentheirpursuitofporkin theirfinalterm,thenvoters wouldbe less likelyto electlegislatorsto thatfinalterm,so it is notsurprisingthat
legislatorslearn how to commitcrediblyto continuedporkprovision.Mills does not address the argument,put forthhere,thatlowerlegislativetenureselectivelylowersthe
abilityto act on incentivesto enactpork.
'70Thereis evidencethatthelongertheyservein office,
the morelikelybothRepublicans and Democratsare to favorspendingbills. See James L. Payne, The Cultureof
Spending: Why Congress Lives Beyond Our Means, 81-87, 176-78 (ICS 1991). See also
at 61 (citedin note10); W. RobertReed and D. Eric Schansberg,Impact

Will,Restoration
of Congressional Tenure Restrictionson Spending, in Grofman,ed, Legislative Term Limits 101, 111 (citedin note5) (Tenurehad small but statistically
effecton ratsignificant
ingsby NationalTaxpayersUnion.).To be sure,not all spendingis pork.But whiledifferentviewsaboutthe desirablelevelofspendingcan explainthe different
spendinglevels betweenthe parties,it is hardto see whytheywouldexplainan increasein spending
overtimebyindividualsin bothparties.
"' See Elhauge,101Yale L J at 41 (citedin note79).

134
The University
ofChicagoLaw Review
[64:83
tal budgetto see ifit includesa wastefulsubsidyto propup that

employer.Moreover,while the district'slegislatorcan credibly
her district(who else would be
claim creditforporkbenefiting
forit?),it is hardto singleherout forblamein approvingporkin
otherdistricts.Thus,a legislator'selectoralchancesare farmore
benefitedby gettingporkin her districtthan theyare hurtby
apallowingporkin otherdistricts.A legislatorwill accordingly
in
in
provepork otherdistricts exchangeforotherlegislators'apthe odds ofreelectionfor
provingporkin her district,improving
thenation.
everyonethoughimpoverishing
The second explanationis that pork reflectsa "univerproblems
salistic"normforreducingcyclingand agenda-setting
with majorityrule.'72Recall that even thougheach individual
legislatorrationallyranksthe policyoptions,a legislativemajority can cyclebetweenthem,thus producingopportunitiesfor
The universalisticnormtriesto reduce
agenda manipulation.173
cyclingand agenda manipulationby sharing agenda-setting
poweramongall members,givingeach powerto selectpolicyin
thearea ofmostconcernto herdistrictthrougha normofmutual
deference.Each memberthus naturallyfavorsher districtwith
porkin thoseareas in whichshe has thepowerto selectpolicy.
Howeverit gets started,porkis hard to stop because of a
collectiveactionproblemamong districts.Even if each district
it knowsthata deprefersto end porkbarrelpoliticsaltogether,
refrainfromit willcostthedistrictits normal
cisionto virtuously
share of porkand have littleeffecton the overalllevel of pork.
The otherdistrictswill continueto provideporkto each other,
and the lowercostsfromnot providingporkto the districtthat
whethertheyopt
optsout willbe sharedamongall the districts,
out or not.Thus even a legislatorwho perfectly
representedthe
interestsof his districtwould continueto providepork under
these circumstances.
Each districtindividuallywantsporkeven
theywouldliketo end it.'74
thoughcollectively
of territorially
These motivationsforpork are a byproduct
electedlegislatorsand/orcyclingproblemsinherentin majority
'72See, for example, Barry R. Weingast, A Rational Choice Perspectiveon Congressional Norms, 23 Am J Pol Sci 245, 250-55 (1979); Emerson M.S. Niou and Peter C. Ordeshook, Uniuersalismin Congress,29 Am J Pol Sci 246, 256-57 (1985).
'73
See textaccompanying
notes129-30.
One mightimagine a collectiveagreement to refrainfromprovidingpork. But pork

is impossible to definein an enforceableway. Further,such an agreement would only be
enforceable against futurelegislatures if it were incorporatedinto the Constitution. In
any event, whether done through constitutional amendment or statute, a collective
agreementto curb pork can be obtained only by overcominga legislativeveto that legislators have everyincentiveto exercise to preserve theirincumbencyadvantage.
171

1997]
135
rule,neitherofwhichtermlimitschanges.175
Thus, the incentives
to promiseor providesuch porkwould seem to be no lower for
challengersseekinginitial supportand newcomersseekingreelectionbeforetheyhit theirtermlimitthan they are forlongtermincumbents.
But legislativetenureshouldaffectthe abilityto act on these
motivesbymakingit harderto maintainthe reciprocallogrolling
and mutual deferenceneeded to enact pork.Votingforpork in
otherdistrictscarriesthis risk:ifthe otherlegislatorsdo not reciprocateby votingforporkin yourdistrict,thenyou make your
districtworseofffornothing.Legislatorswill thus approveeach
other'sporkonlyiftheycan strikeand enforcea mutuallybeneficialdeal or patternof logrolling.Term limitsdisruptthis patternin twoways.'76First,theylimittheprospectivetimein office.
This makes it less likelythat the otherlegislatorwill be around
to carryout his end of the bargain.More generally,the shorter
and moreuncertainthe periodof futurerepeat play, the lower
the incentiveto cooperatein the present.'77
Second,termlimits
loweraveragepast tenure.The less timelegislatorshave served
together,the less familiartheywill be witheach other,the more
difficulty
theywill have strikingdeals, and the less theycan deor enforcingdeals. Likewise,the
velop reputationsforfulfilling
mutual deferenceneeded to adopt a universalisticnormis underminedby reducingthe repeated experienceand trust that
seniorityprovides.
In addition,at least some porkis so egregious(it confersa
much larger cost than the benefitto the favoreddistrict)that
only a legislatorwith an extremelylarge seniorityadvantage
could ever get it through.'78
An extremelysenior legislatorcan
obtain such egregiousporkbecause he can offeror threatenfar
175 Getting
ridofdistrictrepresentation
mightbe an alternativereformthatwouldreducepork.However,becauserepresentatives
wouldcovera muchlargerarea, theywould
be less able to adjustforregionalvariations,and voterswouldhave moredifficulty
monitoringthem.The lowerporkwouldthuslikelycomeat the costofincreasingvoters'ideologicaldistancefromtheirrepresentatives.
176See generallyReed and Schansberg,
Impactof CongressionalTenureRestrictions,
in Grofman,
ed,LegislativeTermLimitsat 101-02(citedin note170).
171See Peter Bernholz,
On thestabilityoflogrolling
outcomesin stochasticgames,33
Pub Choice65, 75-78(No 3 1978).At theextreme,termlimitseliminatethe incentivesto
cooperatealtogetherby creatingan incentiveto defectin the finalperiod,knowledgeof
whichcreatesan incentiveto defectin thepenultimate
period,whichbyextensionunravels all cooperationback to the firstperiod.See L.G. Telser,A Theoryof Self-enforcing
53 J Bus 27, 29 (1980).
Agreements,
'78Forexample,SenatorRobertByrdbrought$5 billionin federalspendingto West
Virginiain threeyears,includingtransferring
thenationalfingerprinting
lab ofthe F.B.I.
and movinga significant
partofthe C.I.A. staff.See Calamita,Comment,8 J L & Pol at
575 (citedin note9); Hills,53 U PittL Rev at 145n 196 (citedin note61).

136
The University
ofChicagoLaw Review
[64:83
morein returnthan a juniorlegislatorcan. Because termlimits

in tenure,theyalso lowerthe poslowerthe rangeofdifferences
in relativeseniority.
This shouldpreventsomeof
sibledifference
themostegregiousporkfrombeingenacted.
Still,because the premisethat seniorityaffectsoverallpork
levelsis morecontestablethanthe observationthat seniorityafofany level ofpork,I stressthat the argufectsthe distribution
mentfortermlimitsstandswithoutit. Even iftermlimitshave
no effecton the total level of pork,theystill have the salutary
benefitsof reducingthe pressureto vote forideologicallydiverand ofallowingdistrictsto effectively
engentseniorincumbents
gage in bargainsthat oust seniorincumbentsnoxiousto other
districts.
But, wouldn'tdecreasinglegislativetenurealso reducethe
benefits?Probablyless so. By
amountofnonporkgovernmental
definition,
nonporkis approvednotbecause ofthe concentration
ofits benefitson somedistrictsbutbecause the politicalforcesin
favorofit are generallystrongerthan thoseopposedto it. It is
thusfarmorelikelythata majorityoflegislatorswill independentlyfavorthe nonporkno matterwhatthe otherlegislatorsdo.
dependsless thanporkon the sortofability
Nonporkaccordingly
deals that higheroveralllegislato strikeand enforcelogrolling
tivetenureenhances.Sometimesnonporkwillbe favoredbyonly
oflegislators,whowillthushave to strikea deal with
a minority
otherdistricts.But it shouldbe somewhateasier to do so because, the largerthe ratio of benefitsto costs,the greaterthe
numberofdistrictslikelyto be benefitedand the moretheycan
in a deal.
offer
Moreover,some factorssuggestless senior legislatorswill
enact morenonpork.First,theymay have moretime to spend
enactingnonporkiftheyspendless timeon pork.Second,withouttermlimits,highlyseniorlegislatorscan oftenuse theirgreat
legislativecloutto blocknonporklegislation.By reducingthe divergencein legislativeclout,termlimitslessen the likelihood
oflegislatorshave thecloutto blockthewishesof
thata minority
a majority
coalition.
Nonetheless,it remainspossiblethat a loweraveragelegislativetenuremightcause some,evenifrelativelysmaller,reduction in nonpork.But not all nonporkis desirablein the sense
that its benefitsexceed its costs, under whatevernormative
measure one cares to use.'79Some is undesirable,perhaps because it is pushedby specialinterestgroupswieldingdispropor'See PartII.A.2.

1997]
137
tionateinfluenceacross all districts.180

Thus, even if nonporkis
mostlydesirable,the factthat it is a mixedbag coupledwiththe
factthat its level is less affectedby legislativetenuresuggests
offsetby
that any declinein nonporkis likelyto be normatively
the declinein the indisputablyundesirablepork.If nonporkis
mostlyundesirable,then the decline in its level merelyadds to
oftermlimits.
the desirability
This helps explainwhat to some has seemedan inexplicable
tendencyforleadershipviews on termlimitsto split along liberal-conservative
lines. While term limits supportershave includedsuchleft-liberal
figuresas JerryBrown,Ralph Nader,Bob
Kerrey,Anne Richards,Paul Wellstone,Douglas Wilder, and
Paul Tsongas,liberalleaders are generallyless supportivethan
conservatives.18'
Why should this be when on their face term
limitsare ideologically
neutral?On the otherhand,iftermlimits
are just a matterof liberalversus conservativepolitics,why do
termlimitsattractfar morevotes than conservativesgenerally
do? Afterall, the 75 percentofAmericanswho favortermlimits
in polls'82is muchhigherthan the percentageevervotingRepublican.
The analysis above provides an answer. Term limits decrease ideologicaldivergence,
porklevels,and (perhaps)nonpork
levels as well. Conservativenormativebaselines attachless positive value to governmental
benefitsand morenegativevalue to
coststhan liberalones, thus leadingconservatives
governmental
to classifymoregovernment
actionas porkand morenonporkas
undesirable.Conservativesare accordingly
likelyto findall three
effectspositive,makingtermlimitsan unambiguousboon.183
Liberals would findthe firsttwo effectsdesirablebut not the last,
and wouldfindthe size ofthe secondpositiveeffectsmallerthan
conservativessince theyclassifyfewergovernmental
benefitsas
pork.The conflicting
effectsare likelyto split liberalleaders, as
indeed theyhave been splitin practice.But the firsttwo effects
may well dominateif, as I predict,any decline in nonporkis
180SeePartIV.B.1.
18See Garrett,
81 CornellL Rev at 628 n 12 (citedin note9) (collecting
sourcesnoting
oppositionofleadingDemocrats,and supportin Republicanplatform
and Contractwith
America).
182
Calamita,Comment,8 J L & Pol at 589 (citedin note 9). See also Benjamin
Without
Sheffner,
Hope ofWin,TermLimitsBackersManeuverfora VoteThisFall in the
Senate,RollCall 5 (July27, 1995)(74 percent).
"; Note thatconservatives
withthe goal ofreducingnonporklevelsby loweringaverage legislativetenurecannotachieveit throughordinaryvotingbecause it too poses collectiveactionproblems.Such a goal is thusconsistentwiththeirvoteforbothincumbents
and termlimits.

138
The University
ofChicagoLaw Review
[64:83
small. And in facttypicalpolls show that amongvoters,as opposed to leadership,supportfortermlimitsamong Democrats

(71 percent)and Independents(78 percent)is only marginally
smallerthansupportamongRepublicans(79 percent).'84
Anotherexplanationis partisan strategy:that Republican
leaders favoredterm limitsonlybecause they did not control
Congressand thoughttermlimitswouldimprovetheirodds.But
thisexplanationis hardto square withthe factthatmostDemocratlegislatorscontinueto opposetermlimitsnow that theyare
and mostRepublicanlegislatorsand votersconin the minority
tinueto supportthemeven thoughtheyare now in the congressional majority.Further,termlimitswerepushedeven in states
Indeed, 45
in which Republicanshad legislativemajorities.185
out of
the
party
by
percentof termlimitsbills were sponsored
And Democratvotersweremarpowerin the state legislature.186
ginallyless likelyto votefortermlimitsthan Republicanvoters
thelegislature.187
regardlessofwhichpartycontrolled
of
theoryfocusesnoton theeffect
A relatedpartisanstrategy
termlimitson reelectionsduringthe termlimitperiod,but on
the factthat termlimitscreate open seats when theyoust incumbents.This theorypointsout that,regardlessof whetherit
a partycan do betterwithtermlimits
holdsa majority,
currently
if it does worse than the otherpartywhen its incumbentsare
runningbutbetterthantheotherpartywhenopenseats are conthiswas trueforRepublicansin ConAnd empirically,
tested.188
gressfrom1984 to 1990.189But it is hardto puttoo muchstockin
this because the same relationshipdoes not hold over a longer
8 J L & Pol at 589 (citedin note9). Supportis also not
"See Calamita,Comment,
amongblacks(72 percent)and whites(75 percent),and actuallya little
verydifferent
higheramongwomen(76 percent)than men (73 percent),amongelderly(84 percent)
thanyoung(73 percent),and amongthepoor(77 percent)thantherich(71 percent).Id.
Roll Call at 5 (citedin note182) (notingsupportacrossdemographic
See also Sheffner,
groups); John H. Fund, Term Limitation: An Idea Whose Time Has Come, in Benjamin
and Malbin,eds,LimitingLegislativeTerms225, 225 (citedin note5) (quotingCBS poas statinghe had "'neverseen an issue on whichtherewas so littledemoliticaldirector
in percentid at 227 (earlierpollshowingonly5 percentdifference
graphicvariation'");
fortermlimits).
ages ofliberalsand conservatives
' See, forexample,Calamita,Comment,
8 J L & Pol at 589-90n 158(citedin note9)
(Colorado); Daniel Friedman and Donald Wittman,Why voters vote for incumbentsbut
against incumbency: A rational choice explanation, 57 J Pub Econ 67, 77 (1995)
(Wyoming).
'86Benjamin and Malbin, Term Limits forLawmakers, in Benjamin and Malbin, eds,
LimitingLegislative Terms at 7 (cited in note 12).
'87 See Keith Boeckelman and Gina Corell, An Analysis of Term Limitation Elections,
in Grofman,
ed,LegislativeTermLimits185,190(citedin note5).
" See Gilmour and Rothstein,Model of Partisan Balance, in Grofman,ed, Legislative
TermLimitsat 156(citedin note139).

forstateraces.
No data are offered
1' Id at 159-60.

1997]
139
time span. From 1970 to 1990, or from1954 to 1990,Democrats

were farbetterat winningopen seats than Republicanswere.'90
Republicansinterestedin termlimitsforpartisanadvantage in
1990 wouldthushave had to be gamblingthatthe priorsix years
a betterforecastofthefuturethanlongertimeperiods.
offered
Moreover,because the open-seattheoryclaims an effectnot
in each electionbut onlyin the open-seatelectionsfollowingthe
last termunder termlimits,the effectis limited.The theory's
proponentsthemselvesestimate that twelve-yearterm limits
would gain Republicansfourseats in a 435-memberCongress.191
It seems doubtfulthatminorpickupcouldbe the main motivator
forthe termlimitsmovement.Nor does the theoryexplainwhy,
if advantagingRepublicanswere the main motive,supportfor
termlimitsremainsso highamongDemocrats.Afterall, in state
legislatures,44 percentof term limits bills were sponsoredby
Democrats.'92
And Democratvotersare almost as likelyto support termlimitsin polls and electionsas Republicans.193What
mightmotivatethemis leftunexplainedbythe partisanstrategy
theory.
4. Alternativecollectiveagreements.
The existenceof a collectiveaction problemdoes not necessarilymean a particularcollectiveagreementis justifiable.There
may be alternative,less restrictive,
means ofcollectivecoordination. If votersin different
districtscould reach implicitagreementsto get rid of onlyincumbentswho are ideologicallydiver'90Id (1970 to 1990);Will,Restoration
at 218 (citedin note 10) (1954 to 1990). Democratsdid,however,do betterin incumbentraces even in longertimespans.See Gilmour
ModelofPartisanBalance,in Grofman,
and Rothstein,
ed,LegislativeTermLimitsat 160
(citedin note139).The authorsdo not,however,explorewhythismightbe so. The analysis heresuggeststhatit is eitherbecause beingin the majoritypartyexacerbatesincumnote138,or becauseRepublicansare less likely
bencyadvantages,see textaccompanying
to favorspendingon distributable
governmental
benefitsand thustheirincumbency
confersless ofan advantage,see Part II.B.2. Evidencefromhow incumbentRepublicansdo
in the 1996electionmayenableus to choosebetweenthesehypotheses.
191See Gilmour
and Rothstein,
ModelofPartisanBalance,in Grofman,
ed,Legislative
TermLimitsat 164 (citedin note 139) (based on long-term
data from90th-lOOth
Congress).
"92Benjaminand Malbin,TermLimitsforLawmakers,in Benjaminand Malbin,eds,
LimitingLegislativeTermsat 7 (citedin note12).
193
See textaccompanying
note 184 (notingpolls showing71 percentDemocratsupport,79 percentRepublican);JohnDavid Rausch,Jr.and GaryW. Copeland,TermLimits
in Oklahoma,California,and Coloradoin 1990,in Grofman,
ed, LegislativeTermLimits
199,208-10(citedin note5) (One Californiatermlimitsproposalgot50 percentofDemocrat votes and 51 percentof Republicanvotes; Coloradotermlimitsgot 63 percentof
Democratvotes,73 percentofRepublicanvotes;Oklahomatermlimitsgot 73 percentof
Democratvotes,79 percentofRepublicanvotes.).

140
The University
ofChicagoLaw Review
[64:83
or highlynoxiousto otherdistricts,they
gent,pork-providing,
couldgain the benefitsoftermlimitswithoutsuffering
as often
fromhavingnovicesin office.
However,such agreementswould be impossibleto reach,
muchless enforce.Too manyvoterswouldhave to be involvedto
reach any collectiveagreement.Moreover,because votingis siand is by
multaneous,involvesmassivenumbersofparticipants,
secretballot,votersin one stateor districtcannotobservethe actions of votersin anotherdistrictbeforemakingtheir choice.
Hence theycouldnotknowwhetherany such implicitagreement
was beingkeptbythe otherdistrictsuntilafterthe election.Nor
wouldmembersin such a diffusegrouphave reputationaleffects,
or institutionalarrangementsto instrategicinterdependence,
surecooperation.
Termlimitsare a second-bestsolutionto this problem.Because they offera concreteprospectivestandard, collective
agreementto termlimitsthroughstateballotinitiativesis possible. And enforcement
requiresno ongoingmonitoring
by other
votersor simultaneouscoordination,
butjust the applicationofa
simplerule. The downsideis that termlimitsare overinclusive
and underinclusive.Overinclusivebecause they sometimes
screenout seniorincumbentswho are ideologicallycompatible
withtheirdistrictsand notproneto porkprovisionor noxiousto
otherdistricts.Underinclusive
because collectiveactionproblems
inremainthatimposesome pressureto voteforpork-providing
cumbentsdespitesomeideologicaldivergence.But all legal rules
And since we cannot hope to
are over-and underinclusive.194
achievethe first-best
solution,it hardlyserves as a meaningful
standard for critique.Reducingthe magnitudeof ideological
slack and amountoftotalporkmorethan sufflcestojustifyterm
limits.
Alternatively,
we mighttryto tackleparticularproblemslike pork-witharguablymoretailoredremedies.But porkis exto define,sinceit requiresa commonnormative
tremelydifficult
baseline formeasuringthe benefitsand costs of legislationnationwideand withindistricts.Moreover,
proceduralremedieslike
theline itemvetomaynotbe so well-tailored
afterall. An executive can, afterall, use the line item veto when porkis not involvedand refrainfromusingit whenporkis involved.It too is
thusunder-and overinclusive.
Indeed,some analystsarguethat
"'See generally Stephen McG. Bundy and Einer Elhauge, Knowledge about Legal
Sanctions, 92 Mich L Rev 261, 268-79 (1993); Einer Elhauge, The TriggeringFunction of
Sale of ControlDoctrine,59 U Chi L Rev 1465, 1495-97 (1992).

1997]
141
the line item veto does not reduce legislativepork but merely
shiftspowerfromthe legislativeto executivebranch.The main
reason is that executivesseekingvotes fora bill have incentives
to promisenotto exercisetheline itemvetoagainst an individual
legislator'sporkif that legislatorwill supportthe bill.'95In any
event,even if the line itemveto would reduce porklevels somewhat, that does not eliminatethe advantage of reducingthem
furtherwith termlimits.Nor would it negate all the otheradvantagesoftermlimitsdiscussedin Parts II and III.
mightseem anotheralternativemeasPeriodicredistricting
ure. But it wouldbe disruptiveand wouldalso notbe possiblefor
such as U.S. Senators.Moreover,thoughwe wouldexall offices,
pectit to resultin the morefrequentousteroflegislators,thatis
because theywouldhave a hardertimecomingclose to the political views of shiftingelectorates.There would still be seniority
advantagesand thus the penaltyforrejectinga seniorlegislator
would remain.Indeed, studies have shownthat the decrease in
in legislativeraces has been no higherin states
competitiveness
than in statesthat did not,thus suggestingthat
thatredistricted
currentgerrymandering
is not the main cause of the declinein
competitiveness.196
C. VariationsAmongDistricts,Eras & PoliticalJurisdictions

Not all districtssupporttermlimitsand some do so morevociferouslythan others.Nor are term limits equally popular at
everytimein historyor in everystate or nation.If we extendthe
in certainfeatures,we
above analysisby allowingfordifferences
can help predictand explain such variations.Most of what folin theAppendix.
lowsis also derivedmathematically
1. Varyingdistricts.
My basic modelassumed all districtswere similar,but obvidistrictsvaryin the
ouslythis does not hold. Most noteworthy,
seniorityof their representatives.They may also vary in how
much they care about ideologicaldivergencecomparedto gov"9'See, for example, Maxwell L. Stearns, The Public Choice Case Against the Item
Veto,49 Wash & Lee L Rev 385,417-18(1992).
"See Charles R. Kesler, Bad Housekeeping: The Case against Congressional Term
Limits, in Benjamin and Malbin, eds, Limiting Legislative Terms 241, 246 (cited in note
its seniorfedto redistrict

5). Moreover,any one statehas collectiveactiondisincentives
eral legislators.See Part III.D. Withtermlimits,it becamepossibleto coordinateparallel
coordination
in multiplestates because termlimitsoffereda six- to twelve-year
efforts
periodis not possiblewithredistrictperiod.See Part II.D. But such a longcoordination
aftereach census.
immediately
takeeffect
ingbecausethenewdistricts

142
The University
ofChicagoLaw Review
[64:83
in how muchofthe costs ofporktheybear,

ernmentalbenefits,
in
the
ofnonporktheydeemsociallydesirable.
and
proportion
A districtwith less senior than average representation
shouldfavortermlimitsfortwo separatereasons.First,as with
all districts,termlimitsreducethe pressureto voteforan ideologicallydivergentincumbent.Althougha junior incumbent
should be less ideologicallydivergentthan more seniorincumbentsin otherdistricts,a districtis penalizedforlosingany degree of seniorityit has. Moreover,the penaltyis larger the
greaterthe seniorityof the othermembersbecause what deterA newcomerthus
minesporkdistribution
is relativeseniority.197
suffersa lower senioritydisadvantagewith term limits than
has less
withoutthem.Accordingly,
evenifa district'sincumbent
tenurethan the maximuma termlimitwould set, termlimits
still decreasethe district'sideologicalslack by limitingthe seniorityofotherdistricts'legislatorsand thus reducingthe size of
the penaltythe districtwould sufferif it oustedits ownlegislator. This diminishedslack need not result in the ousterof an
ideologically
divergent
incumbent.Often,it will insteadresultin
an ideologicalshiftbythatincumbentto take accountofherlowered slack. Eitherway,the districtbenefitsfromloweredideologicaldivergence.
Second,a districtwithless seniorthan averagerepresentationreceivesless than an average share of governmental
benefits.Termlimitswillbringits seniorityand shareofgovernmental benefitscloserto average. This makes the districtbetteroff
evenifit does notexperience
or care about ideologicaldivergence
and evenifdecreasedseniority
does notdecreasetheoveralllevel
A districtcan thusfindaffirmative
ofpork.198
value in gettingrid
ofseniorrepresentatives
fromotherdistrictseven if termlimits
leave the districtno happierwithits own representation.199
This
helps explainwhypro term-limits
advertisingin manydistricts
oftentargetsa fewprominent
legislatorsfromotherdistricts.In
'97SeePartILIB.1.
See also Appendix.
Ifdecreasedaveragelegislativetenuredoesdecreasetheoverall
levelofpork,thatfeatureis all the moreattractiveto suchdistricts
becausetheysuffer
mostfromhighporklevels.
'" See Elhauge,WallSt J at A14 (citedin note115). See also Friedman
and Wittman,
57 J Pub Econat 70-73(citedin note185) (notingthismotivation
fortermlimits).In contrastto thisArticle,
Friedmanand Wittmanassumethattheonlymotivefortermlimits
is to redistribute
power(herefromseniordistricts
tojuniordistricts)
and ignoretheways
in whichtermlimitscan solvevariouscollectiveactionproblemsand makegovernment
workbetterto thebenefit
ofall districts.
See id at 81. Theythuscan explainsomeofthe
variationin supportin different
districts
butnottheoveralllevelofsupport.See id at 82.
See also id at 72, 75 (conceding
thateffect
ofdistrictseniority
is weak,on orderof1 to 3
percent).

1997]
143
California,for example, term limits proponentspitched term

limits statewide as a way of gettingrid of AssemblySpeaker
WillieBrown,so muchso that his legal challengeto termlimits
claimedtheywerea "billofattainder.
"200
Whethera districtwithmoreseniorthan average representationwill favoror oppose termlimitsis less clear. On the one
hand,it sufferseven greaterideologicalslack than districtswith
morejunior representation.Thus, term limits would confera
greaterbenefiton it. On the otherhand, a seniordistrictwould
suffersome positivecosts because termlimitswould reduce its
share of any governmentalbenefitsdistributedby tendingto
equalize seniority.Whetherthe benefitoutweighsthe cost depends on the particulardistrict.201
Consistentwith these conflictingeffects,seniordistrictstend to oppose termlimitsmore
oftenthanjunior districts,but not by a verylarge margin.One
studyof the Californiavote fortermlimits,forexample,found
thatdistrictswithseniorlegislatorswere only1 to 3 percentless
likelyto vote forterm limitsthan districtswithjunior legislators.202
Wherea seniordistrictfallswill likelydependon how much
it cares about ideologicalfitcomparedto porkand how muchof
the costsofporkthe districtpays. If it does not care about ideologicalfitat all, it will enjoyno benefitfromreducingideological
divergence.It will thus surelybe a net loserifoverallporklevels
are unaffected
by termlimits.If termlimitsreduce porklevels,
the loss of its above-averagepork share may be offsetby the
greaterefficiency.
But not always.And even whenit is, the overall resultdependson what share ofporkcosts it pays. If,forexample,a districtpays littleofthe taxes fundingpork,the district
does not benefitfromlower overall pork levels forit is a net
gainerfromporkgenerallyand notjust in its district.Poor districtsmightdisproportionately
have this profileiftheypay little
taxes and disproportionately
value porkbecause theydo nothave
the luxuryofexpressingpurelyideologicalbenefitat the expense
' See Charles M. Price,The GuillotineComesto California:Term-Limit
Politicsin
theGoldenState,in Benjaminand Malbin,eds,LimitingLegislativeTerms117, 121, 125
(citedin note5).
' See also Appendix
(makingsame pointmathematically).
Even if the collectiveactionproblemsalone wouldmake seniordistrictson balanceless enthusiasticthanjunior
districtsabouttermlimits,entrybarrierproblemsshouldoffsetthis to some extentbecause theymake termlimitsmoreattractiveto seniordistricts.See textaccompanying
notes290-93.
21)2
Friedmanand Wittman,57 J Pub Econ at 75 (citedin note 185). One caveat
aboutthisevidence:because the termlimitswereprospective
and thus did notoust any
legislatorsforsix years,it is notclear that the districtswithseniorlegislatorsin 1990
whentermlimitswereenactedexpectedto have seniorlegislatorsin 1996.

144
The University
ofChicagoLaw Review
[64:83
of economicinterest.203
Richerdistricts,in contrast,mightdisproportionately
have the oppositeprofile:payinga higherthan
averageshare oftaxes and thus porkcostsand gaininga belowaverage benefitfrompork comparedto ideologicalcorrespondence.This mightfurtherexplainwhy,even thoughtheirideological contentis neutralon the surface,termlimitsare more
likelyto be opposedbyliberalsthanbyconservatives.204
In any event,a districtwithrelativelylittleideologicalconcernand a low share ofporkcostscan standa muchlargerideologicalgulfbetweenit and its representative,
and wouldthusbe
and a large share of
expectedto have veryseniorrepresentation
pork.It wouldalso predictably
opposetermlimitsand do worse
on other
underthem.But the districtimposesclear externalities
it getsonlybecause
districts.It wantsthe typeofrepresentation
the costs ofit are borneby the otherdistricts.Those otherdistrictsmightthusturnto termlimitsto ridthemselvesofthisexternality.
Finally,districtsmightvaryin theirbeliefsabout the proportionofnonporkthatis sociallydesirable.Districtsthat think
is low (the conservativedistricts)are morelikely
the proportion
Districtsthatthinkthe proportion
is high
to favortermlimits.205
(the liberaldistricts)are morelikelyto opposetermlimits,but
maywell stillbe attractedbecause theyfindany effecton nonofthe declinesin porkand
porklevelsoffsetbythe combination
ideologicaldivergence.206
2. Varyingtimesandjurisdictions.
Legislativeterm limits have not been equally popular in
everypoliticaljurisdictionor everyhistoricalperiod.Theyhave
passed in somestatesand notothers.Theyhave been popularin
theUnitedStates butin fewothernations.207
Andtheyhave only
districtsmighthave similarincentivesiftheyreceivemuchporkbut pay
"3Elderly
littlein taxes.
See also Part II.B.3. Anotherpossibleconnection
is that(perhapsbecausetheVotingRightsAmendment
attemptsto securesafeseats forminorities)
reelectionratesare
higherforblacksthanwhites.See DanielA. Farber,Review,BlackFaces,BlackInterests:
The RepresentationofAfricanAmericans in Congress, 11 Const Comm 613, 619-20 (1995)
(reviewing
CarolM. Swain'sBlackFaces,BlackInterests).
Thisshouldtendto makeblack
representatives
moresenior.Ofcourse,forthereasonsnotedabove,seniordistrictswill
notnecessarily
opposetermlimitsbuttheyare morelikelyto do so thanjuniordistricts.
See also note184(notingthatsupportfortermlimitsis onlyslightly
loweramongblacks
(72 percent)thanwhites(75 percent)).
25 See PartHL.B.3.
206Id.
2 See Carey,Term Limits in Costa Rica, in Grofman,
ed,Legislative Term Limits at
321 (citedin note 132) (The onlynationswithlegislativetermlimitsare Costa Rica,

1997]
145
recentlybecomepopular in the United States. Before 1990, no

statehad legislativetermlimits.From1990 to 1995,twenty-four
statesenactedthem.Whatexplainsthisvariationand timing?
The ideologicalslack reduced by termlimits is higherthe
longerthe prevailinglevels of incumbenttenure because such
longtenuresincreasethe penaltyin each districtofreplacingan
I would thus expect overall enthusiasmforterm
incumbent.208
limitsto be higherwhen tenureand reelectionlevels are high.
And thoselevels were steadilyincreasingin the years leadingup
to 1990. In the state legislatures,turnoverrates that as late as
the 1960s had been in the 30 to 40 percentrangedroppedto 16 to
20 percentby 1988, and reelectionrates oftenexceeded 90 perIn some states,turnoverwas farlower.In Colorado,one
cent.209
of the firstthreestates to enact legislativetermlimitsin 1990,
In anotherof
no state senatorhad lost reelectionin ten years.210
these threestates, California,only3 out of 273 had lost in the
priorsix years-none in 1988.21
In Congress,reelectionrates were even higherthan in the
average state legislature.In the periodbefore1990, House reelectionrates were over90 percentforsixteenyears runningand
This was unpreceover 95 percentfor eight years running.212
dentedin U.S. history.Indeed,in the 150 years before1968, the
reelectionrate had neveronce exceeded95 percent.213The term
limitsmovementtook offafterthe 1988 elections,when an alltime high of 98 percent of House incumbentsrunningwere
reelected,a rate higherthan reelectionrates in the old Soviet
Mexico,Ecuador,and thePhilippines.).
See Appendix.
'Benjamin and Malbin,TermLimitsforLawmakers,in Benjaminand Malbin,eds,
LimitingLegislativeTermsat 5 (citedin note12); David H. Everson,TheImpactofTerm
Limitationson the States: Cuttingthe Underbrushor Choppingthe Tall Timber?,in
Benjaminand Malbin,eds, LimitingLegislativeTerms189, 191, 194 (cited in note 5);
in Benjaminand Malbin,eds,LimitingLegislativeTermsat 232Fund,TermLimitation,
et al, For WhomtheBell Tolls: TermLimitsand
33 (citedin note184); GaryF. Moncrief,
in Grofman,
ed,LegislativeTermLimits119,127 (citedin note5).
StateLegislatures,
210Benjaminand Malbin,TermLimitsforLawmakers,in Benjaminand Malbin,eds,
LimitingLegislativeTermsat 5 (citedin note 12); Fund,TermLimitation,in Benjamin
and Malbin,eds,LimitingLegislativeTermsat 233 (citedin note184).
211Benjaminand Malbin,TermLimitsforLawmakers,in Benjaminand Malbin,eds,
LimitingLegislativeTermsat 5 (citedin note 12); Benjaminand Malbin,eds, Limiting
LegislativeTermsappendixA at 276 (citedin note5).
212Benjaminand Malbin,eds, Limiting
LegislativeTermstable B-2 at 293 (cited in
note5).
whenthe reelection
213 Id at 291-93.Therewere a coupleofyearsin our earlyhistory
rotationwas moreprevalentand no
rateexceeded95 percentbutin thoseyearsvoluntary
ran forreelection.Id at 291. The overallturnover
morethan sixty-eight
representatives
ratein thoseyearsrangedfrom30 to 46 percent,whichcomparesto 8 to 12 percentturnoverfrom1983to 1991.

146
The University
ofChicagoLaw Review
[64:83
Indeed, duringthe 1988 to 1990 period a federal

Politburo.2"4
legislatorwas statisticallyalmost as likelyto be sentencedto
prison as to lose reelection:five were sent to prison,six lost
Andofthosesix wholost,fivewereundersomesort
reelection.2"5
Turnoverwas higherin Britain'sHouse of
of ethical cloud.2"6
Lords.2"7
Such highreelectionrates naturallyresultin highlevels of
noncollectenureand thushighpenaltiesforoustingincumbents
tively.Between1953 and 1991,the averagetenurewentfrom9.8
to 12.4 yearsin theHouse,and from8.5 to 11.1 yearsin the Senate.2"8The effectsat the highestlevels of senioritywere even
who
more striking.The House went from119 representatives
that
had served12 or moreyearsto 198.219(Beforethis century,
numberhad never exceeded23.220) The Senate went from19
senatorswith 12-yearsseniorityto 49.221 The early 1990s thus
markedthe firsttimein historythat the numberof legislators
was nearly50 percentin bothchambers.
with12-yearsseniority
the enthusiasmfortermlimitsshoulddiminish
Conversely,
the lowerreelectionand turnoverrates are. This is one reason
the term limit movementpredictablylost steam after many
freshmentook officein the 1994 congressionalelection.And it
framerswere less interestedin
may be whythe constitutional
termlimits.At thattime,tenurein the new Congresswas zero,
Moreover,the strongnorm,
and past turnoverhad been high.222
the exampleofGeorgeWashingtonand ThomasJefferfollowing
Throughrotateout oflegislativeoffice.223
son,was to voluntarily
8 J L & Pol at 568 (citedin note9); CaliforniaBallotPamComment,
2'4Calamita,
in Benjaminand Malbin,eds,LimitingLegislativeTermsappendixA-5at
phlet,reprinted
275-76(citedin note5). The firsttermlimitsinitiativeswerepassed in 1990,the first
electionyearafterthe1988elections.
215Lee Iacocca,WeCan'tEvenThrow
theRascals Out,LA TimesB7 (May18,1990).
216See Fund, Term Limitation, in Benjamin and Malbin, eds, Limiting Legislative
Termsat 230-31(citedin note184).

BallotPamphlet,reprinted
in Benjaminand Malbin,eds, Limiting
217See California
LegislativeTermsappendixA-5at 275-76(citedin note5).
218 See
Will,Restorationat 74-76(citedin note10).
219
Id at 80.
22 Id at 78-79.
223Id at 83.
'See PartI.A.2.
2
See Petracca, Rotation in Office,in Benjamin and Malbin, eds, LimitingLegislative
in Benjaminand Malbin,eds,
Termsat 35-38(citedin note10); Fund,TermLimitation,
LimitingLegislative Terms at 237 (cited in note 184); Tabarrok, Term Limits and Political Conflict,in Grofman,ed, Legislative Term Limits at 243-44 n 7 (cited in note 150).
tenurein the past may have been that lower

Anotherreason forlowercongressional
moredifficult.
increasedtravelcostsand madekeepingin touchwitha district
technology
In addition,thecapitalcityonlyrecently
becamea pleasantplace to live,see Will,Restosessionslastedtwoto
rationat 11-12(citedin note10),and untilthe 1930s,congressional

1997]
147
out the nineteenthcentury,congressionalturnoverwas typically

Even AbrahamLincolnleftafterone
40 to 50 percentper term.224
termin Congressunderinformalrotationnorms.225
Likewise,the larger the total amount of distributablegovernmentalbenefitsper district,the larger the penaltyfor not
having a senior representativeand the greaterthe ideological
slack.226
States or nationswithlarge amountsoftotal distributable governmental
benefitsthus have greaterincentivesto favor
termlimitsto reducesuch ideologicalslack. Such politicaljurisdictionswould also benefitmore fromany pork reductionproduced by an overalldecrease in seniority.Accordingly,
since the
smaller size of government
made distributablebenefitsfar less
in the late 1700s, the collectiveactionproblemmotisignificant
vatingtermlimitswas likewisefarless pressingduringthe constitutionalframingthanit is now.227
Indeed,long tenureand high porklevels are mutuallyreinforcing.
Elevatedporklevels are likelyto increaseaveragetenure
byimposinga greaterpenaltyon incumbentreplacement.Longer
averagetenureincreasesthe amountofporkproduced.This suggests the pressurefortermlimitsin the UnitedStates will continueto grow,notabate, in the future.
The enthusiasm for term limits will also be higher the
greaterthevalue thatvotersput on ideologicalfit.228
Termlimits
should thus be more popular in more ideologicallycontentious
times.This helps explainthe suddenpopularityoftermlimitsin
1990 on the heels ofthe end ofa Cold War periodthat tendedto
submergedomesticideologicalconflict.To the extentvotersbelieve decreasingaveragelegislativetenurewill lowerthe level of
nonporkdistributablegovernmentalbenefits,term limits will
also be morepopular the more conservativevotersare in their
beliefsaboutthe proportion
ofnonporkthat is sociallydesirable,
as votersrecentlyseem to have become.
threemonths,requiringthe legislatorto returnto some otherline of work,see Fund,

Term Limitation, in Benjamin and Malbin, eds, Limiting Legislative Terms at 228 (cited
in note 184). In state legislatures,past highturnoverhas been attributedto lowerpay,

smallerstaffs,and shortersessions,all of whichwere purposefully
changedin reforms
beginningin the 1960sthat by the 1980s producedthe low turnoverthat inspiredterm
limits. See Everson, Impact of Term Limitations, in Benjamin and Malbin, eds, Limiting
LegislativeTermsat 191-94(citedin note209).

4Kesler,Bad Housekeeping, in Bernardand Malbin,eds,Limiting Legislative Terms
at 244 (citedin note196).
225Id.
See also Appendix.

See also PartI.A.2.
228 See also Appendix.
226
227

148
The University
ofChicagoLaw Review
[64:83
It bears emphasis that the collectiveaction problemsdescribedabove applyonlyforpoliticaljurisdictionsthat elect legislatorsby district.Not everynation elects legislatorsin this
fashion.For example,to the extenta nation uses a systemof
proportional
representation
wherelegislatorsare chosenfroma
partylist,thenthe effectsofrelativesenioritydo notcoercevoters toretainseniorincumbents.
One wouldthusexpectthatterm
limitswouldbe less popularin suchnations.229
Likewise,ifa city
elects all its commissionersat-largethrougha citywidevote
ratherthanby precinct,thenthe cityis less likelyto favorterm
limits.230
In addition,the collectiveactionproblemsrequirethat seniorityconferextralegislativecloutmakingit especiallydifficult
In parliamentary
fordistrictsto give up seniorrepresentatives.
systemslike Britain's,whereseniorrepresentatives
maybe releelevated to
gated to the back benchand junior representatives
the government,
the connectionbetweenseniorityand clout is
less reliable.One wouldthusexpecttermlimitsto be less popular in suchparliamentary
systemsthanin systemsthatseparate
theexecutiveand legislativepowers.
D. The CollectiveActionProblemofState Limitson Federal
Officials
The analysisso farprovidesan explanationforwhya state
mightpass termlimitson statelegislators,or whythe entirenationmight(throughconstitutional
amendmentor,ifpermissible,
federalstatute)pass termlimitson federallegislators.In those
instancesthe collectiveactionis takenat the level ofthe collectivityaffectedby the legislature,so that the votershave incentivesto weighthe fullbenefitsand costsoftermlimits.But why
do statevoterseverpass termlimitson theirstate'srepresentativesto Congress?Afterall, an individualstate decidingwhether
to limitthe termsof its U.S. Senatorsseems subjectto similar
collectiveactionproblemsas a statedecidingwhetherto voteout
of officea particularsenior senator.23'That is, if a particular
2"However, sometimes parties in a proportionalrepresentationgovernmentcan organize themselves to providepork by dividingjurisdictionsinto subunits, each of which is
assigned a deputy to provide particularized benefits.See Carey, Term Limits in Costa
Rica, in Grofman,ed, Legislative Term Limits at 328-31 (cited in note 132) (findingthat
Costa Rican parties organize in this fashion). In such a country,term limits mightbe attractiveto reduce pork levels, and have in factbeen used in Costa Rica.
" The entrybarriers described in Part III may, however, provide sufficientgrounds
fortermlimitseven in such cities.
2 'A state deciding whether to reelect a senior member of the House of Representatives is subject to somewhat greater collectiveaction problemsbecause the smaller size of

1997]
149
state passes termlimitsthat oust its federalofficials,it lessens

its seniorityand share of legislativepowerwhetheror not the
otherstates have enacted termlimits on theirfederalofficials.
Whatevercollectiveaction problemspressure them to reelect
senior Senators thus seem likelyto also pressure them to nix
termlimitsthatwouldoustthosesame Senators.
This explains whystate-enactedtermlimitsnevertake the
formof retroactivelimits that would immediatelyoust senior
of them have been profederalofficials.Rather,all twenty-four
spectivelimitson the futurenumberof years those legislators
can serve. If,forexample,a legislatorhas already servedeighteen years, he can serve another twelve under a prospective
termlimit.Votersknowthe difference.
The onlylegtwelve-year
islativetermlimitsproposalsever defeatedby voters-in Washingtonin 1991 and Utah in 1994-were also the onlyones that
And those measures lost bewould have applied retroactively.232
cause offearthat the relativelegislativecloutofthe state's congressionaldelegationwould sufferfromretroactivelimits.Term
limitsopponentsmade such argumentsthe centralfocusoftheir
opposition,stressingthatthe state wouldlose seniorityand leadershippositionsin Congress,includingthe Speaker ofthe House,
and thus be committing
"unilateraldisarmament"by lessening
the state'slegislativeclouton federalissues ofparticularinterest
In Washington,where the most detailed polling
to the state.233
districts(1) makesthe ratioofdistributable
benefitsto costsincurredin thejurisdiction
larger,(2) means a largershare of any distributable
benefitreductionmustbe shared
coopwithotherjurisdictions
whethertheycooperateor not,and (3) makes anyinformal
erationeven less feasiblebecause 435 ratherthanjust 50 regionswouldhave to coordidelegation
nate. Still,a state decidingwhetherto adopttermlimitsforits congressional
facesa collectiveactionproblemsimilarin kind(thoughless in degree)to thatit facesin
decidingwhetherto reelectthem.
2 In thirty-four
opportunities,
votershave rejectedtermlimitsonlythreetimes:once
each in Washington,
Utah, and Mississippi.See Bill Hord,Outside Cash Backed Term
Limits Effort,Omaha WorldHerald 9SF (Sept 16, 1996). The rejectedWashingtonand
Utah initiativeswereretroactive,
id, and theUtah votealso reflectedthe factthatUtah
termlimitslaw witha triggerclause. See notes238, 243. The
alreadyhad a prospective
Mississippirejectioncame afterthe SupremeCourtinvalidatedthe limitson federalofficials, and did not involvea pure legislativetermlimitsproposalbecause the initiative
would also have limitedthe termsof local officialssuch as sheriffswho do not have
genreelectionsthatpose collectiveactionproblems.This extensionto suchlocal officials
eratedmuchofthe oppositionto theMississippiinitiative.See Reed Branson,Term Limits Backers Vow Narrower Effort,The CommercialAppeal B2 (Nov 9, 1995); Editorial,
Scaring Mississippi Voters,Wall St J A18 (Oct 30, 1995).See also Part II.C.2 (notingthat
thecollectiveactionproblemsariseonlyforelectionsto representdistrictswithina larger
politicalentity).All threevotesare thusconsistentwiththe collectiveactionexplanation
put forthhere.Note thatthe politicalentrybarrierproblemdiscussedin Part III could
sincestateswouldhave incentivesto reduceentrybarrinotalone explainprospectivity
ers as soonas possible.
2> Olson,Term Limits Fail, in Benjaminand Malbin,eds,Limiting Legislative Terms

150
The University
ofChicagoLaw Review
[64:83
data were collected,polls beforethe electionshowedthat large

percentagesof potentialsupportersof termlimitscould be dissuaded by the clout argument.234
And polls conductedafterthe
defeatof retroactivetermlimitsshowedthat 52 percentof the
"no")voteswere cast because ofone versionor otherofthe relaFifteenpercentof those voting"no" said
tive cloutargument.235
ifthetermlimitswereimposed
theywouldhave voteddifferently
nationwide.236
When presentedwith a proposalforprospective
termlimitsin 1992,Washingtonvotersacceptedthem.237
And in
Utah,thevoterrejectionreflected
the factthatUtah alreadyhad
termlimits.238
adoptedprospective
Of course,prospectivetermlimitsimposedby a state on its
officials
are notthe same as termlimitsimposednationwide.But
suchprospective
termlimitshave been enactedas partofsimulStatetaneous termlimitsmovementsin manyotherstates.239
imposedtermlimitscan thusbe seen as an offerbythe enacting
states to cooperatein removinglong-termincumbentsonly if
otherstatesdo the same. If otherstatesdo notrespondby passcan be withdrawnby reing similarlimits,the original"offer"
pealingthetermlimitsin thosestatesthatfirstenactedthem.
The prospectof such repeal is not an academicfantasy.It
And
was explicitlycontemplatedby term limits advocates.240
at 222 (citedin note10);TonySemerad,
at 79-82,92-93(citedin note5); Will,Restoration
Rivals Battling Over Ballot Measures, Salt Lake Trib Bi (Nov 6, 1994); Dan Harrie, Term
LimitsDraw FreshOpposition,
Salt Lake TribBi (Oct 19, 1994);Editorial-Commentary,
UtahHas TermLimitsNow,Salt Lake TribA30 (Sept 2, 1994).Washington
state was
in federalissuessincethefederalgovernment
particularly
interested
owned29 percentof
itslandand had strongcontroloverits waterand electricity.
Olson,TermLimitsFail, in
Benjaminand Malbin,eds,LimitingLegislativeTermsat 79-82,92-93(citedin note5).
'Olson, Term Limits Fail, in Benjaminand Malbin,eds,Limiting Legislative Terms
at 83 (citedin note5).
35 Id at 85 (21 percent
citedlostcloutofstate,19 percentthatstatewouldbe doingso
alone,8 percentlosingSpeakerofHouse,and 4 percentlosinginfluence
on variousissues
affected
byclout).
See Will,Restoration
at 222-23(citedin note10).
"Brief forthe State Petitioner,
US TermLimits(Nos 93-1456,93-1828),1994 WL
444683at *3 (citedin note10).
'See
Tony Semerad, Term Limits Aren'tDead Utah Demos, Salt Lake Trib Al (Nov
10, 1994).Alsoconsistent
withthecollectiveactionexplanation,
theexistingtermlimits
law containsa trigger
clause.Id. See also note243 and accompanying
text.
"'Termlimitsactiviststhemselvesrecognized
thattheyexacerbatedcollectiveaction
concernsaboutlosingrelativelegislativecloutin their1991loss becauseWashington
was
theonlystatevotingon theissue thatyear.StuartRothenberg,
Transplanting
TermLimits: Political Mobilization and Grass-RootsPolitics, in Benjamin and Malbin, eds, Limit-
ingLegislativeTerms97, 110 (citedin note5). Theyput termlimitsinitiativeson the

ballotin fourteen
statesin 1992,all ofwhichthenwonpassage,including
in Washington.
See Boeckelman and Corell,Analysis of TermLimitationElections, in Grofman,ed, Legislative TermLimits at 185 (cited in note 187).
Transplanting Term Limits, in Benjamin and Malbin, eds, Limiting
240Rothenberg,

1997]
151
there was related precedentforit: in 1790, Pennsylvaniawas

forcedto repeal its termlimiton its federalrepresentativesbeforeit took effectin the new Congresswhen no otherstate folMore recently,
lowed suit underthe new federalConstitution.24'
whena majorityofstates failedto followits lead in enactingfederal termlimitsby 1995, Coloradoreactedby delayingthe start
date forcountingprospectivetermlimits from1991 to 1995.242
The identityofthe state is telling.Coloradowas the firststate to
face the possible loss of clout because, of the three states that
had enacted termlimitsin 1990, it was the onlyone to impose
themon its congressionaldelegation.The issue has sincebecome
mootforstate-enactedtermlimitssince US TermLimitsinvalidated theirapplicationto Congress.
As the term limits movementbecame more regularized,
states even startedto includeexplicit"triggerclauses," making
termlimits on federallegislatorseffectiveonly aftera certain
numberofotherstates also imposedsuch limits.243
This was a diofthe rootcollectiveactionproblem.Indeed,one
rectrecognition
ofthe states was Washington,whereclout argumentstorpedoed
the firstattempt.Anotherof the states witha triggering
clause
was Utah, and voters rejected an initiativethat would have
droppedthis triggerclause in large part because ofopponentartermlimitswouldcostthe state
gumentsthat suchunconditional
cloutin Congress.2"Likewise,in Missouri,the triggering
provision was responsiveto criticism(froma formertermlimitsupporter)thatunilaterallylimitingthe termsofMissouri'scongressional delegationwould lose the state legislativeclout in Congress.245
In contrast,a state that ousts its senior federallegislator

throughordinaryvotingcannotregainhis seniorityby bringing
LegislativeTermsat 110 (cited in note 239), quotingone advocateas statingin 1991:
stateswillhave six or eightyearsto lookat theissue.Theywillbe
"Without
retroactivity,
able to repeallimitsbeforetheygo intoeffect
ifotherstatesdon'talso adoptthem."
241
textaccompanying
note48. Pennsylvaniahad had the termlimitssince 1776,
see id, but theyhad notpresenteda collectiveactionproblemunderthe ArticlesofConfederation
becausethosemandatedrotationforall states.See note46.
2 See
ColoConst,ArtXVIII, ? 9a.
2See Alaska Stat ? 15.30.180(a)(1995) (twenty-four
otherstates);Mo Const,ArtIII,
? 45(a), cl 1 (one-halfofthe states);NH Rev Stat Ann? 653:3-4(1996),suspendedbyNH
Legis ch 108:6(1995) untiltwenty-four
otherstatesenacttermlimits;Utah Code Ann ?
20A-10-301(1995) (Termlimitwill take effectwhenat least twenty-four
stateshave establishedlegislativetermlimits.Utah Legis ch 264 ? 4 (1994)); Wash Rev Code Ann ?
29.68.015-16(Termlimitwilltake effect
whennineotherstateshave passed termlimits.
WashLaws ch 1 (1991) (InitiativeMeasureNo 573)).
' See Semerad,Salt Lake Tribat Bi (citedin note233); Editorial-Commentary,
Salt
Lake Tribat A30 (citedin note233).
8 J L & Pol at 593 n 184 (citedin note9).
See Calamita,Comment,

152
The University
ofChicagoLaw Review
[64:83
him back if the otherstates fail to oust theirseniorlegislators,

and it wouldin anyeventlose thevalue ofhis seniority
in theinterim before the next election. States tryingto coordinate
throughordinaryvotingwould also have a much shortertime
frameto coordinatethe mutual ousters.This was one of the
problemswiththe rotationprovisionPennsylvaniawas forcedto
abandonin 1790: it gave the state onlytwoyears to coordinate
otherstatesin parallelactiononce the federalConstitution
was
adopted. Because all recentlyenacted legislativeterm limits
wereno less than six years,theyprovidedat least a six-yearperiodto securesimilarlimitsin otherstates.In keepingwiththis
timeframe,all twenty-four
state-enactedlegislativetermlimits
werepassed in the 1990-95period.246
All states mighteventuallyhope to benefitfromsuch statetermlimits.But it followsfrom
by-statepassage ofcongressional
this analysisthatno state is uniquelyinterestedin the passage
ofcongressional
termlimitsin its state;the matteris one ofcollectivenational interest.247
We should thus expectto see that
supportfortermlimitswould be morelikelyto come fromnationalgroups,while supportforthe oppositionis morelikelyto
comefromlocal groupsworriedaboutlosingtheirclout.Andthat
is in factwhatwe see.248
Granted,such parallel state-enactedtermlimitsare an imperfectsolutionto the underlyingcollectiveactionproblem,requiringcumbersomecoordinationamong many states, each of
whichhas incentivesto cheat by rescindingits termlimitsto
gain a legislativeadvantageover otherstates. Nonetheless,it
was rationallyviewedas the onlyfeasiblemeans open. The alternativeofa federalstatutemandatingtermlimitsnot onlyrein
quiredthe approvalof federallegislatorswitha self-interest
opposingsuchlimitsbut also was foreclosed
bydictain Powell.249
And constitutional
amendmentproceduresbothimposedonerous
obstaclesand gave that same self-interested
supermajority
fedSee note 6.
246
247Afterall, enacting term limits is a matter of social contractingaround a collective
action problem,and all parties to a contracthave an incentiveto make it happen; indeed,

it can be expected that each party will be more attentive to definingthe duties of the
otherthan his own.
2"See Olson, Term Limits Fail, in Benjamin and Malbin, eds, Limiting Legislative
Terms at 86-87 (cited in note 5) (96 percent of fundingfor 1991 term limits campaign in
Washingtoncame fromout of state compared to 52 percent of oppositionfunding);Rothenberg, Transplanting Term Limits, in Benjamin and Malbin, eds, Limiting Legislative
Terms at 97-101 (cited in note 239) (describing national term limits groups); Hord,
Omaha WorldHerald at 9SF (cited in note 232).
249See
PartI.A.5.

1997]
153
veto on the proposedchange.250

That
eral legislaturean effective
routeofstate-enactedterm
voterswerewillingto trythe difficult
ofthe alternativesand to
limitsis a testamentto the infeasibility
just how great voterfrustrationwith legislativeentrenchment
had become.
In addition,the states that enactedtermlimitson theirfederal officialsmaywell have hopedto changethe incentivesofthe
in Confederallegislature.If states withmajorityrepresentation
gress had enactedfederaltermlimits,and the remainingstates
failed to followsuit, then a congressionalmajoritywould have
founditselfrelativelydisfavoredby a systemthat rewardedseniority.That legislativemajoritywould thus have had strongincentivesto abolishanyinternallegislativerules that accentuated
the advantages of seniority.25'
Such a rule change alone would
have lessened ideological divergence.This helps explain why
some states choseto triggertheircongressionaltermlimitsupon
adoptionbya majorityofstates.
If congressionalterm limits had been enacted by states
holdingtwo-thirdsof Congress,their federal legislatorswould
have had incentivesand sufficient
votes to propose a national
constitutional
amendmentmandatingsimilartermlimitson federal officials.This would have been true even if the federallegislatorsfromterm-limited
states were relativelyseniorbecause
theywouldhave gainednothingfromlaboringunderprospective
termlimitsmoreonerousthan legislatorsfromotherstates. For
example,a legislatorwith eighteen-years
experiencewho could
have served only twelve more years under state-imposedterm
limitswouldhave lost nothingpersonallyfromvotingfortwelveyear constitutional
termlimits.Indeed,he wouldhave improved
prospectsforhis state in a way that would have aided his next
reelection.
250
See PartI.B.4.
251In fact,beforethe SupremeCourtstruckdownstate-imposed
termlimits,House of
rule changesrelaxedseniorityadvantages.Indeed,historically,
Representatives
a large
influxoffreshmanlegislatorshas producedchangesto rules favoringseniorlegislators.
In 1974, an influxoffreshmanDemocratsresultedin the eliminationor weakeningof
manyformalseniority
advantages,thoughthe informal
ones persisted.See Linda Cohen
and MatthewSpitzer,TermLimits,80 Georgetown
L J 477, 508-09n 88 (1992). In 1994,a
similarinfluxof House Republicansgave Speaker Gingrichpowerto select committee
chairsand memberswithoutregardto seniority.
See CraigD. Margolis,House OutofOr-
der: CommitteeReform in the Modern House of Representatives, 11 J L & Pol 273, 335
(1995). Ultimately,
theseprovedto be modest,see note 135,butthatis probablybecause
withouttermlimitseven freshmanlegislatorshave an incentiveto maintaina system
that confersseniorityadvantagesbecause it increasestheirodds of reelection,see text
accompanying
notes134-35.

154
The University
ofChicagoLaw Review
[64:83
What if no such actionby the federallegislaturehad been

forthcoming
duringthe six-to twelve-year
periodofprospectivity
providedby state-enactedcongressionaltermlimits?Then the
enactingstates couldstillhave rescindedtheirfederaltermlimits beforetheysufferedany actual loss froma relativelack of
in theirrepresentation.
seniority
In short,state-enacted
federaltermlimitswerehardlya perfectmeans to a nationwidelimit.But theymay have been not
onlyan attractivealternativeto directlyshootingfora constitutionalamendmentbut the mostlikelypath to obtainingsuch an
this path was
amendmentin the long run. Unfortunately,
in
US
Term
Limits.
blockedbytheSupremeCourt'sdecision
To be sure,thereare alternativepossiblestrategies.Activists have, forexample,petitionedforinitiativesinstructinga
state's representativesin Congressto vote for a term limits
A similarstrategysucceeded
amendmentto the Constitution.252
in gettingthe SeventeenthAmendmentenacted. But such instructionsare ofcoursenotbinding.And unlikewiththe SeventeenthAmendment,
federallegislatorshave directpersonalincentives to vote against the amendmentabsent any stateimposedlimiton theirterms.The collectiveactionproblemdescribedabove would,in any event,coercevotersforany individual districtto reelecttheirincumbenteven if she failedto follow
on termlimits.Nonetheless,it maybe the best
theirinstruction
remainingstrategyafterUS TermLimits.
III.
POLITICALENTRYBARRIERS
A. Incumbency
as an EntryBarrier
Entrybarrierscan be high in politicsbecause incumbents
To some
have the considerableadvantageof name recognition.
extent,thisbrandnameadvantageresultsfromlegal favoritism.
Federal law helps incumbentsbuild brandnameby givingthem
It also
frankingprivileges,large staffs,and recordingstudios.253
it
hard
chalthat
make
for
sets spendingand contribution
limits
lengersto overcometheirdisadvantagein name recognition.254
"2See Paul Jacob, Memorandum to Term Limit Activists (Jan 18, 1996). See also
Sheffner,Roll Call at 6 (cited in note 107).
' US TermLimits, 115 S Ct at 1911 (Thomas dissenting)(citation omitted).
"4Id at 1912 (citation omitted). Because of Supreme Court decisions invalidating
mandatoryspending limits,currentlyspending limits only apply to candidates who agree
to them in exchange forfederal funds. See Fred Wertheimerand Susan Weiss Manes,
Campaign Finance Reform:A Key to Restoringthe Health of Our Democracy,94 Colum L
Rev 1126, 1131 (1994). Since public financinghas not yet been extended to Congress, this
currentlymeans the limitseffectively
apply only to the President.

1997]
155
But even withoutsuch legal advantages, the officeinevitably

givesan edge.Duringtheirtenure,incumbentswillhave had opportunitiesto speak to Rotaryclubs,presideoverschool-opening
Their tenurein office
ceremonies,and meet likelycontributors.
also allows them to build supportby performing
casework for
constituentsand to put themselvesand their positionsin the
news bygivingspeechesin Congress,issuingstatements,holding
press conferences,
or appearingon talk radio or televisionprograms.255
The resultis that incumbency"is the mostimportantdeterminantofcongressionalelectionoutcomes."256

The longerthe incumbentserves,the longershe can use such advantagesto improve her name recognitionand reputation.Entry barriers
shouldthusbe higherthe longerthe incumbent'stenure.This is
consistentwithempiricalstudies(discussedabove) that estimate
the incumbencyadvantage at 10 to 12 percentof the vote, and
show ideologicaldivergenceincreasingwith tenure.257
It is also
consistentwiththe astonishingly
highreelectionrates recounted
above.258
But perhapsmore specificevidenceof entrybarriersis the
relativepaucityof serious challengers.Of the House members
seeking reelectionin 1990, 19.5 percenthad no major party
challengerat all.259An additional 38.9 percenthad partychallengerswhocouldnot even raise morethan $25,000to challenge
the incumbent.260
And 32.5 percent had challengerswho had
raised morethan $25,000 but less than halfwhatthe incumbent
had raised.261
The resultwas that only9.1 percentofthe incumbentsrunningfacedentrybywhat we mightgenerouslycall serious challengers:thoseable to raise at least halfwhat the incumbent raised. Not surprisingly,
96 percentofthe incumbentsrunningwonreelection.262
' See generallyDavid R. Mayhew,Congress:The ElectoralConnection49-73 (Yale

1974) (Legislators'desireto maximizereelectionleads themto threemain activities:advertising,
providing
particularized
benefitslikeporkand casework,and position-taking.).
2 Paul S. Herrnson,
Congressional
Elections205 (CQ 1995).
See PartII.B.2.
See Part11.C.2.
" See Wertheimer
and Manes,94 ColumL Rev at 1134-35(citedin note254).
2S7
2Id.
261
Id.
262Id. In 1992,the statisticswereonlyslightly

less alarming.Only20 percentofchallengerscouldraise at least halfofwhatincumbents
raised,and 93 percentofrunningincumbentswonreelection.
See id.

156
The University
ofChicagoLaw Review
[64:83
B. WhyEntryBarriersMatter
In economicmarkets,we can distinguishtwo reasons why
entrybarriersare deemed potentiallyharmful.First, barriers
mightprecludetheentryofan additionalfirmthatwouldmakea
This improvesthe permarketmorecompetitive.
noncompetitive
marketby increasingoutputand
formanceof a noncompetitive
than
loweringpricesevenifthe new entrantis no moreefficient
existingfirms.Second,barriersmightprevententryby firms
in the market.This improves
moreefficient
thanthosecurrently
even if the currentmarkethas manycommarketperformance
petitorsand even if the new entrantreplaces an existingfirm
and thusproducesno net increasein the numberoffirmsin the
market.
evenifless desirablethanincumbent.
1. Morecompetition
The firstreason mightseem inapplicableto politicalmarkets. Unlike in productmarkets,only one representativecan
forpoliticaloffice
servethe districtat a time.Thus, competition
fora monopolypositionin a
mightseem muchlike competition
natural monopoly.We want to make sure we end up with the
monopolist,
but we cannotultimatelyhave more
most efficient
firm
in
one
market.
than
the
However,havingmorecandidatesrunningmighthelp present a broaderand betterdefinedset ofviews and issues in any
election.For example,Ross Perot'spresidentialcampaignin 1992
arguablyforcedClintonand Bush to address the deficitmore
than theyotherwisewouldhave. This could improvethe operationofthe politicalmarketevenifonlyone candidatecan winin
brandnameadvantagemaybe
theend. If so, thenan incumbent's
undesirable(because it detersentry)even thoughvotersprefer
in actual elections.
(and thusvotefor)thebrandnameincumbent
Thus,like the collectiveactionproblem,thisis nota problemredressablethroughordinary
voting.
a chalMoreover,even ifless desirablethan the incumbent,
lengerexcludedby entrybarriersmightbe moredesirablethan
the challengersnowrunning.Suppose 4 representsthe degreeof
fitbetweenthepublicpositionsa candidateis willingto take and
his electorate'sviews, and 0incumbent) > O(excluded challenger) > 0(present
whoprivately
favorspoliciesthatwould,if
challenger). An incumbent
he pursuedthem,producea 0 lowerthan any of these can, because ofentrybarriers,adopt a publicpositionas low as 0(present

1997]
157
challenger).263
If the entrybarrierswere eliminatedor the incumbent
barred fromreelection,the public position of the ultimately

electedofficialwouldhave a 0 at least as highas 4(excludedchallenger).
For example,even ifBob Dole wouldhave won the 1996 Republican nominationregardlessof whetheror not Colin Powell had
run, an entryby Powell likelywould have produceda shiftin
Dole's positions.
Such a shiftin positioncan producemoresatisfactory
representationforan electorateeven if,with the brandnameadvantage, the incumbentnot onlywon all electionsagainst existing
challengers,but wouldhave won againstthe excludedchallenger
as well. Thus, again, this rationalefortermlimitsis not susceptibleto the paradoxofbeingremediablebyordinaryvoting.
2. Moredesirablethanincumbent.
The second reason forcondemningentrybarriershas been

questionedin economicmarkets.264 A moreefficient
potentialentrant,the argumentgoes,shouldalways be able to contractwith
a less efficientexistingfirmto overcomethe entrybarrier.A
more efficientpotentialtaxicab owner should, for example,be
able to buy a medallionfroman existingcab owner at a price
profitableto both. Or a more efficientfirmcan buy an established brandnamefroman existingfirm.Of course,oftenlegal
restrictionson transferexist. Not all licenses can be sold. And
the purchaseofan existingcompanyor brandnameby a potential
entrantcan violateantitrustlaw ifwithoutit independententry
would have occurred.But where such restrictionsdo not exist,
the objectionis valid: because holdinga transferablemarketadvantage involvesthe opportunity
cost of foregoingits possible
sale price,the existingfirmsuffersno largercosts than a potential entrant.Thus,the entrybarrierwouldnotpreventa moreefficiententrantfromswitchingplaces witha less efficient
existing
firm.
But politicalbrandnamescannotbe sold in this fashion.265
Incumbentscan createan effective
politicalbrandnameby developing popularityforthe views and reputationassociated with
theirname throughadvertisingand position-taking.
But an in263
thecollectiveactionproblemsidentified
in Part II, he couldofcourseadoptan
even lower p. But the argumenthere followseven if no such collectiveactionproblems
exist,as in theelectionfora singleexecutiveto governa politicaljurisdiction.
See Harold Demsetz, Barriers To Entry,72 Am Econ Rev 47, 49-52 (1982).
See John R. Lott, Jr., The Effectof Nontransferable PropertyRights on the Efficiency of Political Markets: Some Evidence, 32 J Pub Econ 231, 232 (1987); John R. Lott,
Jr.,Brandnames and barriers to entryin political markets,51 Pub Choice 87, 88 (1986).
'
21

158
The University
ofChicagoLaw Review
[64:83
cumbentcannottransferhis name and reputationto others.The

closestthingto such a transferis whena spouse runsto fillthe
and it is interestingto note
seat of a deceased representative,
that electionyieldsforthemare far higherthan formostnew
candidates.266
But that is an involuntarytransfer.The closest
transferis endorsingothers,and "thereis a
thingto a voluntary
betweenbeing endorsedby Ronald Reagan for
huge difference
67
thePresidency
and runningas RonaldReagan."2
evenifa politiciancouldtransfer
his reputation,
Moreover,
a
politicalentrantcouldnot easily transferthe fundsto buy it. A
successfulentrantin business marketscan anticipatemaking
the capitalizedvalue ofwhichhe
moneyfromsales to consumers,
firmforits brandname.In contrast,the
can payto an incumbent
"proceeds"ofa successfulpoliticalentrantare mainlynotmoney
but rather votes and political support, which cannot be
monetizedand transferred
to the incumbent.Even if a political
entrantdoes raise supportin the formof monetarycampaign
he cannotlegallyspend thatto pay an incumbent
contributions,
to resignhis offilce.
Such paymentsare impermissible
fora very
orpoliticalreputations
goodreason:we wouldhardlywantoffices
auctionedoffto the highestmonetarybidder.That would only
exacerbaterent-seeking
behaviorand the advantageof interest
groupsbetterpositionedto raise funds.Whatwe wantis forthe
officeto go to the candidatewho can offerthe highestpolitical
support.But politicalsupportdoes notcomein a formthatcan be
forhis brandnameadvantage.
exchangedto an incumbent
Because ofthese obstaclesto transferability,
the incumbent
will runforreelectionratherthan sell his brandnameto a lessknownchallengerwho woulddo a betterjob. Nontransferability
can accordingly
make brandnameadvantage an entrybarrier
thatpreventsa moredesirablechallengerfromwinningan election.
Still,one mightjustifiablyquestionwhethersuch a political
brandnameadvantageis undesirableon balance.268
Votersand
The secondclosestanalog is whenchildrenof politiciansrun foroffice.Interestingly,politicians'
childrenenterpoliticsat higherthan averagefrequency,
but the vast
majority
do notrunat all. Lott,32 J Pub Econ at 232 (citedin note265). This is consistentwiththe proposition
thatpoliticalbrandnameis difficult
to transfer
to anyone,includingchildren,
butis evenmoredifficult
to transfer
to personsoutsidethefamily.
See Lottand Davis,74 Pub Choiceat 475 (citedin note154).
See, forexample,Tabarrok,14 CatoJ at 343-44(citedin note12).Lott'sbrandname
modeldoes notanswerthisquestionbecause (1) it simplyassumesthe brandnameincreasesthechancesofreelectionwithoutbenefiting
the voters;and (2) it onlyaddresses
thepossibility
ofa moreefficient
challengerbeingexcluded.Lott,51 Pub Choiceat 88-89
(citedin note265).

1997]
159
consumerswho choose an established brandname over lessknowncompetitors

mustdo so because theyfeelthat on average
it makes them betteroff.The less-knowncompetitormightbe
much betterat deliveringdesirable governmentalservices,but
he mightbe muchworse.Giventhe expectedoutcomeand relative risks of choosingone or the other,they preferthe brandname. Thus, people stop at McDonald's because it offersthe assuranceofa certainqualityand priceeventhoughtheymightget
mixat an unknownrestauranttheypassed
a betterquality/price
on the way. If McDonald'swereunable to sell its brandnameto a
moreefficient
it wouldhardlyseem desirableto adopt
franchiser,
the equivalentoftermlimits,prohibiting
McDonald's fromusing
its brandnameforlongerthan twelveyears. The squelchingof
the brandnamewould deprivethe marketof informationthat
makes consumersbetteroffon balance.269
If theyweren'tbetter
off,theywouldchoose(votefor)theunknowncompetitor
overthe
brandname.270
The problem,however,is that voters may never have a
choicebetweenthe moredesirablechallengerand the brandname
incumbent.Such politicalchallengersmay nevereven enterthe
race because theyconcludethat the costs of creatingtheirown
brandnameare too high givenotheruses of theirtime and resources. In economicmarkets,entrantscan more easily recoup
such costswithincreasedsales and shouldthus make the necesto do so. But in posary brandnameinvestmentif it is efficient
litical markets,brandnameinvestmentsoftenconferuncompensated positiveexternalitiesor are compensatedwith votes and
politicalsupportthat do not cover the necessarymonetaryinvestment.Moreover,politicalchallengersknow that the incumbent can respondto any investmenttheymake in a rival brandname withincreasedinvestmentsin the incumbent'sown brandname. Accordingly,
entrybarriersmay mean that actual elections presentvoterswith a choice betweenbrandnameincumbents and less desirableunknownchallengers.Voters'failureto
voteforunknownchallengersoverbrandnameincumbentsin actual electionsdoes not mean theycould not get more desirable
if the incumbent'sbrandcandidates,and betterrepresentation,
name advantagewerereducedor eliminated.
...CompareDemsetz,72 Am Econ Rev at 50 (citedin note264) (makes no sense to
talk ofentrybarrierswhenthe barrieris reallycaused by information
costs that are a
costofproduction
thateach firmfacesequally).
270
See Daniel Friedman and Donald Wittman,Term Limits as Political Redistribution,
in Grofman,
ed, LegislativeTermLimits229, 229-30(citedin note 5) (relyingon such
logicto dismissbrandnameentrybarrierargumentfortermlimits).

160
The University
ofChicagoLaw Review
[64:83
C. TermLimitsto LowerEntryBarriers
Termlimitsare one solutionto this entrybarrierproblem.
fromrunningforoffice,
incumbents
By preventing
long-term
they
lessen the possible brandnameadvantage and thus lower the
heightof politicalentrybarriers.Some brandnameadvantage
maypersist.Term-limited
legislatorscan use theirbrandnameto
seek otherpoliticalofficesor, under rotationversionsof term
limits,can even seek the same officeafterwaitingsome six to
twelveyears. But politicalfame fades quicklyover time and
across politicalboundaries.How manyofus knowthe names of
whoservedtwelveyearsago orin otherdistricts?
representatives
Moreover,a reputationwill oftenbe forservinga particulargenerationor districtand thusnotworkin othertimesand places.27'
Termlimitswillthuslowerentrybarriers.
To theextentfamedoes carryoverin different
regions,such
famewill notbe unique and no incumbentwill enjoya safe seat.
Indeed,one of the great advantagesof termlimitsis that they
betweenfamousterm-limited
producereal competition
offlcials
forotherpoliticalposts. Currently,normalrisk aversiongives
prominentpoliticiansincentivesto stay in theirsecure offices
and bide theirtimeuntilanotherseat becomesvacantby resigwhich is what the evidenceshows they
nation or retirement,
fromsuch safe sinecures,termlimitsendo.272
them
By ousting
couragewell knownpoliticiansto competewitheach othermore
often.273
271 In thenation
(CostaRica) withthelongesthistory
oflegislative
termlimits,only13
percentof term-limited
legislatorshave come back to serve anotherterm.See Carey,
Term Limits in Costa Rica, in Grofman,ed, Legislative Term Limits at 323 (cited in note
132).
272See Peverill Squire, Challengers in U.S. Senate Elections, 14 Legis Stud Q 531, 532
(1989)(Senatechallenges);GaryC. Jacobsonand SamuelKernell,Strategy

and Choicein
Elections32 (Yale 1981) (House challenges);GaryW. Copeland,Term
Congressional
Limitations and Political Careers in Oklahoma: In, Out, Up, or Down, in Benjamin and
Malbin,eds,LimitingLegislativeTerms139, 147-48(citedin note5). This rebutsan objectionsometimes

madeto termlimits:thattheywillcause challengers
to deferrunning
untilthe incumbent's
termlimitcomesup. See Fowler,Commenton Competition,
in
Benjaminand Malbin,eds,LegislativeTermLimitsat 183-84(citedin note12); Hibbing,
Careerism in Congress, in Dodd and Oppenheimer, eds, Congress Reconsidered at 69
(citedin note 151); BernardGrofmanand Neil Sutherland,The Effectof TermLimits

When Competitionis Endogenized: A Preliminary Model, in Grofman,ed, Legislative
TermLimits175,175 (citedin note5). Althoughthismaysometimes

happen,anydampeningof entrywouldsurelybe less than that whichoccurswithouttermlimitswhen
challengers
waitforincumbents
to comeofretirement
age. Andthepoliticians
oustedin
one officeby termlimitscannotdeferchallengesto otherincumbents
unless theyare
willingto be outofoffice
(and awayfromits fame-producing
advantages)fora significant
period.
The Great Money Chase (CommonCause April1995)(oftwentyS73ee K. Alexander,
sevenCalifornia
legislators
decidingnotto runforreelection
in 1994giventheprospectof

1997]
161
By reducingthe costs of entryfornewcomers,lower entry

barriers should also produce additional entrymore generally.
True, in some sense, higherentrybarriersalso encourageentry
byincreasingthebenefitsofsuccessfulentry.For example,being
a monopolistwhoenjoysthe advantageofentrybarriersis so attractivethatwe mightexpectmorefirmsto tryforit.274
Likewise,
the prospectof a thirty-year
legislativecareer may be more atcareer and thus intractivethan a twelve-year-term-limited
crease the benefitsof politicalentry.Such logic has persuaded
othersthat termlimitsdiscourageentry.275
But the net effectof
reducingentrybarriersshouldstillbe additionalentryforno less
than six reasons.
First,some realism.With only9.1 percentof congressional
incumbentsfacingseriouschallengersin 1990,276currentlevels of
low. Such low levels make it
politicalentryare extraordinarily
implausiblethat incentiveswithouttermlimitsare doingmuch
to encourageentry.Andtheymake it hard to believethingscould
getanyworsewithtermlimits.
Second, I doubt many people runningfor political office
would be deterredby a law limitingthemto twelveyears in office. The average tenure of sittingcongresspersonswas 11.14
years in 1994.277Whilethis underestimatesmattersby including
memberswho have not yet completedtheir service,effortsto
adjust forthis using 1977-91continuationrates producea predictedmean completedtenureof 13.5 years in the House.278
This
does not offermuchmoreinducementthan the maximumcareer
available under a twelve-yeartermlimit,thoughof course the
average tenureundertermlimitswould be less.279
The disincenhavingtheirtermlimitedin 1996,twenty-two
ran forotherelectiveoffices,
oftenagainst
otherincumbents);BernardGrofman,
Introduction,
in Grofman,ed, LegislativeTerm
Limitsat 16 n 21 (citedin note5) (oftwenty-two
CaliforniaAssemblymembersretiring
underthespecteroftermlimits,eighteenran forotheroffices).
" Such rent-seeking
behavioris in factgenerallyunderstoodto be inefficient.
The
cost ofcompeting
to becomea monopolist
increasesthe deadweightloss ofmonopolyby
eatingup even the monopolist's
surplus.See RichardA. Posner,The Social CostsofMonopolyand Regulation,in JamesM. Buchanan,RobertD. Tollison,and GordonTullock,
eds,Towarda TheoryoftheRent-Seeking
Society71 (Texas A&M 1980).
275See Fowler,Comment
on Competition,
in Benjaminand Malbin,eds,LimitingLegislativeTermsat 182,184(citedin note12).
276See textaccompanying
notes259-62.
177See Elizabeth Garrett,TermLimitationsand the Mythof the Citizen-Legislator
(preliminary
draft1996) 17 & n 33 (on filewithU Chi L Rev).
278 See W. Robert
An AnalysisoftheImpactofCongresReed and D. Eric Schansberg,
sional TermLimits,32 Econ Inq 79, 82 (1994).
" One estimate,assumingthatreelectionrates are unalteredby termlimits,is that
averagecompletedtenurewoulddeclineto 6.3 years. Id. That shouldbe even lowerif
reelectionrates go down.But thatbringsus to the third,moreimportant,
pointin the

162
The University
ofChicagoLaw Review
[64:83
tiveis even less forstate legislatorsbecause even withoutterm

limitsonly24 percentservemorethantwelveyears.280
a lowermean completedtenure
Third,and moreimportant,
overstatesthe deterrentto entrybecause it measures the expectedtenureofan incumbent.But the expectedyears in office
frombeinga challengerare what mattersin evaluatingthe attractionofentry.281
And since over90 percentofchallengerscurless than twoyears.
rentlylose,thatexpectedvalue is currently
Thus, even a six-yeartermlimitmay not mattermuchto newofwinning,and
comers.Whatmattersfarmoreis theprobability
thatshouldgo up withtermlimitseitherbecausemoreseats will
be open or because the less seniorincumbentswill have lower
brandnameadvantages.Thus, even if the benefitsof successful
entrygo downundertermlimits,the expectedbenefitsof entry
shouldstillgo up.
Fourth,the prospectof being turnedout may not matter
muchto candidates.It is not as if electedofficepays morethan
the sortof privatejobs term-limited
politicianscould get after
Ofcourse,politiciansmayrunto use an offlceas
servingin office.
a steppingstone to higherpoliticaloffices.But if so, theyare
likelyto have to maketheirmovesbytheend oftheirtermlimits
Andtermlimitsin thosehigherpoliticalofflcesshould
anyway.282
increasetheirchancesto moveup.
Fifth,a longerlegislativecareerwill notbe moreattractive
to everyone.Many may onlywant to enterpublicservicefora
limitedtimebeforereturningto privatelifebut want to make
sure they can accomplishsomethingwhile they are in offlce.
Such personswouldbe moreattractedto an officethatlasts only
six to twelveyearsbut assures themofreal powerby the end of
thatperiodthanto an offlcethat mightlast thirtyyearsbut requiresthemto waitdecadesbeforerisingto real power.A twelvetext.
Combiningupper and lower state houses, only 30 percent serve as long as twelve
years, see Moncrief,et al, For Whomthe Bell Tolls, in Grofman,ed, Legislative Term Limits at 121-23 (cited in note 209), and one-fifthof these leave withoutgoing on to a thirteenthyear, id at 128 n 5. See also Copeland, Political Careers in Oklahoma, in Benjamin
and Malbin, eds, LimitingLegislative Terms at 145 (cited in note 272) (findingOklahoma
legislators unworriedabout term limits because few expect to serve more than twelve
years).
2 (Expected years in officefromchallenging) = (probabilityof winning challenge) *
(expectedyears in offlceifwin and become incumbent).
282
Copeland, Political Careers in Oklahoma, in Benjamin and Malbin, eds, Limiting Legislative Terms at 145-46 (cited in note 272) (notingthat twelve-yearlimit unlikely
to affectOklahoma politicians with progressive ambition since there are "no recent examples of a state legislator with at least twelve years of service seeking election to the
U.S. Congress or to another statewide offlce").

1997]
163
year termlimit,forexample,wouldreducethe timenecessaryto

rank in the 80th percentileof seniority(where positionslike
committeechairs normallygo) fromsixteento twenty-two
years,
this
result
in
a
net
increase
to six to eightyears.283
Indeed,
may
in the value of holdingofficeeven forthose solelyinterestedin
wieldingthemaximumpoliticalpower.284
Morefundamentally,
eveniftermlimitsdid reducethebenefitsof entry,this would at mostbe a feedbackeffectthat could
notoffsetthe main effect.The higherbenefitsofentryunderthe
currentsystemhold onlyso long as entrybarriercosts actually
preventothersfromentering,whichmust mean entrycosts exceed entrybenefits.A legislativeseat can offera thirty-year
careercarrotonlyifin factwinningthat seat precludesmeaningful
forthe nextthirtyyears.285
competition
Such entrybarriersthus
cannotencourageentrymorethantheydiscourageit.
Finally,termlimitsshould encourageentryby makingthe
date incumbentswill depart officemore definite.Challengers
have some incentivesto run even beforetheybelieve theycan
win an electionbecause such priorchallengesincreasetheirown
reputationand brandnamein futureelections.But withoutterm
limits this is a risky strategy.The challengercannot be sure
when the incumbentwill depart,and if the incumbentstays a
long time any past reputationalinvestmentby a challengerwill
have depreciatedbythetimethe seat becomesopen.286
Withterm
limits,the challengerhas more certaintythat his investments
willpay offin a futureelectiononcethe incumbentdeparts.
But we need not relysolelyon theoryabout whetherlower
entrycostswithtermlimitswillbe offsetbylowerentrybenefits,
for some empiricalevidence is available. And it supportsthe
claimthattermlimitson balance encourageentry.AfterCalifornia imposed term limits in 1990, thus restrictingthe possible
prospectivecareer,candidatefilingsincreasedby 25 percentfor
the state senate and 50 percentforthe state assembly.287
The in' See Reed and Schansberg,32 Econ Inquiryat 84 (citedin note278). In addition,the
ofattaininga leadershippositionwouldmorethandoubleundertermlimits,
probability
althoughthe lengthoftimein such positionswouldbe less than halfthe timewithout
termlimits.Id at 83 n 8.
2Id at 90.
" Notealso thatin thoseonce-every-thirty-years
elections,candidateshave an incentiveto spendfarmorebecause theygeta thirty-year
return,whichincreasesentrybarriers forcandidateswithless money.
See JohnR. Lott,Jr.and KermitDaniel, TermLimitsand ElectoralCompetitiveness6 (forthcoming
Pub Choice,Jan 1997).
27 See Dan Greenberg,
TermLimits:The OnlyWayto Clean Up Congress,Heritage
FoundationBackgrounder1, 8 (Aug 10, 1994). See Lott and Daniel, TermLimitsat 15
(citedin note 286) (findingthat major partycandidaciesincreasedby 24 percentof a

164
TheUniversity
ofChicagoLaw Review
[64:83
crease shouldbe evengreateronce the termlimitsstartturning

outincumbents
thisyear.
Pursuant to the reasoningabove, additionalentryshould
improvethe politicalmarketin threeseparateways:by helping
refineissues, by producingmoresatisfactory
position-taking
by
representatives,
or byallowingentryby representatives
whocan
do a betterjob. Consistentwiththesepredictions,
aftertheadoptionoftermlimitsin California,
votemarginsdecreasedby 10-22
percentofa standarddeviationand the probability
ofan incumbent beingdefeatedincreasedby 14-37 percent.288
The new entrantsmustexpendtimeand resourcesto developa new brandname conveying
theirviews.But theywill no longerbe discouraged fromdoingso by the large brandnameadvantagesenjoyed
incumbents.
bylong-term
This rationalefor term limits is strongerthe larger the
brandnameadvantages associated with an office.Given the
enormousmedia attentionfocusedon the President,it is hardly
surprisingthat termlimitswere adoptedforthat federaloffice
first.The rationaleis also strongerforofficesthatusuallymark
the end ofa politicalcareer,such as President,sincesuch termlimitedofficeholders
are less likelyto take theirbrandnameadBecause brandnameadvanvantageto electionsforotheroffices.
tages rise withtenure,termlimitsshouldalso be morepopular
the longerthe tenureofincumbents.It is thus predictablethat
presidentialtermlimitsbecamean issue at theend ofRoosevelt's
longtenure,and thatcongressional
termlimitsbecamean issue
in the early 1990s when legislativeturnoverhad becomeparticularlylow.
Of course,it is also truethatthe largerthe brandnameadtenureis likelyto be. Thus,other
vantage,thelongerincumbent
factors(such as technologicalchanges) may have increasedincumbents'advantage,whichin turn producedthe long tenure.
Whateverthe source,the enthusiasmfortermlimitswouldrise
withincumbents'
advantage,and thereis an "existingscholarly
consensusthat incumbencyadvantage is much larger than it
oncewas."289
Some findno detectableincumbency
advantageuntil after1950; othersfindthat an advantage of 1-2 percentin
1900has recently
become10-12percent.290
standard deviation and the numberof elections with only one candidate decreased by 89
percentofa standard deviation).
'See Lott and Daniel, TermLimits at 14-15 (cited in note 286).
"9Gelmanand King, 34 Am J Pol Sci at 1142, 1157-58 (cited in note 148).
2Id at 1158.

1997]
165
To some extent,the effectof tenureon brandnamebarriers

offsetsan effectdiscussedin Part II.C.1. Recall that thereI concluded that districtswith below-averageseniorityshould be for
termlimitswhereas districtswith above-averagesenioritymay
or may not be fortermlimits.Standingalone, this would have
led one to predictthe seniordistrictswould be less enthusiastic
about termlimits.But highseniorityalso means the districtsuffers fromhigh brandnameadvantages and entrybarriersthat
will make it moreenthusiasticabout termlimits.It is thus unclear whetheron balance we should expect any correlationbetweenseniorityand supportoftermlimits.291
Finally,some mightfindthe entrybarrierargumentso persuasive that it obviatesany need to relyon the collectiveaction
problemsdescribedin Part II. But the evidenceis to the contrary.
First,legislativereelectionrates are muchhigherthan executive
reelectionrates. This is hard to explainunless thereis some additionalfactorcontributing
to incumbencyadvantagesin legislative races. Second,the trendtowarddecreasinglegislativecompetitivenessin the last fewdecades is hard to explainwithentry
barriersbecause the evidenceshows that incumbentlegislators
are no betterknownnow than theywere then.292
This is not necessarilyconclusive,because it may take a lot moreinvestmentto
securethe same level ofbrandnamerecognition
in districtsthat
have grownmorepopulatedand distractedovertime.But it does
tendto suggestthat the collectiveactionproblemsunique to legislaturesoffera betterexplanationofthistrendand ofthe timing
ofthe 1990-95explosionin proposalsforlegislativetermlimits.293
In contrast,executivetermlimits,whichrest solelyon the entry
barrierproblem,have been moreor less equally popularforthe
last fifty
years.
D.
Alternative
EntryBarrierSolutions
1. Repealinglegal incumbency
advantages.
It mightseemthata moretailoredresponseto politicalentry

barrierswould be to repeal the laws givingincumbentsspecial
advantages such as frankingprivilegesand large staffs.294
But
291See textaccompanying
note202 (notingthatdifference
betweenseniorand junior
districtsupportfortermlimitswas only1 to 3 percentin California).
2 See
Kesler,Bad Housekeeping, in Benjaminand Malbin,eds,Legislative Term Limits at 246 (citedin note196)(discussingstudiesconductedbyJohnFerejohn).
"The termlimitspassed duringthisperiodwerethe firstlegislativetermlimitsenactedin theUnitedStatessincethe 1700s.See BernardGrofman,
TermLimitsDebate,in
Grofman,
ed,LegislativeTermLimitsat 1 (citedin note5).
' Some state termlimitsinitiativeshave in factincludedreductionsin legislative

166
The University
ofChicagoLaw Review
[64:83
to communithereare goodreasonsto allow our representatives

and to have staffs.Nor would such
cate withtheirconstituents
measureseliminatethe brandnameadvantagesof incumbents,
mostofwhichare inherentin holdingofflce.295
In any event,the statesenactingtermlimitson federalofficials mayhave doneso in partbecause theylack the powerto repeal federallaws givingbrandnameadvantagesto incumbents.
rule change,federalstatDoingso wouldrequirea congressional
amendment,
any ofwhichrequiresthe apute,or constitutional
provaloffederallegislatorsill-disposedto lessen theirabilityto
win reelection.If federallegislatorsfromenough states were
subjected to term limits, then their incentivesmight have
but the Supreme
forsuchrepealsto occur,296
changedsufflciently
thatoption.
Courthas nowforeclosed
2. Ballotaccess restrictions.
Anothermore tailoredresponse mightbe not to outright
prohibitincumbentsfromrunningfor reelection,but to deny
themthe rightto have theirnames on the ballot.Arguably,this
more directlyoffsetsincumbents'brandnameadvantage,while
still allowingthe votersto retain,witha write-incampaign,an
prefer.Indeed,theArkansaslaw
whomtheystrongly
incumbent
struckdownin US TermLimitswas just such a law, as werethe
Such statutesseem
termlimitsenactedin manyotherstates.297
motivatedbythebrandnameadvantageproblem.
directly
to runa write-incampaigneffectively
Forcingtheincumbent
advantageto challengers(theirname is
givesa namerecognition
on the ballot)thatmightbe thoughtto roughlyoffsetthat ofincumbents.This shouldencourageentryby morechallengers.But
ifthedisadvantageofnotbeingon theballotis muchlargerthan
incumbents'normal brandnameadvantage, then such a rule
would effectively
prohibitincumbentswho exceeded the term
limitfromrunningagain,thusproducingthe same effectsas absolutetermlimits.The US TermLimitsmajoritywas convinced
thiswas true,notingthatin 21,300congressionalelectionsonly
But theevidencewas unsix write-in
candidateshad everwon.298
had ever run as
clear because onlytwo incumbentcongressmen
write-incandidates,one ofwhomgot27 percentofthe voteafter
staff.See, forexample, Polsby,Am Prospect at 40 (cited in note 5).
2 See Part III.A.
'See also Part II.D.
" See US TermLimits, 115 S Ct at 1909 n 39 (Thomas dissenting).
Id at 1868 n 43 (majorityopinion).

1997]
167
losinghis party'sprimary,the otherofwhomwon in an 83 percentlandslide!299

However,even if the disadvantageof denyingballot access
does roughlyequal the brandname advantage of the average
long-term
the effectscan be perverse.A highlysenior
incumbent,
incumbentwill have a largerthan averagebrandname
long-term
advantageand stillbe able to prevailwhile the less seniorlongtermincumbentsfall.But it is the highlyseniorincumbentswho
have the greatestpork-providing
abilityand greatestideological
slack and fromwhomvotersin otherdistrictsmost want to be
rid.300
Nor do such ballot restrictionsdo anythingto reducethe
substantivecollectiveaction pressure (describedin Part II) to
voteforlong-term
incumbentswitha greatershare oflegislative
clout.And ballotaccess restrictions
mightalso perverselyinduce
incumbentsto devoteeven moreeffortto buildinga brandname
capable ofovercoming
the obstacle.Still,even ifthehighlysenior
incumbentsretaina net brandnameadvantagedespitethe ballot
access restriction,
theiradvantage should be lowerthan before,
and thus politicalentrywill be higher,withall the beneficialeffectsidentifiedabove.
Nonetheless,such restrictionsseem to run contraryto the
propositionin Buckleyv Valeo that the governmentcannotrestrictthe speech of some to "enhance the relativevoice of othFor the pointof such a ballot restrictionis not to further
ers."'301
administrative
reduceconfusion,or otherwisemake it
efficiency,
easier forvotersto become informedabout the candidates by
limitingthe ballot to candidates with serious voter support.302
Rather,the pointis to denysomevotersthe information
thatthe
incumbentis runningby restricting
his abilityto say so on the
ballot.Andtherestriction
is justifiedonjust the groundsdeemed
impermissiblein Buckley,that otherwisethe relative voice of
otherswould be too weak. Term limitsthus seem preferableto
ballotaccess restrictions.
3. Campaignfinancereform.
Oftenit is suggestedthat the real solutionis campaignfinance reform.303
Buckleyv Valeo could be overturnedto impose
Id at 1910(Thomasdissenting).
See Part II.
""424 US 1,48-49(1976).
302 See Tribe, AmericanConstitutional
Law ? 13-18 at 1097 (cited in note 37)
(describing
legitimatepurposesofballotregulation).
"'See, forexample,Carole Jean Uhlaner,Alternatives
to TermLimits,in Grofman,
30(
ed, Legislative Term Limits 347, 347-49 (cited in note 5).

168
The University
ofChicagoLaw Review
[64:83
strictequal spendinglimitson candidates,or evaded by giving

discountsto candidateswho agreedto conform
largeadvertising
to spendinglimits.Public fundingcouldbe extendedto legislative races and used to equalize spendingbetweenincumbents
and challengers.Or both candidates could be given equal
amountsof freetelevisiontime to help equalize theiraccess to
voters.
seemattractive.Incumbentsget
At firstglance,suchreforms
For example,in 1990 Political
the lion's share of contributions.
to
ActionCommitteesdirected92 percentof theircontributions
in 1990 only9.1 perAnd,as alreadymentioned,
incumbents.304
centof House incumbentsfaceda challengerwho had managed
to raise as much as half of what the incumbenthad raised.305
Moreover,campaignfinancereformmaybe salutaryforreasons
unrelatedto incumbencyadvantages,includingfreeingcandidates fromraisingmoneyand reducingall candidates'dependenceon campaigncontributors.
Nonetheless,to suggestsuchreformas a substituteforterm
limitsmisses the pointabout entrybarriers.Because of incumbents' brandnameadvantages,a winningchallengertypically
spends.306
mustspendfarmorethantheincumbent
thatassuresequal spendingor
Anycampaignfinancereform
equal advertisingtimein the electionhelpspreservethe incumbencyadvantage,not eliminateit. For incumbentsbegin each
theirpositionhad
electionbetterknownbecause of everything
to do and say in thepast. This is why
giventhemtheopportunity
tend to back them:everyonelikes to investin the
contributors
would only
likelywinner.But limitingcampaigncontributions
thebrandnameadfromfurther
magnifying
preventincumbents
The brandnameadvantage actuallycannotbe overvantage.307
8 J L & Pol at 559 (citedin note9). In 1992,PACs con"See Calamita,Comment,
and Manes,94
See Wertheimer
thanchallengers.
tentimesmoreto incumbents
tributed
ColumL Revat 1134(citedin note254).
runningfaced
305 See textaccompanying
notes259-62.A full58 percentofincumbents
oronewhoraisedless than$25,000.Id.
eitherno challenger
"See Stephen E. Gottlieb, The Dilemma of Election Campaign Finance Reform,18
has been estimatedat approxiHofstraL Rev 213, 222 (1989) (IIV]alue ofincumbency
See GaryC. Jacobson,
district.").
matelyone milliondollarsfora federalcongressional
Money in Congressional Elections 33-162 (Yale 1980); Gary Jacobson, Enough Is Too
Much: Money and Competitionin House Elections, in Kay Schlozman, ed, Elections in
America173,192-93(Allen& Unwin1987).The 1994electionswerean exceptionto this

withthirty-four
RepublicanchallengerswhodefeatedDemocraticingeneraltendency,
what incumbentsspent. See BradleyA.
cumbentsspending,on average,two-thirds
Smith, Faulty Assumptions and Undemocratic Consequences of Campaign Finance Re-
form,105Yale L J 1049,1065(1996).
tenurewas flatuntilthemidtimeseriesevidenceshowsthatcongressional
37In fact,
1970s,whenit beganto increasesuddenly,see W. RobertReed and D. Eric Schansberg,

1997]
169
come unless the challengercan advertisefar more than the incumbentin the campaign.308
Still, to those who decrythe high amountsof currentcampaign expenditures,and the dependenceon contributors
theyimply,it is worthnotingthat campaignexpendituresgo down significantly
under termlimits.To a surprisingdegree.In California, average campaign expenditures fell from $309,000 to
$215,000.309This might seem surprisingbecause the six-year
limitwas imposedin 1990 prospectively
and thus ousted no one
until 1996. But under such termlimits,candidates startingin
1992 wereno longercompetingfora potentiallylifetimeposition.
The shorterexpectedtenuremeans theyhad incentivesto spend
less in eitherdefendinga seat or challengingan incumbent.As
termlimitskick in fully,spendingshould decrease furtherbecause with less senior incumbentsthe brandname advantage
neededto be overcomewithcampaignadvertisingwill decline.
4. Conclusion.
None of the alternativesseems a betterdevice forreducing
political entrybarriersthan term limits. In addition,none of
themwoulddo anythingabout the collectiveactionproblemsdescribedin Part II. Those problemscreate a substantivecoercion
The behaviorofcongressional
tenureovertime:1953-1991,73 Pub Choice183, 184 (1992),
whichroughlycoincideswiththe implementation
ofthe firstseriousroundofcampaign
financelimitations.See Federal ElectionCampaignActAmendments
of 1974,Pub L No
93-443,88 Stat 1263 (1974), codifiedat scatteredsectionsof Titles 2, 18 and 26 of the
in part,byBuckleyv Valeo,424 US 1
UnitedStates Code and declaredunconstitutional,
(1976).
308To reallycreate a level playingfield,reform
would have to allow challengersto
spendfarmorethan incumbents.One intriguing
possibilityis to make spendinglimits
lowerthe highera candidate'stenureor total expendituresin past races. See JohnR.
The ImportanceofSunk NonLott,Jr.,ExplainingChallengers'CampaignExpenditures:
transferable
Brandname,17 Pub Fin Q 108, 115 (1989). Not surprisingly,
this formof
campaignfinancereform
has gonenowherein legislaturesand is notbeingpushedbyreformactivists.In additionto this practicalproblem,the proposalfaces a technicaland
conceptualproblem.The technicalone is how muchweightto give to tenureand past
campaignexpenditures.The relatedconceptualproblemis that we cannotdecidehow
muchweightto give unless we knowwhat goal we wish to achieve.The goal generally
is to equalize the chancesthat a challengerand incumbent
posited(at least implicitly)
have ofwinning.But thatshouldbe our goal onlyifwe knowbothare equallydeserving
ofwinning,whichwe have no reasonto believe.Indeed,standingalone,the factthatthe
incumbentalreadywon one electionsuggestsshe is moredeservingofoffice.But recognizingthatforcesus intothe conceptually
unanswerablequestionofjust howmuchofincumbents'electoraladvantageis merited.
"See Lottand Daniel, TermLimitsat 11 (citedin note286) (in constant1982 dollars,
comparingthreeelectionsbeforetermlimitsenactedto threeelectionsaftertermlimits
enacted).See also MarkAnderson,Changeson theWay:Preliminary
Effects
ofTermLimits in Arizona,4:4 Term LimitsOutlookSeries 9-10 (Nov 1995) (After1992 passage of
termlimits,averagespendingdeclinedfrom$17,900in 1990to $14,500in 1994.).

The University
ofChicagoLaw Review
170
[64:83
to vote forincumbentsthat is not redressedby any effective

changein politicaladvertising.Onlytermlimitssimultaneously
tackleboththecollectiveactionand theentrybarrierproblems.
IV.
STRUCTURALOBJECTIONS TO TERM LIMITS
Most of the objectionsto termlimitshave been arguments

forexpertiseand experience.These have been pittedagainstthe
oppositeintuitionthat freshnewcomerswould be more princiI have littleto add to this debateotherthan to notethe
pled.310
inherently
unresolvablenatureof such disagreementsand that
itselfwithvoters'sineitherside does a goodjob of reconciling
multaneousapprovalof termlimitsand votes to retainincumThe voterswhoapprovedtermlimitsmusthave rejected
bents.31'
theviewofopponentsthattheloss ofexpertiseand experienceis
intolerable.The voterswho voted to retain incumbentsmust
have been unconvinced
bythe claim,standingalone,thatprinciplednewcomers
are betterdecisionmakers.
But many argumentsagainst term limits are structural.
Theyarguenotthatvotersare makingunwisechoices,but that
the processofdecisionmaking
set up by termlimitsis flawed.I
addresssuchargumentsnext.
A. They'reStillUndemocratic
One responseto Parts II and III mightbe that,even ifthey
show whytermlimitsmightbe desirable,termlimitsare still
undemocratic.
It is importantto pin downpreciselywhat might
be undemocratic
aboutthem.Termlimitsare, afterall, adopted
by democratic
majorities,and almostalwaysby directvoterballot.Thus,theyreflectat least the viewsofthe currentelectorate
and are not imposedundemocratically
on that electorateby its
governmental
agents.Nordo theyraise thetraditionalconcernof
restrictions
thatmightinterfere
withvoters'capacityto displace
incumbentsin the future.If anything,the problemwith term
limitsis thattheydisplacethoseincumbents
toooften.
Rather,thecritiqueboilsdownto threeconcerns.First,term
limitsmightallowtheviewsofthecurrentelectorateto interfere
witha futureelectorate'spreference
to retainits incumbents.312
Second,termlimitsmightallow a state or nationalmajorityto
imposeits viewson contrary
districtsthatwouldpreferto retain
30 See textaccompanyingnotes 10-11.
"' See textaccompanyingnotes 12-13.
311See, forexample, Nelson W. Polsby, Some ArgumentsAgainst Congressional Term

Limitations, 16 Harv J L & Pub Pol 101, 104 (1993).

1997]
171
their incumbents.Third, accuratelyreflectingan electorate's

preferences
maynot advance Burkeanconceptionsofrepresentativedemocracy.
1. The interference
withfutureretention.
Termlimitsundeniablymightinterfere
withthe preferences
of a futureelectorateto retain its incumbents.If the theoryof
term limits were that the preferencesof currentelectorates
shouldprevailover the preferencesoffutureelectoratesregardingtheirfuturerepresentation,
thentermlimitswouldclearlybe
undemocratic.3"3
But thisis nottheirtheoryrightlyunderstood.
What the analysis in Parts II and III shows is that we cannot choose a systemthat leaves the preferencesof the future
electorateunconstrained.If termlimitsare not adopted,the future electorate'sdemocraticchoiceswill be coercedby collective
actionproblemsand constrainedby entrybarriers.The collective
actionproblemscan cause votersto keep incumbentswho have
divergentideologicalviews and provide excessive pork. Term
limitscan thus help the futureelectoratechooserepresentatives
who betterrepresenttheirdemocraticpreferenceson ideological
and pork matters.And entrybarriers can exclude preferable
candidatesfromthe ballot and allow incumbentsto avoid issue
definition
and shiftsin positionthatbringthemcloserto electorate preferences.
Loweringthose entrybarrierscan thus produce
representativeswho betterfitthe democraticpreferencesof the
futureelectorate.
Granted,this lessened coercionand constraintcomes at a
cost. Term limitsmay sometimesprecludethe candidacyof an
incumbentwho a completelyunconstrainedand uncoercedelectoratewould preferto keep. But if we have to choosebetweena
capacityto displaceand a capacityto retainincumbents,it seems
plain thatpreservingthe latteris less important.Reducingvoter
3"3SeeParts I.A.3,I.B.1, and I.B.2. To be sure,individualsuse theirpresentautonomy
to constraintheirfutureautonomyall thetime.We can,forexample,imaginean individual simultaneously
eatinga richdessertand signingup to join a weight-watching
program.But thereare limitsto our willingnessto allowindividualsto legallybindtheirfutureselves.I mayowe the programmoneyifI failto showup but I cannotlegallyhave
myfutureselfenjoinedor imprisonedto make sure he stays on the diet. See generally
EinerElhauge,AllocatingHealth Care Morally,82 Cal L Rev 1449, 1527-28(1994); Thomas C. Schelling,Choiceand Consequence96-99(Harvard1984). In any event,allowing
thepresentto bindthefuturein thatcase is morejustifiablebecause theindividualseeks
to bindhimselfto theweight-watching
program.Here,sucha theoryoftermlimitswould
meana currentelectorateseeks to bindfutureelectorateswhoto someextentwillconsist
ofdifferent
individualsentirely(as personsmove,die, and registerto vote),and the currentmajoritycoalitionseeks to binda futuremajoritythat almostcertainlywill include
different
personsas coalitionsshift.

172
The University
ofChicagoLaw Review
[64:83
capacityto retainincumbentsmattersonlyif the incumbentis

uniquelygood at representing
the electorate.In districtsaveraging572,000persons,314
and statesofup to 30 millionresidents,
this seems implausible.The supply of potentialcandidatesis
large enoughthat someoneelse will almost always exist who
shares the same viewsand abilityas the terminatedincumbent,
to thesecondbestrepresentative
is unlikely
or at least thefalloff
to be large.In contrast,a regimethat makes it hard to displace
incumbentsmattersas long as any of the many alternatives
would be preferable.And the dropoffbetweenthe views and
abilityofthe best candidateand the entrenchedincumbentcan
be enormoussince the incumbentcan be far below the second
bestoptionin thejurisdiction.
It mayevenbe thatthe
Certainlythiscalls forsometradeoff.
righttradeoff
todayis different
than the righttradeoff
twohundredyears ago. Back thensmallerpopulationsmeanta smaller
to thesecond
supplyofcandidatesand perhapsa sharperdropoff
best representative.
Further,in thatera, lowertenure,porklevels, and brandnameadvantagesmeantless coercivepressureand
lowerentrybarrierlevels.But the key pointis thatthe tradeoff
effects.
We cannotdismissa
is betweendifferent
antidemocratic
effects.
We can only
votingregimebynotingit has undemocratic
optimizeby choosingthe votingregimethat minimizesoverall
antidemocratic
consequences.
Moreover,the constraintthat termlimitsimposeon future
electoratesis not that large fortheymay always repeal term
limitsif theylike. Almostall termlimitsare imposedby voter
initiative,
and a simplemajorityin a futurevoterinitiativewould
sufficeto liftthem.315
Thus,ifa futureelectoratearrivedat a differentview of the tradeoff
betweenconstraintson its abilityto
displace and retain incumbents,it could always release itself
fromthe shacklesoftheviewheld by the past electorate.In this
respect,ironicallyenough,the SupremeCourt'sdecisionforcing
congressionalterm limits to be imposed via constitutional
amendmentmakes themmoreantidemocratic
because a future
majoritywould not be able to repeal constitutionally
imposed
termlimits.
""See Will,Restorationat 57 (cited in note 10).

3I1In some states, bans on successive initiatives on the same subject matter require
some waiting period beforea priorinitiativecan be repealed, but these are usually not too
long. See ReferendumComm ofHermosa Beach v Cityof Hermosa Beach, 184 Cal App 3d
152, 229 Cal Rptr51, 54 (2d Dist 1986) (one-yearwaiting period).

1997]
173
2. The interference
withindividualdistricts.
The otherconcernis not about the past restrainingthe future,but about the whole restrainingthe part. For example,a
state electorate'sdecisionin favorof termlimitsmay interfere
with an individualdistrict'sdemocraticpreferencein favorof
keepingits incumbentin office.316
It is importantto recognizethat this problemis not raised
everytime an individualdistrictwould, absent termlimits,reThe collectiveactionproblemsidentified
tain its representative.
in Part II mean thateverydistrictmight,absenttermlimits,prefer to keep its incumbent.Nonetheless,each of those districts
may prefertermlimitsbecause it prefersthe representationit
gets with limits to the representationit gets withoutlimits.
Likewise, the entrybarriers describedin Part III may mean
everydistrictprefersits incumbentto the available challengers
but wouldstillbe betteroffwithlowerbarriersand new entry.
Still, it cannotbe gainsaid that sometimesthe problemis
raised. In particular,as Part II.C. 1 noted,some districtsrationally opposetermlimitsbecause of a combinationofhigh seniority,low ideologicalinterest,high valuationofporkbenefits,and
low share ofporkcosts.The numberofsuch districtsmay not be
large.And,again,the democraticcostis unlikelyto be large since
termlimitsonlyprecludeone candidateand thereis no reason to
expectthe nextbest alternativeto be muchworse.Nonetheless,
such a district'sdemocraticpreferencein favorof retainingits
incumbentis thwartedto furtherthe democraticpreferencesof
the state electorateas a whole.
But thisis notunusual. A politicaljurisdictiongenerallycan
enact legislationto furtherthe substantiveinterestsofthejurisdiction'smajorityeven thoughthat conflictswiththe preferences
of particularregionsin the jurisdiction.That is inherentin the
supremacyof state law over municipallaw, and of federallaw
over state law. And it is particularlyjustifiablewhere,as here,
the rationaleforimposingthe views of the whole on the part is
thatthepartis imposingexternalitieson otherregions.3"7
Indeed, there is nothingin the Constitutionthat requires
statesto allowvotersto voteby districtforits federalrepresentatives. The Constitutionrequires only that each state sends a
number of representativesto the House proportionateto its
It is perfectly
foreach state to hold a
constitutional
population.3"8
forexample, Bates v Jones, 904 F Supp 1080, 1095 (N D Cal 1995).
3'7See Part II.C.1.
18US Const, Art I, ? 2, cl 3; US Const, Art I, ? 4; US Const, Amend XIV, ? 2.
316See,

174
The University
ofChicagoLaw Review
[64:83
statewidevote forall its representatives

and the framersfully
expectedmany to do so."' Only seven of the originalthirteen
states initiallydecidedto divideinto districtsthat would each
If the state's electoratecould
vote forits own representative.320
constitutionally
imposeits views about the identityof each representativeon individualdistricts,it surelycan constitutionally
imposethe less restrictive
alternativeoflimitingeach representative'sterms.
3. The interference
withBurkeanrepresentative
democracy.
A morefundamentalobjectionchallengesmyverydefinition
of democracy.We do not, the argumentgoes, want a system
where each representativeslavishlyfollowsthe latest poll reand undeliberative
views ofthe electorflectingthe uninformed
a
ate. Rather,we wantand have systemofrepresentative
democracy,wherevotersdo notdirectly
voteon issues but ratherselect
a representative
to gatherinformation,
deliberateand do what
the representativethinks is righteven if that conflictswith
opinionpoll evidenceof electoratewishes. Under this Burkean
the last step of the arconceptionof representative
democracy,
gumentgoes,ideologicalslack mightseemgoodand its reduction
bytermlimitsthusundesirable.
But thislast step is a falseone. Even a Burkeanconception
of democracymust ultimatelyrest on the conclusionthat it
somehowmakesvotersbetteroff.Thatis, werewe aboutto enter
intoa social contract,we wouldrealizethatour ownuninformed
decisions on futureissues would produce worse decisionsdecisionsless reflective
ofourtruevalues and preferences-than
withtime to make informed
decisionmaking
by representatives
judgmentson our behalf.Thus, directdemocracycould be expectedto producedecisionsthat less accuratelyreflectthe electorate'sconsideredpreferences
thanrepresentative
democracyif
those representatives
complywith a Burkean dutyto act as a
trusteeof the electorate'sinterests.Electionsshouldinstead be
aboutvotingforthebestBurkeandecisionmakers.
Thereis nothingin thistheorythatunderminestermlimits
or supportsthe currentsystem'seffectivepenaltyon ousting
seniorrepresentatives.
If electorateswant Burkean legislators,
3"9See US TermLimits, 115 S Ct at 1903 & n 30 (Thomas dissenting).The state power
to elect at-large representativeswas ended by statute in 1842. See 5 Stat 1491, codifiedat
2 USC ? 2 (c) (1994) (requiringall House elections to be held by district).But the constitutional point remains valid.
31 US Term Limits, 115 S Ct at 1903 (Thomas dissenting).

1997]
175
theycan continueto vote forthem under termlimitsand they

have incentivesto do so sincesuch legislatorson balance produce
betterrepresentation.
Indeed,theyare freerto do so withterm
limitsbecause termlimitslower the penaltyforoustingsenior
representativeswho may be far worseBurkean decisionmakers
than theiropponents.Since this penaltyapplies irrespectiveof
how the senior representativesvote, it does nothingto assure
behave in a moreBurkean fashion.Inthat the representatives
deed, one of the major argumentsby termlimitsproponentsis
thatcurrentlegislatorsare fartoo proneto be nothingmorethan
glorifiedpoll-watchersand that freshlegislatorswould be more
likelyto engagein trueBurkeanrepresentativedemocraticdecisionmaking.32'
In addition,the theorydoes not show that the sortof ideological slack reducedby termlimitsis good. An electoratemay
want a Burkeanlegislator,but that is because it wants someone
who does not followuninformed
polls on each individualissue,
not because it wants someonewho deviates fromthe ideological
views of the electorate.An electoratemay well be betterrepresentedby someonewho shares the electorate'sideologicalviews
and promisesto make an informed
judgmentover particularissues that arise than theywouldbe by someonewho promises,in
Perotian fashion,to have an electronictown meetingto make
each futuredecision. But a penalty of diminishedpower for
oustingincumbentscan produceincumbentswho have different
aspirationsthan the electorate.Such incumbentsneed not use
theirideologicalslack to fulfilltheirdutyto act as trusteeforthe
electorate'sinterestsand views-they can pursue more venal
goals or ideologicalviews at odds withany consideredjudgment
the electoratewouldreach.322
321 See, forexample,
Will,Restoration
at 99-101,107-17,142-45(citedin note10); Michael J. Malbinand GeraldBenjamin,LegislaturesAfterTermLimits,in Benjaminand
Malbin,eds, LimitingLegislativeTerms209, 211 (citedin note 5); AmihaiGlazer and
MartinP. Wattenberg,
How WillTermLimitsAffect
LegislativeWork?,in Grofman,
ed,
LegislativeTermLimits37,40-41(citedin note5). Iffollowing
short-term
pollsconfersan
electoraladvantage,as it mustto conflict
withBurkeandecisionmaking,
thenseniorincumbentshave strongincentivesto followthembecause doingso maximizestheirideologicalslack,whichtheycan use to pursuetheirownviewsor interestson othermatters.
Havingmoreideologicalslack does not mean one is morelikelyto use it in a Burkean
manner.
322 Underanother
use ofthe term,a "Burkean"legislatoralso pursueswhateverpolicies he paternalistically
regardsas in the electorate'sbest interestseven ifthatconflicts
withwhathe believesthe electoratewouldthinkifit werefullyinformed.
See MichaelE.
Levine and JenniferL. Forrence,RegulatoryCapture,Public Interest,and the Public
Agenda:Towarda Synthesis,
6 J L Econ & Org 167, 177 (Special Issue 1990).The attractivenessofthisseems dubious,and to call it "democratic"
seems moredubiousstill.But
evenifvalid,sucha theorymustreston thepremisethatthe Burkeanlegislatoris some-

176
The University
ofChicagoLaw Review
[64:83
There are also more specificproblemswith the argument

that termlimitsare bad because theyreduceideologicalslack.
First,even the most extremeBurkean agrees that representativesshouldfacereelectionand thus be somewhatpoliticallyacbetween
countable.Theythus implicitlyadvocatesome tradeoff
But whatis it about
ideologicalslackand politicalaccountability.
the mix of slack and accountabilitywithoutterm limits that
wouldlead anyoneto concludeit was optimal?Unless the objection to termlimitsrests solely on an unjustifiedbias forthe
bestatusquo, it mustexplainwhat makes the currenttradeoff
tweenideologicalslack and politicalaccountability
optimal.And
Second,ifwe want legsuch an explanationis neverprovided.323
islatorswithmoreideologicalslack,it is notat all clear whywe
in proportion
to each disshouldwantthat advantageconferred
trict'sseniority.
Whyshouldsomedistrictsgettherightdegreeof
slack and notothers?A moreequitablesolutionwouldbe to advocatelongertermsand fewerelectionsforall representatives,
moreideologicalslack equallyon all districts.324
thus conferring
alreadymakes a judgmentabout howofThird,the Constitution
makes the tradeoff
beten to hold electionsand thus implicitly
tween slack and accountability.Senatorsget more slack than
but neithergets ten-yearor lifetimeterms,
Representatives,
whichwouldconfereven moreslack. By reducingthe ideological
slack allowableat elections,termlimitssimplyhelp make those
how a "better"decisionmakerthan an informedpolitywould be. It thus does not counsel
against term limits because there is no reason to think that without term limits such
"better"decisionmakerswould be selected (their selection may be penalized) or that the
increased ideological slack would be used to make "better"decisions (it may be used to
make even worse ones than an informedpolitywould). Such a Burkean theoryalso suffersfromall the problemsmentionedin the next textual paragraph.
:"Likewise, opponents oftenargue that term limits would shiftpolitical power from
Congress to the President. This assertion seems highlydebatable. A term-limitedCongress need not be weaker in battles with the President and may even be tougherbecause
it is likely to be both (1) less focused on preservingscraps of pork and (2) more tightly
bound by districts'ideological demands that may forcelegislators to do battle. But the
more fundamentalproblem is that opponents making this argument never explain precisely what it is about the currentbalance of legislative and executive power that leads
them to believe we are now at the optimal point. Surely it is not because it reflectsthe
original constitutionalunderstanding,because since then we have added term limits for
the President but not Congress. Nor, even if we focus solely on experience (as opponents
generally do), do term limits appear to result in a mismatch. Under six-to-twelve-year
term limits,the average legislator would still sit longer than the average president and
much longerthan the average cabinet member.Instead, this argumentalso seems to rest
on the unjustifiedpremise that deviatingfromthe status quo is bad.
""Empirical studies show legislators' ideological slack shrinks significantlyas elections approach. See Kalt and Zupan, 33 J L & Econ at 126-27 (cited in note 145); Levitt,
86 Am Econ Rev at 435-36 (cited in note 146); Joseph P. Kalt and Mark A. Zupan, Captureand Ideologyin theEconomic TheoryofPolitics, 74 Am Econ Rev 279, 298 (1984).

19971
177
and thus betterreflectthe tradeoffimelectionsmoreeffective

plicitin the Constitution.
More generally,I see littlereason to believethat ideological
slack is on balance desirable.We can all recall profile-in-courage
cases where the use of slack seems desirable. But historyalso
There is no a priorireason to
evidencesplentyofprofiles-in-evil.
believe slack is morelikelyto be exerciseddesirablythan undereason.
sirably.True,we mightthinkofa historicallycontingent
The currentlong-term
incumbentstend (tended?)to be moreliberal than the currentelectorate,firstelectedwhenold-fashioned
liberalismwas morepopular.To the extenttheyare (were?)exercising their slack to furtherliberal goals, and one likes those
goals, thenthe slack wouldbe desirable.But that would onlybe
persuasiveto thosewho share thoseliberalviews and are willing
to see themimposedagainst the will of the majority.(This may
be yet anotherreason whytermlimitshave been more popular
than liberals.)And the persuasivenesswould
withconservativesvanish with any reversal in the relative politicalideologiesof
seniorincumbentsand the electorate.
question
Finally,we mustrememberthat the constitutional
turnson whetherthe votersimposingtermlimitswere furtherdemocraticobjective,notjust an objective
ing an impermissible
on whichreasonablepeoplemightdiffer.Even if one could make
an argumentthat having Burkean representativeswith more
ideologicalslack would be desirable,nothingin that argument
explainswhyvotersshouldnotbe freeto rejectthatBurkeanargumentand adopt an electoralsystemproducingrepresentatives
more
withless ideologicalslack. Seekingto make representatives
accuratelyrepresenttheirelectoratesmaynot strikeeveryoneas
desirablebutit surelyis notan unconstitutional
objective.
B. TheyIncreaseInterestGroupInfluence
Disgust withinterestgroupinfluencewas a major impetus
But opponentsargue that term
to the termlimitsmovement.325
limits perverselyincrease, not decrease, interestgroup influence.36 To analyze these claims, we must firstunderstandthe
logicbehindpublicchoicetheory'spredictionthatinterestgroups
willhave disproportionate
politicalinfluence.
325
326
See, forexample, Will, Restoration at 56-57 (cited in note 10).

See, forexample, Polsby, Am Prospect at 42-43 (cited in note 5).

178
ofChicagoLaw Review
The University
[64:83
1. Interestgrouptheory.
One of the puzzles of democracyis that the majorityoften
loses, even whenit has a largerstake in the matter.A smaller
groupwithless aggregateinterestin an issue can nonetheless
beat a largergroupwitha greateraggregateinterest.Oil producers might,forexample,get an inefficient
governmentsubsidy
eventhoughtheyare outnumbered
bythe taxpayerswhopay for
it. Publicchoicetheoryexplainsthisphenomenon
by notingthat
all groupsaffected
by legislationfacea collectiveactionproblem
in politicaleffort,
whichsmall interestgroupsare generallybetterable to overcome.
Obtainingfavorablepoliticaloutcomesrequires significant
oftime,money,and effort.327
contributions
The mostobviousare
whichcan take the formofmoney,supcampaigncontributions,
orvolunteerhelp.These can notonlyhelpinportiveadvertising,
fluencewhogetselectedbutalso cause politiciansto modify
their
views.But evenifsuch campaigncontributions
werebanned,political participationwouldbe costly.It is costlyto monitorand
lobbylegislaturesand regulators,to evaluate the positiveand
negativeconsequencesoftheiractions,and to communicate
this
information
to lawmakersand citizens.Further,it is costly(in
timeand effort)
forthosecitizensto absorbthe information
and
thenact on it byvoting,protesting,
writingtheirrepresentative,
or showingup at hearings.The morecomplexand technicalthe
issue, or the lowerthe preexisting
level ofvoterknowledge,the
willbe. For
highertheinformation
costsofpoliticalparticipation
I will call boththe makingof monetarycampaign
convenience,
contributions
and the expenditureoftime,money,and effort
to
engage in informed political participation making a
to thegroup'spoliticaleffort.
"contribution"
For severalreasons,the individualbenefitsof makingcontributions
to a group'spoliticaleffort
are oftenslighteventhough
the group'sstake is large.To beginwith,because the benefitsof
favorably
resolvinga politicalissue are spread amongthe membersofthebenefited
group,thebenefitscan be low foreach individual groupmembereven thoughtheyare high forthe group
In addition,each individualhas an incentiveto free
collectively.
ride on the contributions
ofothergroupmemberssince an individual'sfailureto contribute
willnotexcludehimfromwhatever
collectivebenefitsflowfromobtainingor defeatinga givengovernmentalaction.Moreover,the likelihoodthat any particular
individual'scontribution
will affectthe outcomeis minuscule.
27
See generallyElhauge, 101 Yale L J at 36 (cited in note 79).

19971
179
the majoritywill oftenremain rationallyignorant

Accordingly,
and apatheticbecause the costs of informedpoliticalparticipationexceedthe personalbenefits.
Special interest groups have several political advantages
over the general public.First,while the general public is large
and diffusely
interestedin manyissues, special interestgroups
are typicallysmall and intenselyinterestedin a limitedset ofissues. Theirmembers'intenseinterestmeans that,forthat set of
issues, theirmembersderivegreaterindividualbenefitfromsuccessful group effortsto influencethe political process. This
makes it morelikelythatpercapita benefitswill exceedthe individual costs of makinga contribution
to the group'spoliticaleffort.It also gives their membersless incentiveto free ride on
othermembers'contributions,
particularlywhen coupledwith a
small groupsize thatmakes it easier forothermembersto notice
whetherindividual membershave failed to contributeto the
group'spoliticalactivities.328
Second,special interestgroupsfacelowercostsin organizing
and enforcing
collectiveagreementsto police freeriding.329
Their
small size makes it easier to identify
otherswho have coinciding
interestsand to reachcollectiveagreementsto share costsby,for
example,requiringall membersto contribute
yearlydues to fund
permanentstaffor to take turnsgivingtestimony,sendingtelegrams, or making campaigncontributions.It also means that
freeridingwill be easier to detectbecause it has a proportionally
larger effectand because membersof small groups are more
likelyto have ongoingpersonalcontact.
Third,special interestgroupsoftenface lower information
costs. Monitoringis less costlybecause special interestgroups
are, bydefinition,
onlyconcernedwithmonitoring
a limitedset of
issues. Perhaps moreimportant,
theirmembersalreadyhave expertiseor day-to-dayinvolvementin the matterat hand. This
makes it easier fortheirmembersto absorb and act on information about potentialgovernmental
action. Oil producersare, for
"Small, intenselyinterestedgroupsare less susceptibleto freeridingin threedifferent senses. See id at 36-39 & nn 24-26.First,theyare morelikelyto have individual
memberswho benefitenoughfromfavorablegovernmental
decisionsto have an independentmotiveto makecontributions
furthering
the group'spoliticalgoals. Second,they
are morelikelyto have memberswhoare sufficiently
large in relationto the size ofthe
groupthattheyhave interdependent
motivesto contribute
because theirfailureto do so
wouldbe noticedbyothermembers,
whowouldrespondbywithdrawing
theirowncontributions.Third,wherean incentiveto freeridedoes exist,it is likelyto be lowerbecause
the divergencebetweenthe costs and benefitsof contributing
to groupeffortswill be
lower,meaningthat the expectedpenaltynecessaryto curbfreeridingwill be loweras
well.
3'29 Id at 38-39,40 n 35.

180
The University
ofChicagoLaw Review
[64:83
judgingwhethera new oil

example,likelyto have littledifficulty
refiningregulationis likelyto benefitthemor not; membersof
because they
the generalpublicwill have much moredifficulty
are less familiarwithhowtheoil industryworks.
To be sure,even special interestgroupsare large enoughto
face massivecollectiveactionproblems.This explainswhygovernmentexpendituresare 1250 to 2500 times largerthan the
made to influencethose expenditures.330
campaigncontributions
politicaloutcomesis relativepoliticalinfluBut whatdetermines
can thus
ence,and groupsthatmake somepoliticalcontributions
beat largergroupsthatdelivernone.
2. Termlimits'ambiguouseffect
on interestgroupinfluence
overcandidates.
have not done a goodjob explaining
Termlimitsproponents
preciselywhytheybelievelimitson termswouldlessen interest
group influence.The tendencyhas been to note all sorts of
government
spendingand regulationand
seeminglyinefficient
would
reduce it.33'But these
that
term
limits
then to assert
claims fail to distinguishbetweengovernmentalinefficiencies
caused by interestgroupinfluenceand those caused by district
collectiveactionand entrybarrierproblems.To someextent,the
problemis that the argumentabout interestgroup influence
overlapsclaims alreadydiscussedin Parts II and III. We thus
330
See John R. Lott, Jr.,A Simple Explanation for Why Campaign Expenditures are
Increasing: The Governmentis GettingBigger 25 (Mar 26, 1996 drafton file with U Chi L
Rev) (puzzling about "[w]hyare Federal budget expenditures2500 times larger than total
Federal legislative campaign expenditures and State budget expenditures 1250 times
larger than total state legislative campaign expenditures?").
3 See, forexample, Will,Restorationat 17-36 (cited in note 10). One frequently-made
argumentis that,even if we do not know why term limits would reduce interestgroup influence,they must reduce it because interest groups typicallylobby against term limits.
See, forexample, Reed and Schansberg, 76 Soc Sci Q at 711 (cited in note 131). See also
Olson, Term Limits Fail, in Benjamin and Malbin, eds, LimitingLegislative Terms at 76,
86-89 (cited in note 5) (detailing that in Washington term limits had only one interest
group supporterwho providedonly 1 percentof all fundingwhile the opposition'ssupport
came almost entirelyfrominterest groups). This revealed preferenceargument is interesting but unpersuasive on its own. Interestgroups have made large sunk investmentsin
influencingthe current set of incumbents. They can be expected to lobby against any
change that reduces the value of those past investments even if the groups would continue to have an advantage in influencingnew legislators that is as large or larger than
before the change. See also Copeland, Political Careers in Oklahoma, in Benjamin and
Malbin, eds, LimitingLegislative Terms at 154 (cited in note 272) (noting that "lobbyists
see the task of influencinglegislators becoming less efficientbut more effectiveunder
term limits"). See also Elizabeth A. Capell, The Impact of Term Limits on the California
Legislature: An InterestGroup Perspective,in Grofman,ed, Legislative Term Limits 67, 81
(cited in note 5) (lobbyist concluding that influencinglegislators will become less efflcient).

1997]
181
and excludethisoverlapto see whetheranything

need to identify
remains to the argumentother than the points already discussed.
Sometimesa particularinterestgroup's interestscoincide
withthe interestsofa district.Thus, federalsubsidiesformohair
productionboth reflectedthe disproportionate
influenceof moin
hair producers all districtsand the disproportionate
benefitof
such subsidiesin certaindistrictswhereproductionwas concenWhere the productof interestgroupinfluencedisprotrated.332
benefitsa districtin a way thatmakes districtbeneportionately
fitsexceed districtcosts even thoughnational costs exceed nationalbenefits,it fitswithinmydefinition
of"pork."We can thus
expectthis to be reducedby termlimitsforthe reasons already
coveredin Part II. But thereis no generalreason to expectinterest groupcaptureto fitor be explainedby porkbarrel politics.
And if it fit,that would add nothingto the explanationalready
in Part II. Let's thusexcludecases ofpork.
offered
Relatedly,the currentincidenceof interestgroupinfluence
mightbear on the proportion
ofnonporkone believesis undesirable. Suppose one believes that legislation favoringinterest
groupsis undesirableand that most nonporkis interestgroup
legislation.Then a personwho believes that decreasingaverage
tenure decreases overall nonporklevels would findterm limits
desirablein partbecause mostofthe nonporkdecreasedwas undesirable.But this benefitof termlimitscomes solely fromthe
implicationsof this beliefabout the overall desirabilityof nonpork,whichI have alreadydiscussed in Part II.B.3. It is not a
claim that termlimitsselectivelydecrease interestgroupinfluence comparedto otherinfluences.
A selective discouragementclaim is made by term limits
proponentswho argue that legislativeserviceresults in legislatorsmorelikelyto favorinterestgroups.For example,legislative
service mightbe viewed as affirmatively
or corindoctrinating
the commonsense or moralprinciplesthat
rupting,undermining
wouldotherwiselead legislatorsto fendoffinterestgroups.333
Alone mightthinkthat those who enjoy makinggovternatively,
ernmenttransfersare morelikelyto want to stay in officewhile
those who do not are morelikelyto want to leave.334
But others
have the opposite intuition.They argue that because term332
See, forexample,Will,Restoration at 23 (citedin note10).

See, for example, James L. Payne, Why Government Spending Grows: The
"Socialization" Hypothesis,44 W Pol Q 487, 487-89 (1991); Mark P. Petracca, Term Limits
Will Put an End to Permanent GovernmentbyIncumbents,Pub Affairs8, 8 (Nov 1990).
33 I am indebted to John Lott forthis point.

182
The University
ofChicagoLaw Review
[64:83
limitedlegislatorswillon averagebe less experiencedand expert,

theywill be moredependenton interestgroupsforinformation
and expertiseaboutthesubjectmattersbeingregulatedand thus
moreeasilybamboozled.335
claims are hard to evaluate. But the more
Such conflicting
are
important
pointis that,to the extenttermlimitsproponents
right,this is merelya subspeciesof ideologicalslack. If senior
legislatorsfavorinterestgroupsmorethan the electoratewould
like,the electoratecouldjust votethemout of officeabsent the
collectiveactionand entrybarrierproblemsdiscussedin Parts II
and III. Andwhilethosecollectiveactionand entrybarrierproblems do provideseniorlegislatorssomefreedomto favorinterest
groupsin ways the electoratedisfavors,that freedommust be
tradedoffagainst otherways of exploitingideologicalslack. A
his electorateby
to offend
legislatorwhohas a greaterpropensity
interestgroupscan deviateless on otherissues fromhis
favoring
electorate'swishes.The reasonswhytermlimitsmightgenerally
decreaseideologicalslack,whetheror not the issue involvesinterestgroups,have alreadybeencoveredin PartsII and III.
However,theremaybe somegroundsforthinkingthatterm
limitscan changethedirectionin whichideologicalslack is exerindividuals
cised. Suppose that,no matterwhat theirseniority,
would have to transgressinternalmoral principlesif theyfavoredspecial interestgroups,whichwouldcause themto suffer
costwere the
cost.Even if this psychological
somepsychological
same forboth seniorand junior legislators,that cost mightbe
worthbearingin exchangefora longlegislativecareerbutnotfor
a shortone. By shortening
legislativecareers,termlimitsthus
reducethepotentialpersonalbenefitsoffavoringspecialinterest
groups.Simplyvotingout seniorincumbentscannotaccomplish
thesame goal,becauselegislatorswouldstilllookto the prospective benefitof a longlegislativecareerunless it is categorically
denied.But again,anylesseningofideologicalslack in the direction of interestgroupswould predictably
be offsetby increased
(and no less damaging)ideologicalslack in otherdirectionsunless the overallamountof ideologicalslack were somehowreduced.The net benefitoftermlimitswouldthus stillcomefrom
decreasedideologicalslack along the lines describedin Parts II
and III.
No, themainsensein whichinterestgrouptheorymightadd
somethingto the previousdiscussionis to explain changes in
whichpreferences
theelectorateexpresses.Interestgrouptheory
33
See Polsby,Am Prospect at 42-43 (cited in note 5).

1997]
183
is, afterall, a theoryabout the politicalparticipationand voting

behaviorthatelectoratesactuallymanifest.Thus, an increase in
interestgroupinfluencemightcause the electorateto expressdifferentpreferencesthan it would otherwise.And if this were
viewed as undesirable,curbinginterestgroup influencewould
resultin the electorateexpressingbetterpreferences.
This would
improvesocial welfareeven if the legislator'sideologicalslack
fromthe electorate'sexpressedpreferences
were unchangedand
overallporklevelswereunaffected.
But thereis littlereasonto thinktermlimitswoulddecrease
interestgroupinfluencein this manner.The factorsthat make
interest group influencedisproportionatewould persist with
termlimits.Special interestgroupswould still have the advantages offewerfreeriderproblemsand lowerorganizationaland
information
costs.Theywouldstillbe able to delivera disproportionateshare of campaigncontributions,
lobbying,and informed
voting.This would give candidates (whethernewcomersor incumbents)runningunder termlimitsjust as much incentiveto
favorinterestgroupsas incumbentsrunningwithouttermlimits.
And it wouldgive interestgroupsjust as muchdisproportionate
influenceoverwhogetselectedas theyhave now.
Indeed,Nelson Polsbypersuasivelyargues that termlimits
will increase,not decrease,interestgroupinfluenceover candidates.336
Because incumbentscan relyon theirabilityto deliver
governmentalbenefitsand name recognitionadvantages, newcomersmust expend much more moneyon campaignsthan incumbentsto get elected.The averagewinningchallengerspends
farmorethan the incumbent.337
This makes incumbentsless deIn effect,
pendenton interestgroupsforcampaigncontributions.
Polsby cleverlyconvertsthe vice of the currentsystem-senior
incumbents'greaterinsularityfrompolitical pressures-into a
virtuebecause it makes them less vulnerableto interestgroup
pressures.
This argumentseems descriptivelyaccurate. The greater
clout and brandnameadvantages that give senior representatives more slack from the electorate's preferencesgenerally
should also give themmore slack fromthe portionsof the electoratebelongingto interestgroups.But shouldwe regardthis as
normatively
attractive?The shieldingfrominterestgroupinfluence is, afterall, purchasedonlyat the costofgenerallyshielding
the politicianfromthe influenceof his electorategenerally.In3e36Id.
33
See text accompanyingnote 305.

184
The University
ofChicagoLaw Review
[64:83
terestgrouptheoryexplainsvariousdistortionsin the political

processbut providesno reasonto thinkthe distortions
wouldbe
smallerwithno politicalaccountability
at all.338
Moreover,one of the costs of a systemthat makes incumbentsindependentis thatit insulatesthemfromtheirpartiesas
well.Underourcurrentsystem,individuallegislatorstreatparty
performance
as a collectivegood.339
To each legislator,her elecofher party'snationalpertoralchancesturnon a combination
formance(especiallyin the Presidency)and her own personal
But theindividuallegislatorcan have relativelylitperformance.
tle effecton improving
the party'sperformance,
and she benefits
fromimprovedpartyperformance
whethershe contributesto it
or not.Thus, legislatorsfocuson improving
theirindividualperformanceby deliveringdistrictpork and on improvingtheir
namerecognition
byraisingfundsfortheirowncampaigns.If incumbencyadvantageswere reducedby termlimits,incumbents
mighthave to rely more on partysupportin elections,which
would improveparties' ability to disciplinelegislators.This
shouldlessen porkbecause the partyhas incentivesto take into
It should
accountthe costs of porkto the national economy.340
costsin ascertaininglegislativepoalso lowervoters'information
sitionsbecause partyplatforms
are fewerand moresalientthan
a myriadofpositionpapersbyindividuallegislators.
Even ifwe put aside thedecreasedfitwiththe partyand the
it is hardlyclear thata politirestoftheelectorate'spreferences,
cian following
his own ideologicalpreferences
is moredesirable
thanone whofollowsthepreferences
ofinterestgroups.First,we
can condemninterestgroupinfluenceas disproportionate
onlyif
we have somenormativebaselinefordetermining
what degreeof
We do not
influenceis "proportionate"
to the group'sinterest.34'
Compare Elhauge, 101 Yale L J at 83-84, 86-87 (cited in note 79).
3 See Mayhew, Congress at 30-32 (cited in note 255).
"0Lessen, but not eliminate, pork because parties still have incentives to provide particularized benefits to districts (like to any interest group) because those benefits are
more concentratedand noticeable than the diffusecosts. See Parts II.B.3; IV.B.1. See also
Carey, Term Limits in Costa Rica, in Grofman,ed, Legislative Term Limits at 328-32
(cited in note 132) (noting that even under proportionalrepresentation system in Costa
Rica, parties organize at-large deputies to provideparticularizedbenefitsto sub-areas for
political gain). But while the costs are less noticeable, they are not entirelyexternalized
the way theyare when an individual legislatorgets pork enacted and inflictscosts on districtsin whichhe does not run. A partythus sufferscosts fromexcess pork. See id at 33233, 336 (findingthat providingmore pork is correlated with reductions both (1) in votes
for the party as a whole and (2) in the chances of post-legislativeappointment by the
party).
"'For my full argument on this point, see Elhauge, 101 Yale L J at 48-66 (cited in
note 79).

1997]
185
always want a group'sinfluenceto be proportionate

to its numbers-consider a racial minorityopposingracist legislationor a
landowneropposingan uncompensatedtaking. To say instead
that we want each group'sinfluenceto be proportionate
to its
economicstake in an issue is the same as judgingall legislation
by its economicefficiency.
And any alternativemeasure ofwhat
we want groupinfluenceto be proportionate
to (forexample,the
group'sutilityor howfarit fallsbelow an egalitarianwealthdistribution)is also the same as applyingthatnormativebaseline to
the legislation.We have no assurance that politicallyinsulated
legislatorswill applythenormativebaselineswe prefer.
Second, diminishinginterestgroupinfluenceneithermagically solves the collectiveaction problemsof the general public
noreliminatesall potentiallydistortingsourcesofpoliticalinfluence. Diminishedinterestgroupinfluenceleaves the media with
a proportionately
greaterrole, whichitselfprovokesgreatereffortsto buymedia outletsor otherwiseinfluencewhat the media
put out. Moreover,the vast majorityoflobbyingforgovernment
spendingis done not by interestgroupsbut by government
offilcials themselves.Thus, one study of witnesses testifyingon
spendingmeasures foundthat 1014 of the 1021 witnesseswho
tooka positiontestifiedin favorofspending,and 63 percentofall
witnesseswere government
officials.342
Even withoutall the witnesses representing
interestgroups,these government
witnesses
would remain.And collectiveaction problemswill still prevent
large diffusegroupsfromfundingwitnessesoftheirown.It is far
fromclear that a Leviathanregime,wheregovernment
officials
lobbyeach other,is preferableto what we have now.At least interestgrouplobbyingkeeps the government
accountableto some
ofthe interestsaffected
byits decisions.
C. The Final Period:Problemor Solution?
Under termlimits,legislatorswould moreoftenfindthemselves in what theyknowis theirfinalterm,holdinglegislative
officewithoutany prospectofreelection.True,underthe current
systemeverylegislatoreventuallyreacheshis finaltermtoo. But
he does so less frequently
since tenuresare longer.Moreover,often he does not knowhe is in his finaltermuntil afterhe loses
an election.Indeed,even the legislatorwho retiresduringhis fi"2See James L. Payne, The congressional brainwashing machine, 100 Pub Interest 1,
4-6 (1990) (47 percentfederalbureaucrats,6 percentotherfederallegislators,10 percent

state and local officials).
Another33 percentconsistedof privateinstitutions
receiving
government
Id at 6. In additionto the 1,021testifying
funding.
foror againstthe spendingbills,39 witnesseswereneutral.Id at 4-6.

186
The University
ofChicagoLaw Review
[64:83
nal termmustconsiderhis electorate'sresponseduringthe pedecision.The

riodofthattermbeforehe reachesthatretirement
onlycomparablesituationwithouttermlimitsis that oflegislatorsaftertheyhave reacheda retirement
decision,and such rethan a finaltermin a regime
tirementscomeup less frequently
oftermlimits.
Termlimitsopponentsview this as a problem,arguingthat
But proponents
viewit as
it willdiminishpoliticalaccountability.
a helpfulsolution,arguingthattermlimitswilldiminishinterest
groupinfluence
because incumbentlegislatorswill moreoftenbe
in whattheyknowis theirlast termand thusneed notpanderto
Note that this is not a claim
interestgroupsforreelection.343
aboutdecreasedinterestgroupinfluenceovernew or incumbent
whoare noncanbut onlyaboutincumbents
candidatesforoffice,
didatesbecause theyare in theirfinalperiod.The claim is also
consistentwiththe simultaneousvotesfortermlimits
perfectly
For the onlyway to obtainthis benefit
and againstincumbents.
is
of termlimits to committo a systemwhere the legislator
knowsshe is enteringher last term.Oustingincumbentlegislatorsthroughordinaryvotingwill not suffice,forlegislatorsremovedfromofficeby losingreelectionhave just as muchincentive to panderto interestgroupsbeforetheirloss as those destinedto winreelection.
But the claim faces the same problemas Polsby'sgeneral
claimthatseniorincumbents'
greaterpoliticalinsulationshields
themfrominterestgroupinfluence.3"
Namely,we have no reason
is preferableto disto thinkthat zero politicalaccountability
tortedpoliticalaccountability,
even whenthe distortionis alleginterestgroupinfluence.Moreover,to the
edlydisproportionate
extentthatinterestgroupinfluencetakes theformofinfluencing
whichtypes
the information
legislatorsreceive,or ofinfluencing
of personsbecomelegislators,it remainsundiminished.Those
legislatorswithideologicalviewscongenialto interestgroupsare
the ones likelyto be reelectedoftenenoughto reacha finalterm
undertermlimits,and theyare unlikelyto changethose views
all at once.
343
See, forexample, Will, Restoration at 32 (cited in note 10). See also Mills, 10 Pub
AffairsQ at 147 (cited in note 115) (same forpork); F.A. Hayek, 3 Law, Legislation, and
Liberty113 (Chicago 1979) (same; advocating a one-termlimitwith a fifteen-year
term).
34 See text accompanyingnote 336. By the same token, a term limits opponent who
argues that senior incumbents'greater political insulation withoutterm limits is a virtue
because it reduces interestgroup influencecannot consistentlyargue that an incumbent's
greaterpoliticalinsulation because ofmore final periods withtermlimitsis a vice.

1997]
187
Indeed,bothsides in the last-perioddebate are undermined

by empiricalstudies showingthat retiringlegislatorsdo not significantlychange their votingbehavior and that any changes
theydo make are similarto changesmade bythe averagelegislaNor,underpresidentialtermlimits,
torrunningforreelection.345
do we see Presidentsmake notablepolicyshiftsin theirsecond
term.Bothpieces ofevidencesuggestthatthe finalperiodis neither as big a solutionnor problemas eitherside suggests.The
lack of politicalaccountability
providedby the need to face reelectiondoes notappear to make muchdifference.
We saw above that there are two possible explanationsfor
how this can be true when ideologicalslack increases withtenOne possibilityis that electedofficialsfeelobligedto comure.346
plywithpromisesmade whenrunningin the electionleadingto
theirfinalterm.If so, theywouldfeelequally obligatedto carry
out the promisestheymade in the electionthat gotthemto their
finaltermundertermlimits.The otherpossibilityis that by the
timetheyhave reachedretirement
age, politicianshave so much
ideologicalslack (because of collectiveaction and entrybarrier
to go whenthe reelecproblems)thattheydon'thave any further
tion constraintis removedentirely.This degree of slack allows
them to vote their own preferencesby their penultimateterm
and leaves themwithlittledesireto changevotes when theyno
longerface reelection.We mightaccordinglyexpecta greaterfinal period change under term limits because politicianswill
reach theirfinaltermsoonerand thus have less slack in their
penultimateterm.But thatis trueonlyto the extentthat senior
legislatorsfacingreelectionwithouttermlimitshave an ideological slack problemjust as seriousas legislatorshave in theirfinal
termundertermlimits.Betterto have that ideologicalslack exploitedforone finaltermundertermlimitsthan forthe last decade or so oftermswithouttermlimits.
Moreover,under termlimitsmany legislatorsin theirfinal
termwould still face politicalaccountabilityin that theyintend
to seek anotherofficeaftertheyleave the legislature.The Senatorwhoplans to runforPresidenthas less incentiveto get inefficient pork forhis state enacted. For that matter,so does the
Senatorwho wants to be Secretaryofthe Treasuryor otherwise
Likewise, the Representative
part of the next Administration.
who plans to run forSenate, or the state legislatorwho plans to
run forstatewideofflceor a federalofflceencompassinga larger
3 See notes 154-55 (collectingsources).
" See text accompanyingnotes 157-65.

188
The University
ofChicagoLaw Review
[64:83
region,has less incentiveto favorpork for her district.This

seems entirelysalutary.And on issues otherthanpork,political
remains:thelegislatorin her finalperiodrunning
accountability
forhigherofficehas incentivesto representher districtaccuInratelyto maintaina reputationforaccuraterepresentation.
systemshows
deed empiricalevidenceundera non-term-limited
thatthosewholeave a legislativeseat notto retirebutto runfor
higherofficeare less likelyto divergeideologicallyfromtheir
Because legislaelectoratesthan thoserunningforreelection.347
tors in their final period under term limits are likely to be
office(at
younger,theyare also morelikelyto aspire forfurther
least sometimedownthe road) ratherthan simpleretirement.348
Thus, whiletermlimitsproducea greaternumberoffinalperiofthose in the
ods, termlimitsalso make a smallerproportion
finalperiodtrulyunaccountable.Whicheffectwill provegreater
is harderto predicta priori.
Finally,it is worthnotingthat most state-imposedterm
limitsdo not impose a final term.They impose a rotationrequirement,forcingthe incumbentto give up his seat (and thus
his seniorityand brandnameadvantage)fora termbut do not
prohibithim fromrunningagain in futureterms.Indeed,ofthe
statesthatimposedtermlimitson theirlegislators,
twenty-four
onlyseven-Arkansas,California,Louisiana, Missouri,Oregon,
NorthDakota,and Oklahoma-imposedthe sortofabsoluteterm
limitsthat creates this final periodproblem.Under the other
seventeenstates' rotationrequirement,incumbentsforcedto
leave officewouldhave had incentivesto refrainfromideological
divergenceto preservetheiroptionofrunningforthe seat again
in thefuture.
In any event,a legislatorundertermlimitsis less likelyto
makeit to his finaltermbecause he is morelikelyto lose reelection.And he is morepoliticallyaccountableduringthe termshe
House term
does run. This is anotherargumentfortwelve-year
limitsoversix-yeartermlimits,sinceit increasesthe proportion
ofyears withno finalperiodproblemfromfourin six to ten in
twelve.Takingall theaboveintoaccount,voterscouldeasilycon"See Zupan, 65 Pub Choice at 173-75, 178 & nn 14, 16, 18 (cited in note 156). See
also Lott and Bronars, 76 Pub Choice at 136-43 (cited in note 154); Van Beek, 19 Pub Fin
Q at 450-56 (cited in note 155). Zupan does show that such aspiring legislators' divergence increases duringtheir last term,which may indicate that theyhave already begun
to appeal to a larger constituency.Zupan, 65 Pub Choice at 176 (cited in note 156). But
that is likely to be socially desirable and in any event the factremains that the overall divergenceforthe last two-yeartermis lower than forlegislators runningforreelection.
348 Even withouttermlimits,42 percentof those retiringfromthe House ofRepresentatives in 1978 aspired forhigheroffice.Zupan, 65 Pub Choice at 169 (cited in note 156).

1997]
189
elude that any finalperiodproblemis small and outweighedby

in nonfinalterms.
moreaccuraterepresentation
Problem
D. The FutureEmployment
Term limitsmay increase the influenceof potentialfuture
Indeed,this problemhas been recogemployerson legislators.349
nized since the timeofAlexanderHamilton,who argued against
rotationon the groundsthat "Whena man knowshe must quit
his station,let his meritbe what it may,he will turnhis attentionchieflyto his ownemolument."350
Still,thereis reason to thinkthat this effectis small. First,
personswho servedin the legislatureare unlikelyto be lacking
forjobs. They will make attractiveemployeeswhetheror not
theyfavoreda particularemployer.
Second,each legislatoronlyneeds one futurejob. Even ifshe
favorsthat futureemployershamelessly,the magnitudeof the
divergencefromthe publicinterestthatthiscan cause is limited.
It is surelysmall potatoes comparedto the divergencepossible
when a legislatorfavorsall the groupswho gave her campaign
And it seems likelyto be smallerthan the divercontributions.
gencecreatedwhencollectiveactionproblemsand entrybarriers
giveincumbentsgreaterideologicalslack overthe wholerange of
possibleissues.
Third,thereare disincentivesto hiringlegislatorswho demonstratedpast disloyaltyto theirdistricts.A persondisloyalin a
legislativeofficemight also be expected to be disloyal when
workingforthe firm,perhaps to obtain an even betterposition
withthe firm'scompetitorsor opponents.A reputationfordoing
for
well at one's currentjob is normallythe best recommendation
futureemployers.Further,the legislatorsmostdesirableto hire
as lobbyistsare oftenpreciselythose who opposedthe firm'sinterestswhenin office.Such formerlegislatorsare morelikelyto
be persuasivewiththeirerstwhilelegislativeallies, who are the
Both effectsmay negate any inones the firmmust win over.351
centiveforlegislatorsto favorfutureemployers.
fuFourth,even if these disincentiveswere not prohibitive,
ture employerswould face legal and economicobstacles if they
Directlyoftriedto exchangejob offersforlegislativefavoritism.
feringa quid proquo wouldbe bribery,punishablewithciviland
"'Becker, Bus Wk at 18 (cited in note 5); Polsby, Am Prospect at 43 (cited in note 5).
US Term Limits,115 S Ct at 1860 (citation omitted).
"' See also Glazer and Wattenberg,How Will Term Limits AffectLegislative Work?,in
Grofman,ed, Legislative Term Limits at 42-43 (cited in note 321) (noting evidence that
industryhires ex-regulatorswho were hard on industry).
3

190
The University
ofChicagoLaw Review
[64:83
criminalsanctions.Employersmighttryto developa reputation

forhiringthose who favoredthemin the legislature.But they
wouldhave incentivesto renegethatmake commitment
difficult.
Moreimportant,
the employersin an industryseekingfavorable
legislative treatmentwould face their own collectiveaction
problem.Each has incentivesto freeride on otheremployers'
hiringoflegislatorswhofavoredthemin the past.
Fifth,whiletermlimitsincreasethenumberofex-legislators
seekingemployment,
termlimitsalso reducethe value of their
legislativeconnections.
Underthe currentsystem,a retiringlegislatormay have contactsin the legislaturefordecades. Under
termlimits,no contactscan last longerthan the termlimitperiod.A legislatorleavingoffice
undertwelve-year
termlimitshas
no contactsaftertwelveyears,and probablyno leadershipcontactsaftereightyears.
More important,
favoringa futureemployeris just one way
ofexercisingideologicalslack.352
Even if term-limited
legislators
have moreincentiveto favorfutureemployers,
seniorlegislators
withouttermlimitshave morefreedomto do so sincetheyhave
moretotalslack.Giventhecombination
ofthosefactors,it is not
clearwhichwouldin factfavorfutureemployersmore.Andeven
if term-limited
legislatorswould favorfutureemployersmore,
thatwouldnotincreasethe totalmagnitudeoftheirslack. Such
favoritism
wouldhave to be compensatedbyreducedexploitation
of slack in otherareas. There is little reason to thinksuch a
changein how slackis exercisedwill resultin a net harmto the
electorate.
In addition,influencebyfutureemployersneed notbe negative. The most importantfutureemployer,or at least the one
withthe fewestcollectiveactionproblemsin offering
futureemis likelyto be each official'spoliticalparty.353
ployment,
To the
extentterm-limited
officials
lookedto themforemployment,
that
wouldbe expectedto increasepartydiscipline.It wouldalso increaseofficials'
motivation
to helpthepartywinthenextelection
(and thus have the powerto offergovernment
employment).354
352See Levineand Forrence,
6 J L Econ& Orgat 179(citedin note322).
3In CostaRica,whichhas a forty-year
oflegislativetermlimits,70 percentof
history
interviewed
legislatorsintendedto seek further
politicalofficeand 91 percentofthose
expectedtheirpoliticalpartiesto securethema politicalappointment.
Carey,TermLim-
its in Costa Rica, in Grofman,ed, Legislative Term Limits at 325 (cited in note 132).
Forty-six
percentofthosewhoservedin the Assemblybetween1949 and 1986 actually
receivedsuchappointments.
Id at 325-26.
' From1974to 1990in CostaRica,64 percentofterm-limited
whoseparty
legislators
wonthenextelectiongota politicalappointment
immediately
following
theirlegislative
service;thosewhosepartylostreceivedan appointment
less than 1 percentofthetime.

1997]
191
Such enhanced party discipline and motivationshould have

prodemocratic
effects.It reducesporkin line withnational electorate preferences,and it lowers voter informationcosts in
choosingbetweencandidates.355
Finally,the empiricalevidencenow available does not indiis a large problem
cate that the prospectof futureemployment
forlegislators.A study of legislatorswho announcedtheir retirementfromthe legislaturefoundthat those planningto take
anotherjob with either the governmentor a private lobbying
than otherretiringlegislators.356
groupdid not vote differently
There is littlereason to thinklegislatorstakingotherjobs after
termlimitswouldbehave any differently.
Voters could thus rationallyconcludethat the futureemploymentproblemhas no significantnegativenet effecton the
and is morethan offsetby the
accuracyoftheirrepresentation,
positiveeffectof termlimits.Or theymightconcludethat any
negative effectcould be correctedwith laws restrictingpostlegislativework(foras long as theircontactsare fresh),just as
the United States now restrictsemployment
by those who leave
the executivebranch.
E. The Short-Term
ThinkingProblem
Othersargue thattermlimitswill increaselegislativeimpaThe premise is that legislatorsare excessivelyshorttience.357
because they must run for reelectionevery two
term-oriented
In thoseelections,votersdismisscommitments
to future
years.358
actionsand judge the legislatorbased onlyon the resultsshe has
The legislatorthus favorsgovernment
actuallydelivered.359
projreturnsthat she can showbeforeher reelecectswithshort-term
Id at 326-27.
355See text accompanyingnotes 339-41.

31 See Lott,52 Pub Choice at 177-79(citedin
note 154) (deviationfromnonretiring

congressmen
less than2 percent).
"'See, forexample,Cohenand Spitzer,80 Georgetown
L J at 479 (citedin note251);
Malbin and Benjamin, Legislatures AfterTerm Limits, in Benjamin and Malbin, eds, LimitingLegislative Terms at 219 (cited in note 321).
358See Morris P. Fiorina, RetrospectiveVoting in American National Elections 25-43
(Yale 1981)(Votersvotebased on already-delivered

governmental
benefits,
notthosethat
willbe deliveredin futurebecause ofcurrentgovernmental
S. Banks,
programs.);
Jeffrey
Linda R. Cohen, and Roger G. Noll, The Politics of Commercial R&D Programs, in Linda
R. Cohenand RogerG. Noll, eds, The Technology

PorkBarrel 53, 61 (Brookings1991)
(Programsgarnermoreelectoralsupportthe earlierthe benefitsare realized.);Linda R.
Cohen and RogerG. Noll, The PoliticalDiscountRate (CenterforEconomicPolicyResearchDiscussionPaper,No 209, 1990).
"' Cohenand Spitzer,80 Georgetown
L J at 487-88(citedin note251).

192
The University
ofChicagoLaw Review
[64:83
tion. Long termprojectsget rejectedeven if theyhave greater

presentvalue.
The currentadvantagesofincumbency
reducethis tendency
make
the
to short-termism
because they
incumbentmorelikely
ofgovernment
projectsin future
to considerthebeneficialeffects
years. Term limits,on this theory,have three baleful effects.
Theyeliminatethe possibilitythe legislatorwill considerbeneficial effectsin years beyondthe termlimit.They reducethe careerbenefitsofpositiveeffectsin yearsbeforethe termlimitbecause thoseeffectscannotlead to as manypotentialreelections.
the legislatorwill give to
Finally,theyreducethe consideration
adany effectsin futureyears because the reducedincumbency
vantagemakesthelegislatorless likelyto reachanyfutureterm.
All will exacerbatethe tendencyof legislatorsto be short-term
oriented.
This is of coursequite similarto the claim that corporate
managementneeds to be protectedfromcorporatetakeoversbeoriented.The
cause they make management"too" short-term
claimssufferfromthe same defect:it is not at all clear whywe
should reject whateverdiscountrate voters and shareholders
giveto futureclaimedbenefits.Theymayrationallyhave differentviewsaboutthe appropriatediscountrate to applyto the futuregenerally.Or theymayrationallyhave a lot less confidence
in claimsofbenefitsin futhan currentmanagement/legislators
about a future
tureyears.If currentvotersreallywereconfident
to vote fora legislator
benefit,theywouldhave no disincentive
whomade it possible.
In anyevent,thereare otherproblemswiththe claim.First,
Without
are somewhatconflicting.
the effectson short-termism
termlimits,a legislatorhas extraincentivesto favorshort-term
programsin his earlyterms.Such programshelp him win the
keyearlyreelectionsthat allowhimto attaina level ofseniority
thatforthe restofhis careergarnershimpoliticalinsulationby
causingcollectiveactionand entrybarrierproblems.Even proponentsofthetheoryconcedeas much.360
Second, the currentsystem'ssupposed benefitof lessened
legislativeimpatiencecomes at a great cost: incumbents'increasedpoliticalinsulation.36'
Whyshouldwe thinkthe costs of
greaterpoliticalinsulationare less than the benefitsof greater
legislatorsmay not discountthe
patience?Politically-insulated
Id at 491 ("[I]ncumbencyadvantages exacerbate the tendency to choose short-run
programs.").
361 Id at 492 (arguing that legislative impatience results precisely because term limits
increase electoral competitiveness).

1997]
193
futureby as much,but theymay also pursue goals completely

in the rate ofdiscontraryto thoseofthe electorate.A difference
countinga commongoal seems unlikelyto matteras muchas differenceoverwhatgoal to choose.
thisis not an argumentthattermlimitsdeMost important,
crease the accuracywith whichthe electorate'spreferencesare
recordedin elections.It is a claim that termlimitsincreasethat
accuracybut that those electoratepreferencesare normatively
inappropriate.A rationalelectoratecould well be expectedto reject such a claim. Whateverits othermerits,nothingin it suggeststermlimitsshouldbe consideredundemocratic.
CONCLUSION
Opponentsof term limits generallybegin with the strong

premise that termlimits are undemocratic.Indeed, even supThe SupremeCourtin US Term
portersoftenconcedeas much.362
Limits was no exception.This Articleaims to show this premise
is false. There are compellingreasons to believetermlimitshelp
the electorateregisterits politicalpreferencesmoreaccurately.
Termlimitsreducecollectiveactionpressuresto votefora senior
incumbentto gain a highershare of legislativeclout.And term
limitslowerentrybarriersthatkeep out challengers.Botheffects
would likelyreduce the ideologicaldivergencebetweenelectorates and theirrepresentatives.
There are, to be sure, some contraryarguments.As Part IV
effectsfromtermlimitsare possible.
shows,some antidemocratic
But no such negativeeffectis likelyto significantly
interfere
with
the electorate'sabilityto get accuraterepresentation.
I thus conclude that termlimits on balance are prodemocratic.
Granted,
this requiresa judgmentcall not everyonemay agree with.But
at a minimumwe need to ask ourselveswhere the burden of
proofshouldlie. It seems plain that democraticprinciplesrequire
withthe electorate's
puttingit againstthosewhowouldinterfere
democratically
expressedwish fortermlimits.It seems equally
plain that this burdenhas not been carried.But I would go further.The matterseems sufficiently
debatablethat it shouldnot
be leftto decisionby courtsapplyinga burdenof proofat all.
Rather,it shouldbe leftwherewe normallyleave such contestable mattersabouthowbest to organizeour society:to the political process.
Of course,the Courtrealizedthat,in the name ofprotecting
it was frustrating
the democraticpreferences
ofthose
democracy,
62
See Will, Restorationat 213 (cited in note 10).

194
The University
ofChicagoLaw Review
[64:83
whovotedfortermlimits.Whatit appearsnotto have realizedis

that it was also likelyinterfering
with the democraticpreferences offutureelectoratesvotingin regularelections.Instead,it
relegatedthemto a systemthat condemnssuch electoratesto
ideologicallydivergentrepresentativesthat less accuratelyreflectthe viewsofthe electoratethanwouldhave been truewith
termlimits.

1997]
195
APPENDIX
various pointsmade in Part II

I here derivemathematically
about the collectiveaction problemsaddressed by term limits.
Assume throughouta legislaturewith n districtseach of which
elects one legislator.The total benefitsof distributablegovernmentalactionare B. The totalcosts are C. Each districti's share
benefitsis Si, rangingfrom0 to n. The average
ofgovernmental
share equals 1 and the sumofthe shares equals n, thatis:
i=n
,Si =n.
i=1
ofthe tenureofdistricti's repreSi is an increasingfunction

sentative,and a decreasingfunctionofthe average tenureofthe
otherdistricts.0 denotesthe degreeoffitbetweenthe ideological
with0 rangingfrom0
viewsofthe districtand its representative,
to 1. The value of perfectideologicalfitis F. Variables will be
subscripted"tl"fortermlimits,and "nl"forno termlimits.
In the beginning,I also make the followingassumptions,
whichare relaxed later. Seniorityaffectsonlythe share of governmentalbenefitsand notthe abilityto advance ideologicalpositions.B and C are constants-thatis, I make no claim yetthat
termlimitsmightaffectthe total amountof distributablegovernmentalbenefitsenactedby the legislature.All districtsshare
benefitsequally,thus foreach district
the costs ofgovernmental
the costsare C/n. Finally,all districtsare assumed to attachthe
benefit.
same value to ideologicalfitand to a givengovernmental
is
The value a districtenjoysfroma representative
SiB/n - C/n + OF.
The districtwill thusvoteforits incumbentovera newcomerif:

- C/n+ Oincumbent)F
> S(newcomer)B/n
S(incumbent)B/n
C/n+ Onewcomer)F
thatis, if:
(S(incumbent)- S(newcomer))B/n > (*newcomer)- IOincumbent))F.-
Since the share of porkis an increasingfunctionof tenure,

S(incumbent) - S(newcomer) > 0, the incumbentcan be reelectedeven
though0pnewcomer)- O(incumbent)> 0, that is, even ifhe offersa poorer

196
ofChicagoLaw Review
The University
[64:83
- S(newcomer)
fitthanthenewcomer.
Moreover,
S(incumbent)
ideological
is largerthe longerthe incumbent'stenure,so that the magnitude of the permissibleideologicalgulfbetweenthe incumbent

thelonger
andanynewcomer,
0(newcomer)- Oincumbent),is also larger
the incumbenthas served.Longertenurethus makes Oincumbent)
lower,indicatinglargerideologicalslack. Also,the largerthe tobenefitsper district,
tal amountof distributablegovernmental
B/n, the more pronouncedan effecttenurehas on ideological
slack.
Wouldthesedistrictsbe betteroffwithtermlimits?Suppose
withthe same tenureand
firstthatall districtshave incumbents
benefits.Thus foreach disthusthe same shareofgovernmental
withouttermlimits,each district'sutiltrict,S = 1. Accordingly,
is (B - C)ln + (n4F. Withtermlimitscovityfromrepresentation
each district'sincumbentswill be oustedat
eringall districts,363
the same timeso thatnonesuffersa disadvantagein benefitdiswill be
tribution,and the district'sutilityfromrepresentation
(B - 0)/n+ wtl. Each districtthus benefitsfromtermlimitsas
longas (ti) > fl). Since termlimitsdecreasethe tenureofmembers (to zero at first),and p decreases with tenure,Otl) > 0nl).
Thus, everydistrictshouldfavortermlimitseven thoughevery
districtalso favorsretainingits incumbent.The degreeofenthusiasm fortermlimitsproposalsshouldbe higherthe longerthe
tenureofsittingincumbents,
and the higherthe overallamount
ofpork,sincethatdecreasesfnl). The enthusiasmfortermlimits
will also be higherthe morevalue votersput on ideologicalfit,
thatis, thehigherF.
Now supposewe relax the assumptionthat seniorityhas no
effect
on relativelegislativeclouton ideologicalissues. Let us call
the degreeoflegislativepowerp, whichgoes from0 to 1 and representsthe effectiveness
ofthe legislatorin gettinghis ideological viewsimplemented
intolegislation.To isolate the ideological
on the share ofgovernmental
benefits.
effect,
put aside theeffect
The ideologicalsatisfaction
a districtwill enjoyfroma representative will be ppF - p(1-4)F, which is the same as 2ppF - pF.
This is positiveas longas q > 0.5. Such an incumbentcan beat a

rival witha higherq even on purelyideologicalconsiderations.
Suppose a rival's0= 1, indicating100 percentcoincidencewith
the ideologicalviews of his electorate.Then the incumbentcan
win on ideologyas long as (20 - l)p(incumbent)F>
P(rival)F,
that is, as
3"If the term limits only cover some of the legislative districts,as with state-imposed
term limits on its federal legislators, collective action problems would remain. See Part
II.D.

1997]
long as (2
l)p(incumbent) > p(rival).
197
As long as the incumbent's
0 > 0.5, a legislatorwithmorep than a newcomerwill be able to

sustain a lower 0. The lowest sustainable incumbent0 will be
0.5 + (p(rivalV2 P(incumbent)).To illustrate,a legislatorwithfourtimes
as muchpoweron ideologicalissues as a rivalwithzero ideological slack couldmaintaina pas low as 0.625; withtwiceas much
powerone as low as 0.75; with50 percentmorepoweras low as
0.83.
What if the seniorityof representativesvaries among districts?Then the question is more complicated.To simplifythe
analysis, put again to one side the link between seniorityand
relativeclouton ideologicalissues. Assume districtsdiscountfuthe extentto whichtheygive pretureresults,withDt reflecting
sentvalue to a resultat futuretimet. Dt is a decreasingfunction
oft, and rangesfrom1 to 0. Then the value a districtputs on its
representation
withouttermlimitswill be:
t=0
Dt(St(nl)B /n -C /n + 4(nl)F).
t=o
Assumingthat with termlimitseach districtwould expect

S = 1, so that the value ofrepresentation
onlyaverageseniority,
withtermlimitsis:
t=00
Dt(B /n - C /n + 4(tl)F).
t=o
Thus, a districtwill favortermlimitsas longas:

t=o
0 < ? Dtf1-St(nl))B /n + Dt(O)(tl)

)-O(n1))F.
t=O
In otherwords,as longas
t=00
t=0o
nA*F > X(St(nl) - 1)B

t=o
t=O
whereAXt= 4t(tl)- ot(nl).364 We knowthe leftside ofthe equationis

positivebecause A > 0 forall values oft.365The rightside ofthe
3&NotethatI
benefitsequals
have assumedthe discountrate forfuturedistributable

198
The University
ofChicagoLaw Review
[64:83
equationshouldbe negativeforany districtwithlowerthan avsinceS - 1 < 0 forlowvalues oft and thereis no

erageseniority,
reasonto expectit to be greaterthan0 forhighvalues oft. Thus,
everydistrictwithbelow-averageseniorityshouldsupportterm
limits.Indeed,theyshouldsupporttermlimitseven if theyput
no value on ideologicalfit(F = 0) becausetheywouldstillbenefit
froma moreevendistribution
ofgovernmental
benefits.
But what if the district'sseniorityis higherthan average?
Then the rightside of the inequalityshould be positive,since
S - 1 > 0 forlow values oft and thereis no reasonto expectit to
be lowerthan0 forhighvalues oft. Still,thisdoes notmean that
we shouldexpecta districtwitha highlyseniorrepresentative
to
opposetermlimits,forrelativelyhighseniority
increasesbothSt
and AX at low values oft, thus increasingbothsides ofthe inequality.Districtswith highlyseniorrepresentatives
have both
largerincentivesto favortermlimitsbecause theybear larger
ideologicalcosts and largerincentivesto opposetermlimitsbecause theyenjoya greatershare ofgovernmental
benefitswithout limits.A districtwitha seniorrepresentative
will be more
likelyto oppose termlimitsthe lowerits F, meaningit cares
relativelylittleabout ideologicalslack. The level of B/n, indicatingthelevelofgovernmental
benefitsat stake,has conflicting
effects.
IncreasingB/n increasesthe magnitudeofthe seniority
effecton the rightside ofthe equation.But it also decreases0(ni)
byincreasingideologicalslack.The overalllevel ofgovernmental
benefitsthus has no clear effecton whetherdistrictswithrelativelyseniorrepresentatives
willsupportor opposetermlimits.
However,thereis no doubtthat all the districtsas a whole
are betteroffwithtermlimitseven if a feware not.That is because for all districtsthe relativecomparisonat any time is
whether
i=n
i=n
i=l
i=l
,Si(tl)B /n - C /n + 0-(ti
)F > 2Si(nl)B /n - C /n + 0'(nlF.
the discount rate for future ideological fit. If the voters discount future distributable
benefitsmore than futureideological fit,termlimits are more attractive.If they discount
ideologymore,termlimits are less attractive.
'Even if a district begins with a representative with tenure lower than the term
limit (forexample, a districtwith a four-yearincumbentconsideringa twelve-yearterm
limit),its 0 should still be lower with term limits because the decliningseniorityof other
representativesmakes the seniorityof its representativesomewhat less important.See
PartII.C.1.

199
1997]
i=n
Since XSi = n, thisis trueas longas AO> 0, whichwe

i=1
thisis
have alreadyseen to be true.The intuitionunderlying
ofgovernissues aboutthe distribution
that,overall districts,
mentalbenefitsbalanceout,but the ideologicalcostofunlimited
termsremains.
Now let us considerthe effectofseniorityon the overalllevel
activity,some of whichis pork and some nonof governmental
ofporkare Bpork,thetotalcostsCpork.By
pork.The totalbenefits
Bnonpork and Cwnpork denotethe
definition,
Bpork < Cpork. Likewise,
action.AssumeBpork
benefitsand costsofnonporkgovernmental
and Cporkare not constantbut are ratherincreasingfunctionsof
in the legislature.Because
the averagetenureofrepresentatives
each incrementalexpenditureon pork by definitioncosts more
than it benefits,this assumptionimpliesthat Cpork- Bporkis also
an increasingfunctionofaverage legislativetenure.In contrast,
and Cnonporkare assumed to be constantsunaffected
by avBnonpork
The districtsas a whole are betteroff
erage legislativetenure.366
as longas
i=n
XSitl)(Bpork(tn) + Bnonpork)l n -(Cpork(tl) + Cnonpork)/ n + 4ti(tl)F >

i=l
i=n
) +rBnonpork
ESi(n1 )(BPork(n1
) /n- (Cpork(nl) + Cnonpork
) /n
+ ot(nlF
i=n
Since
Si
i=1
n, thisis the same as whether:
"If lowerlegislativetenuredoes lowerBnonporkand Cnonpork,thenthe effectwill be

negativeto the extentBnonpork > Cnonpork, and positiveto the extentCnonpork > Bnonpork. The
governmental
benefitsare sociallydeformerwillbe trueifmostnonporkdistributable
sirable,thelatterifmostare undesirable.ComparePart IV.B. Those whofindmostgovernmentalactivityundesirable(i.e., conservatives)are thus particularlylikelyto find
shouldbe
termlimitsdesirable.In anyevent,the degreeofchangeto Bnonporkand Cnonpork
smallerthan the degreeof changeto Bpork and Cporkforthe reasons describedin Part
II.B.3. Thus, even if the lessened governmentalactivityis on balance desirable,the
on porkwilllikelybalanceit out,especiallysinceall theporkis undesirable
greatereffect
and thenonporkis a mixedbag.

200
The University
ofChicagoLaw Review
[64:83
- Cpork(tl)
+ nfavg(tl)F>
Bpork(tl)
- Cpork(nl)
+ nf4avg(nl)F
Bpork(nl)
whereJ'vgequals the sum ofeach district's4- dividedbyn. That

was higherwithtermlimitsthanwithoutthemwas already
Oavg
sufficient
reasonforthepoliticaljurisdiction
as a wholeto be bet-
teroffwithtermlimits.Theadditional
factor
thatBpork-
CporkiS
also greater(i.e., less negative)with termlimitsadds another

reason.Similarly,if all districts'incumbents
have the same tenure,thenthe effecton porklevels makes termlimitseven more
beneficialforeach district.
Ifseniority
varies,the answeragain dependson whetherthe
Junior
districtshave relativelyjunior or seniorrepresentation.
districtswill benefiteven morefromtermlimitsbecause lower
overall levels of pork are distributionally
favorableto them.
Seniordistrictsare again morecomplicated.
Theywillfavorterm
limitsif:
t=0
nXD#(Bpork(ti) + Bnonpork)/l
t=O
t=0
n +
(Cpork(tl)+ Cnonpork)f
) >
otk(tl)F
n + Ot(nl)F).
)n/ -(Cpork(nl)+ Cnonpork)/
XD#(St(n1)(Bpork(nl) + Bnonpork
t=O
That is, if:

t=oo
XnA/F
t=O
+ Cpork(nl)-Cpork(tl) >
t=o0
ZSt(nl)Bpork(nfl - Bpork(tl)+ (St(nl) - 1)Bnonpork.
t=O
The leftside ofthe equationrepresentswhatseniordistricts

have to gain fromtermlimits,lowerideologicaldivergenceand
porkcosts.It is positive,bothbecauseAOis positiveas previously
discussedand because Cporkis higherwithouttermlimitsthan
withthem.The rightside ofthe equationrepresentswhatsenior
districtshave to lose fromtermlimits-theirpreviouslyhigher
share ofdistributable
benefits.It is also positivebecause St > 1
forlowvaluesoft and Bporkis higherwithout
termlimitsthan
If
with them. each additionalexpenditureon pork costs more

1997]
201
than St timesits benefits,thenCpork(nl)- Cpork(tl)shouldbe greater

- Bpork(tl)
for all values of t. In these circumthan St(nl)Bpork(nl)
stances,the effectofloweringporklevels will make termlimits
moreattractiveto seniordistricts.Even ifadditionalporkexpendituresare less thanSt timeshigher,the seniordistrictmay still
findthe effectoftermlimitson porklevels beneficialbecause of
the reducedideologicalslack valued at nA4F. And at higherlevels of t, the higherincrementalcosts of pork should make term
limits even more attractivedespite the district'sinitiallyhigh
seniority
because St will overtimeaverage 1.


Are Term Limits Undemocratic

Uploaded by

Document Information

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

Are Term Limits Undemocratic

Uploaded by

Copyright:

Available Formats

Are Term Limits Undemocratic?

Author(s): Einer Elhauge

This content downloaded from 153.90.89.164 on Mon, 7 Oct 2013 05:51:49 AM