PL14CH14-Hicken ARI 18 April 2011 9:31

ClientelismAllen Hicken

Department of Political Science, University of Michigan, Ann Arbor, Michigan 48109;email: ahicken@umich.edu

Annu. Rev. Polit. Sci. 2011. 14:289310

First published online as a Review in Advance onMarch 17, 2011

The Annual Review of Political Science is online atpolisci.annualreviews.org

This articles doi:10.1146/annurev.polisci.031908.220508

Copyright c 2011 by Annual Reviews.All rights reserved

1094-2939/11/0615-0289$20.00

Keywords

patronage, patron-client relationships, particularism, contingency,development

Abstract

Clientelism is characterized by the combination of particularistic tar-geting and contingency-based exchange. This method of contingentexchange thrives in both autocracies and democracies. It exists in alarge variety of cultural contexts. Confronted with economic develop-ment, clientelism fades away in some political contexts but adapts andsurvives in others. This article explores our understanding of the ori-gins and dynamics of clientelism, focusing on the relationships betweenclientelism and democracy and between clientelism and development.It then evaluates the connection between clientelism and a variety ofpolitical and economic outcomes, including democratic accountability,corruption, and public goods provision. It concludes by outlining someremaining empirical and theoretical challenges and highlighting recentinnovations in data collection and empirical methods.

289

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

Clientelism exists in all polities. The form it takes,its extent, and its political functions vary enor-mously, however, across time and place.

Nicholas van de Walle (2007, p. 50)

INTRODUCTION

Clientelism has proven highly adaptable todifferent political, economic, and cultural envi-ronments.Thismethod of contingent exchangethrives in both autocracies and democracies(and in everything in between); it exists in alarge variety of cultural contexts; and in the faceof economic development it often adapts andendures (contrary to the expectations of earlieranalyses). Clientelism also affects thingswe careabout. For some, it is a subverter of democracyand economic development and an indicatorof a suboptimal pattern of politics. For others,clientelism undermines the ability of citizensto hold elected ofcials individually and collec-tively accountable and induces them to keep thedictatorial and corrupt in power out of fear andnarrow self-interest. Clientelism also divertsscarce resources that might otherwise be usedto further economic development and gener-ates incentives for keeping constituents poorand dependent. It both reects and feeds highlevels of corruption and ultimately underminespublic trust in democratic institutions.

These are just a few of the claims made inthe vast and growing literature on clientelism.In the pages that follow, I review the logicof some of these arguments and assess theextent to which these claims have been veriedempirically. Before doing so, however, I mustdene the parameters of the phenomenon un-der study. The following section describes thelack of conceptual clarity and consensus in theeld, discusses the common elements of manydenitions, and attempts to situate the conceptof clientelism within a universe of relatedterms, e.g., particularism, pork, and vote-buying. I then discuss the various accounts ofthe origins of clientelism, focusing specicallyon the relationships between clientelism anddemocracy and between clientelism and devel-

opment. Why does clientelism emerge in somesettings but not others? How do we accountfor variation across time or across space in thereliance on clientelist practices? This is wherethe literature has made the most progress inrecent years, through a combination of carefultheorizing and creative approaches to evaluat-ing the empirical implications of these theories.I then turn to the question of the consequencesof clientelism. I conclude by outlining someof the challenges to empirically evaluatingtheoretical claims, and I highlight some ofthe recent innovations in data collection andempirical methods as they relate to clientelism.

WHAT IS CLIENTELISM?

If there is one theme that appears again andagain in Annual Review articles, it is the lack ofconceptual clarity and consensus that plaguesthe study of whatever topic is under review.Clientelism is no different. In fact, the termis often used in the literature colloquially withlittle attempt to dene it. The term serves asshorthand for systems, institutions, or individ-uals that are somehow less than ideal (e.g., clien-telist party system, clientelist political party, orclientelist politician). And yet these colloquialuses are grounded in a literature replete withattempts to dene and delimit the concept ofclientelism. There is no generally accepted def-inition, but many denitions highlight the fol-lowing as key elements of clientelist relation-ships: dyadic relationships, contingency, hierar-chy, and iteration. Below, I discuss each of theseelements in turn, highlighting shifts in the liter-ature over the relative weight of each element.

Dyadic Relationships

For early scholars of clientelism, the socialrelationship between patron and client wasparamount. The emphasis was on direct, face-to-face interactions and transactions betweenthe patron and client. For example, Scott (1972,p. 92) speaks of clientelism as an instrumentalfriendship, while Lande (1977, p. xx) refers todyadic alliances. Clients, it was assumed, had

290 Hicken

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

a close personal relationship with the patron(Mainwaring 1999). Recent work has compli-cated this picture considerably. Although thereis still an acknowledgment of the importanceof personal, face-to-face relationships, the em-phasis has shifted toward discussions of brokersand networks. Many clients may have little orno direct contact with their patron, but insteadthe two are connected through a chain of bro-ker relationships (Weingrod 1968, Kitschelt &Wilkinson 2007a, Stokes 2007a, Muno 2010).Some of these clientelist machines can be quitecomplex, reaching from the summits of na-tional politics down to the municipal level(Kitschelt 2000, p. 849). Those at the top ofthe pyramid generate resources that are chan-neled down the pyramid, while votes and otherforms of fealty ow upward (Krishna 2007). Atthe local level, patrons typically rely on bro-kers who have standing in the community andare deeply imbedded in local networks. Thesemay be local government ofcials, landowners,respected business people, or other local nota-bles (e.g.,Moerman 1969, Curtis 1971, Krishna2007, Scheiner 2007, Wilkinson 2007).

Nevertheless, as long and complex as thisbroker network may become, most scholarsstress that at the core are personal, dyadic re-lationships between individualse.g., betweenthe patron and high-level brokers, betweenhigh- and lower-level brokers, and ultimatelybetween brokers and individual clients. Thereare, however, a few scholars who argue thatdyadic relationships are not a necessary ele-ment of clientelism. For example, Kitschelt &Wilkinson (2007a) argue that, under certainconditions, club goods (benets directed atgroups of individuals, which can be withheldfrom other groups but not withheld fromindividuals within the group) can be a form ofclientelist exchange.1 Such collective clien-telism is distinguished from pork or program-matic redistributive benets by the contingent

1Specically, club goods are clientelist if the exchange sat-ises the following conditions: predictability, elasticity, andcompetition (Kitschelt &Wilkinson 2007a, pp. 1415). Clubgoods are discussed further below.

nature of the exchange (Stokes 2007a), and itis to that element of clientelism I turn next.

Contingency

The element that every denition of clientelismhas in common is the contingent or reciprocalnature of the patron-client exchange. Thedelivery of a good or service on the part ofboth the patron and client is in direct responseto a delivery of a reciprocal benet by theother party, or the credible promise of such abenet (Piattoni 2001a, Robinson & Verdier2003, Roniger 2004). As mentioned above, itis the quid pro quo nature of the exchange thatdistinguishes it from other forms of politicalparticularism. Other forms of particularismtarget specic groups, but in the case of clien-telism, that targeting always comes with stringsattached. Politicians supply benets only toindividuals or groups that support or promiseto support the politician. Likewise, the clientsupports only that politician who delivers, orpromises to deliver, a valued benet in returnfor the clients electoral support.

The nature of the goods and services thatare being exchanged is generally left unspeciedin denitions of clientelism. The list of mate-rial goods that can be offered voters is limitedonly by politicians and voters imaginations,and can range from cash to cookware to corru-gated metal.2 Nonmaterial benets can includejobs (Robinson & Verdier 2003, van de Walle2007), access to public services such as housing,education, or healthcare (Hicken & Simmons2008), protection ( Jamal 2007), or interventionwith the bureaucracy (Chubb 1982).

Finally, although the exchangemust be con-tingent in order to be classed as clientelism, itneed not be immediate. There is generally a lagbetween when the voter delivers his vote andwhen the politician delivers the promised ben-et, or vice versa, and thus the ability of each

2Schedler (2002) includes an extensive list of material goodsoffered to Mexican voters. See Brusco et al. (2004) for a sim-ilar list of goods offered to Argentinean voters.

www.annualreviews.org Clientelism 291

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

party to monitor and sanction the other is cru-cial (Kitschelt & Wilkinson 2007a).

Hierarchy

Hierarchy has been a crucial element in manydenitions of political clientelism, particularlyin sociological approaches. Scott (1972, p. 92),for example, denes clientelism as a relation-ship in which an individual of higher socio-economic status (patron) uses his own inu-ence and resources to provide protection orbenets, or both, for a person of lower status(client) who, for his part, reciprocates by offer-ing generous support and assistance, includingpersonal service, to the patron. Lande (1977,p. xx) likewise emphasizes the importance of hi-erarchy, dening patron-client relationships asa vertical dyadic alliance; i.e., an alliance be-tween two persons of unequal status, power orresources each of whomnds it useful to have asan ally someone superior or inferior to himself.(See also Eisenstadt & Roniger 1984, p. 48.)Yet among these approaches there is a recog-nition that even highly constrained clients cansometimes turn the terms of the relationship totheir advantage (Scott 1977a, Silverman 1977,Waterbury 1977).

As mentioned above, much of the recentwork in the eld has focused on the contin-gent nature of clientelism and does not typi-cally include an explicit reference to hierarchyor asymmetry in the formal denition.3 How-ever, nearly all such studies take for grantedthat the relationship between patron and clientis asymmetricwith the patron possessing in-formation, resources, or prestige that the clientlacks. Kitschelt & Wilkinson (2007a, p. 7) arean exception to this general pattern. Theyconceive the patron-client relationship as a

3Kitschelt & Wilkinson (2007a, p. 2) dene clientelism asthe direct exchange of a citizens vote in return for directpayments or continuing access to employment, goods, andservices, while Stokes (2007a, p. 605) offers the followingdenition: the proffering of material goods in return forelectoral support, where the criterion of distribution that thepatron uses is simply: did you (will you) support me?

principal-agent relationship, with the client asthe principal and the patron, or politician, asagent. This conceptualization certainly corre-spondswith standardmodels of democratic del-egation, where voters select representatives andthen endeavor to hold them accountable. How-ever, at rst blush this seems at odds with themore common view of how the patron-clientrelationship functions. The client-as-principalmodel seems to imply that the client has theultimate power advantage in the relationship,whereas most other studies place the patronrmly above the client. Stokes (2005), for exam-ple, argues that in clientelist systems the stan-dard democratic accountability framework isturned on its head. Rather than voters holdingparties andpoliticians accountable for their per-formance, it is parties and politicians that holdvoters accountable for their vote. When juxta-posed with this reality, Kitschelt &Wilkinsonsclient-as-principal framework serves to under-score the way in which patron-client relation-ships can pervert accountability (Stokes 2005)and ultimately distort democracy.4

Iteration

One of the key factors that differentiates clien-telist exchange, from, say, a bribe demandedby a policeman, or pencils handed out bypolitical candidates to potential voters at thecounty fair, is the ongoing nature of the rela-tionship. The former are (hopefully) one-offinteractions, with neither party having a strongexpectation of interacting in the future. Bycontrast, clientelism is at its core an iteratedinteraction, with each side anticipating futureinteractions as they make decisions about theirbehavior today. This has a couple of importantimplications. Iteration provides a mechanismfor overcoming one of the key challenges of

4When Kitschelt & Wilkinson turn to the dynamics of clien-telism, they reverse the patron-client relationship. Politiciansbecome the effective principal, concerned with shirking onthe part of voters, and thus need to nd ways to monitor vot-ers and ensure compliance (Kitschelt & Wilkinson 2007a,pp. 1417).

292 Hicken

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

clientelismthe lack of exchange simultaneity.Whether it is a bag of rice in exchange fora promised vote, or a vote in exchange fora promised job, clientelist exchange usuallyrequires one of the parties to trust that theother will deliver on their promises. Suchpromises are more credible when there is anexpectation that the relationship will be anongoing one. Repeated interaction reinforcessocial norms of reciprocity, which somescholars identify as a feature of clientelism(e.g., Eisenstadt & Roniger 1984). Iterationalso provides both the patron and client withinformation about the reliability of the otherand gives each party the opportunity to punishthe other for defecting. Ongoing relationshipsallow politicians, for example, to leverage socialnetworks to monitor the behavior of clients(e.g., is the client attending rallies? Does shecome to the polls? What bumper sticker isdisplayed in her home window?) and come toan informed decision about whether clientsare following through (or are likely to followthrough) with their promised support.

Iteration is directly related to two of thethree components of clientelism discussedby Kitschelt & Wilkinson (2007a, p. 9). Inorder for clientelist exchange to emerge as anequilibrium, they argue, there must be bothpredictability and monitoring. Iteration facili-tates both. The problem that all politicians facewhere the ballot is truly secret is uncertaintyabout whether voters actually vote the way theysay they will. Unable to actually observe howvotes are cast, politicians must develop reason-able predictions about voters behavior in orderfor clientelism to work. To be specic, theymust be able to make two kinds of predictions.First, as discussed above, politicians need tobe condent that voters who receive a benetwill deliver the expected vote. Second, politi-cians must develop reliable predictions abouthow a targeted clientelist benet will affectvoter behaviorwhat Kitschelt & Wilkinson(2007a, p. 13) term vote choice elasticity.How responsive will a potential voter be to anoffer of a clientelist benetor, put differently,

how big an offer does a politician need to makein order to bring a potential voter over to hiscamp? Repeated interactions over time allowpoliticians to observe which voters keep theirpromises and which voters can be swayed, andto calibrate the size of an offer needed to swaythose voters. Ultimately, then, iteration enablesboth voter and politician to make reasonablepredictions about the others behavior, andthe combination of iterative exchange andsocial networks enables politicians and politicalparties to overcome the strictures imposed bythe secret ballot (Stokes 2007a, p. 611).5

Volition?

The concept of volition as a possible element ofclientelism is strongly debated in the literature(for the positive case, see Eisenstadt & Roniger1984). At its core, this is a question about thenature of the glue that holds clientelist relation-ships together: power/force, needs/demands,or voluntary obligations (Muno 2010). Likeany relationship, clientelism contains certaincosts of ending the relationship. The questionis whether these costs are so prohibitively highas to deprive voluntary of any recognizablemeaning. If we approach patron-client rela-tionships as a mutually benecial contingentexchange, as rational choice approaches typi-cally do, then speaking of volition makes sense.Successful clientelist exchange is a mutuallyreinforcing equilibrium, with each side free toexit if they become dissatised with the natureof the relationship.

Other approaches focus on power asymme-tries and argue that in many cases these are sosevere that it is not useful to think of clientelismas voluntaryfor example, where the threat ofviolence or legal ramications are attached tononcompliance on the part of the client (e.g.,

5As a corollary, where voters and politicians trust each otherto deliver on their promises, this trust raises the costs (de-creases the vote choice elasticity) for potential rivals lookingto sway voters to their side (Kitschelt & Wilkinson 2007a,p. 13).

www.annualreviews.org Clientelism 293

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

slavery or bonded servitude).6 However, evenshort of the threat of violence, the nature ofthe relationship may be such that clients aredeprived of any meaningful opportunity to ex-ercise exit, or even voice (Hirschman 1970).Patrons possess a variety of tools to enforceclients compliance, from social ostracism towithholding of material benets, but a client,acting alone, cannot generally force a patronto honor his commitments. Collective actionby multiple clients is required to punish apatron, but the way clientelist systems are or-ganized makes such collective action extremelydifcult (Lyne 2007). In fact, in equilibrium, thedifculty of taking votes away from a clientelistincumbent is so severe as to sometimes deterany credible challengers fromentering, thus de-priving voters of meaningful exit options (Lyne2007). Hence, the perverse accountability re-ferred to earlier (Stokes 2005).

Given the observational equivalence prob-lems inherent in compliance issues generally, itcan be difcult to ascertain the degree of vol-untary compliance on the part of clients. Onthe one hand, one can argue that voters haveno meaningful agency to exercise when there isonly one credible choice. On the other hand, apatrons dominance may arise because that in-dividual or party is truly the best, most credibleprovider of resources the clients value. Like-wise, the rise of a credible challengermay signalthat voters nowhave a true choice, or itmay alsoindicate that voters evaluations of the patronhave changed such that they are now willing toentertain competing offers for their loyalty.

Clientelism and OtherTypes of Exchange

With the key elements of clientelism in hand,we can now turn to differentiating clientelist ex-change from other forms of political exchange.It is not the distributive or targeted nature of

6Some scholars reject such relationships as examples of clien-telism (e.g., Muno 2010, Piattoni 2001b), but others chooseto include them (Stokes 2007a).

clientelism that sets it apart. Rather, it is thecriterion by which targeting decisions are madethat distinguishes clientelism from other formsof distributive, redistributive, or particularisticpolitics (see Stokes 2007a, p. 605; Grzymala-Busse 2008). Parties may craft packages ofprogrammatic policies designed to targetspecic groups of voters. Candidates or partiesmay appeal to ethnic, religious, regional, orpartisan identities. Politicians may target porkto their home constituency. Populist politiciansmay promise more resources to previouslymarginalized segments of society. All of thesetypes of appeals target certain groups overothers, and all are carried out with electoralconsiderations in mindi.e., politicians expectthat the targeted groups will be more likelyto support them in the next election. What isunique about clientelist exchange is that thechief criterion for receiving the targeted benetis political support, typically voting. This isnot true of other strategies, in which the chiefcriterion is membership in the targeted con-stituency. To identify her target, the politicianin a nonclientelist environment asks, in effect,questions like: Do you live in my district? Areyou a member of my ethnic or religious group?Are you a farmer? Are you poor? Not: Didyou vote for me (see Chandra 2004). Certaingroups receive targeted benets, to be sure,but no member within the targeted group canbe excluded from the benet on the basis of alack of support for the politician or party.

It is not a surprise that the latter descriptionsounds like the denition of a club good. Manyof the potential benets politicians can providefall into this category. But to reiterate, it is notthe nature of the benet being offered that de-termines whether an exchange is clientelistitis the terms on which it is offered. The deliveryof a purely private good by a politician to anindividual would not be considered clientelismif it came without electoral strings attached.Likewise, whether or not club goods areconsidered clientelism depends on the criteriafor distributing those goods. Programmaticclub goods are disbursed according to well-dened rules and without regard to partisan

294 Hicken

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

characteristics or voting history. The discre-tion given to politicians over the distributionof these benets is minimal. (Unemploymentbenets in the United States are a classic exam-ple of a programmatic club good.) By contrast,clientelist club goods comewith substantial dis-cretion by politicians over disbursement, anddelivery of benets is contingent on the groupspast or promised political support (Kitschelt &Wilkinson 2007a).

A couple of examples from SoutheastAsia help to illustrate the distinction betweenclientelism and other types of targeted benets.In Thailand, part of Thaksin Shinawatraselectoral strategy was promising a basket ofclub goods targeted to poor, rural votersaninnovative strategy in the Thai context (Hicken2009). Policies such as the 30-baht healthcarescheme and a debt moratorium for farmerstargeted certain groups, with the costs ofthe programs borne by other groups, butthe benets of the policies were not directlycontingent on a vote for Thaksins Thai RakThai party. Qualied voters in constituenciesthat supported the opposition still had accessto the promised health cards and debt relief. Bycontrast, in the 1980s, the Singapore govern-ment announced a change to its vote-countingsystemvotes would now be counted andreported at the ward level, which in Singaporeroughly equates to an apartment block. As thevast majority of Singapore citizens live in publichousing estates, this meant that the govern-ment would possess fairly detailed data aboutthe distribution of its support. The ruling PAPparty wasted no time in explicitly tying housingservices to support for the PAP. Apartmentcomplexes that supported the opposition couldexpect to be last on the list for upgrades andimprovements (Tremewan 1994).

How is clientelism distinct from the relatedconcepts of patronage and vote buying? Formany authors, patronage and clientelism arelargely synonymous and are used interchange-ably (seeKitschelt&Wilkinson 2007c; Piattoni2001b, p. 6). However, in a subset of the liter-ature there is an important distinction betweenpatronage and clientelism. Some dene patron-

age narrowly as the exchange of public sectorjobs for political support (what was known inthe nineteenth-century American context asthe spoils system), whereas clientelism includesnot only jobs but also other state resources(goods, services, decisions, etc.) (Piattoni2001b).7 Others tie the term patronage to theuse of resources and benets that ow frompublic ofce (Mainwaring 1999, Stokes 2007a,van de Walle 2007). Here the distinctionhinges on the position of the patron and thetypes of resources at his disposal. In patronage,the patron must be an ofce holder or at leasthave access to state resources. In clientelism,the patron may or may not be an ofce holder,and so may not be able to credibly promise todeliver public resources. Absent access to thoseresources, the patron must rely on alternativemeans of exchange (e.g., private resources,party resources, etc.).8 Different authors drawthis distinction in various ways, but a commonthread is that clientelism is a much broaderphenomenon than patronage, with patronagesimply one specic type of clientelist exchange.

Where does vote buying fall? The key toclassifying vote buying rests on our judgmentabout whether the exchange of resources for avote corresponds to the elements of clientelismoutlined above, particularly the elements ofcontingency and iteration.9 If the exchangeof resources for votes is part of an ongoingrelationship, and the criterion for awarding theresource is how the recipient casts (or promisesto cast) his vote, then we can comfortablycategorize vote buying as a particular typeof clientelism. If, however, vote buying is anexercise akin to an entry fee, which all serious

7Robinson & Verdier (2003, p. 2), by contrast, label theexchange of a public sector job for political support asclientelism.8Medina & Stokes (2007) draw a similar distinction betweenresources that the patron controls independent of electoraloutcomes (what they term economic monopolies) and re-sources that depend on the patron holding public ofce (po-litical monopolies). The latter would be patronage, and bothwould fall under the rubric of clientelism.9On vote buying see Schaffer (2007), Schaffer & Schedler(2007), and Hicken (2007).

www.annualreviews.org Clientelism 295

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

Contingent? National PublicGoods &

ProgrammaticPolicies

NY

Targeted Benefits?

Programmatic Redistributive andClub Goods,

Noncontingent Vote Buying,Pork

ClientelismIncl: patronage, clientelist votebuying, clientelist club goods

NY

Figure 1Clientelism versus other benets.

candidates must pay but which carries nospecic obligation for the recipients, then thecontingency element is missing and we wouldnot consider it clientelism (see Callahan &McCargo 1996). How do we judge whetherthere is contingency attached to a vote-buyingexchange? First, we can look at how targetedthe vote buying is. If the good or service is avail-able to all comers, e.g., tee shirts handed out tospectators along a parade route, then althoughthe benet may be intended to sway peoplesvote, it does not rise to the level of clientelism.Second, we can observe whether voters ac-cept vote-buying offers from more than onecandidate/party. Where this is common, itis reasonable to infer that the expectation ofreciprocity on the part of voters and candidatesis weak at best. Finally, we can observe theeffort candidates make tomonitor the targets oftheir vote-buying efforts. The more resourcesdevoted by the candidate to monitoringand compliance, the more likely it is that thevote-buying exchange is clientelist in character.

Although contingency is necessary in orderfor vote buying to be considered clientelism, Iwould argue that it is not sufcient. The rela-tionship between vote buyer and seller must bemore than a one-off transaction. There mustbe some recognition by both parties that therelationship is, or is likely to be, ongoing. For

example, brokersmay offer to buy voter identitycards, or stand outside polling places offeringto pay voters for depositing a premarked ballot(while returning with the blank ballot they re-ceive from the electionofcial).These exampleswould certainly qualify as vote buying undermost denitions, and are certainly contingenttransactions, but do not necessarily constituteclientelist exchange. They are essentiallyarms-length transactions with no expectationof future interactionakin to accepting anoffer of a new camera in exchange for agreeingto listen to a sales pitch about timeshares.

Figure 1 summarizes some of the con-ceptual distinctions discussed above. It isimportant to note that (a) the boundariesbetween these concepts can be blurry, and(b) in practice, politicians and especiallyparties may simultaneously pursue a mix ofstrategiespromoting a program of publicgoods, providing programmatic targetedbenets to particular groups, and deployingclientelist strategies to reach other potentialvoters (Kitschelt 2000, Magaloni et al. 2007).

CLIENTELISM ANDDEMOCRACY

The literature on clientelism can be brokendown into roughly three approaches (Roniger

296 Hicken

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

2004, Kitschelt & Wilkinson 2007a). The rstapproach concerned itself with describing whatit saw as the vestiges of a premodern formof political/social relations (e.g., Gellner &Waterbury 1977, Lande 1977, Schmidt et al.1977, Scott 1977a, Eisenstadt & Lemarchand1981). The focus was on conceptualizationand case studies, with most studies assum-ing that clientelism was bound to disappear ascountries modernized both economically anddemocratically.

Confronted with the reality of clientelismscontinuity, the second approach focusedon understanding how clientelist exchangechanged and adapted to shifting circumstances,describing the ubiquity of clientelism acrosshistorical and political settings, and analyzingclientelism as an explicit political strategy andone type of political/social exchange (Chubb1982, Eisenstadt & Roniger 1984, Kettering1986, Reynolds 1988, Willerton 1992, Roniger2004). We see the shift described above froman emphasis on dyadic relationships to complexclientelist networks embedded in or attachedto various political institutions (i.e., politicalparties, unions, and bureaucracies) (Weingrod1968, Silverman 1977, Tarrow 1977, Clapham1982, Mavrogordatos 1983). This second re-search approach also stands out for its explicitemphasis on a comparative approach in additionto single case studies and for its shift fromdevel-opmental/modernization explanations of clien-telism toward a focus on political institutionsas key independent variables (Shefter 1977).

Building on this rich foundation (and fol-lowing trends in the broader discipline), thecurrent phase of scholarship explicitly aims tobuild and test generalizable arguments aboutthe causes and consequences of clientelism. Aspart of this new emphasis on causal explana-tion, we have seen the development ofmore nu-anced theorizing and amove away from viewingclientelism through a dichotomous lens (i.e.,the presence or absence of clientelism) towardunderstandingwhy andhow the degree andpat-tern of clientelist exchange can vary across andwithin states. Of particular interest to this newgeneration of studies is the way in which clien-

telism coexists with, supports, and even thrivesunder (semi)democratic elections.

It is clear that clientelism can exist in avariety of political settingsfrom autocraticto democratic. Nonetheless, the nature of thepolitical environment inuences clientelistexchange in important ways. For example,the political functions that clientelism fulllscan vary across regime type (van de Walle2007). In democracies, clientelism is a toolfor building a loyal network of supporters. Inautocracies, clientelism also involves creatingsocioeconomic dependence on the regime(e.g., public housing in Singapore), and, as acorollary, political subservience (see Fox 1994,Wintrobe 2000, Grzymala-Busse 2008). Thenature of the regime can also affect the kinds ofbenets offered to voters or the nature of theexchange relationship; for example, clientelismin democratic settings tends to be moretransactional and less hierarchical than whatwe observe in autocratic settings. What drivesmany of the differences between autocratic anddemocratic clientelism, and indeed a key factorthat shapes the incentives to pursue clienteliststrategies within democracies, is the robustnessof political competition.

Competitiveness is dened and opera-tionalized in a variety of ways in theliteratureincluding party-system fragmenta-tion, electoral volatility, margin of victory, andparty turnover in government. Kitschelt &Wilkinson (2007a, p. 28) boil down the indi-cators of party-system competitiveness to thefollowing: when citizens and politicians havestrong incentives to try hard and win support-ers at the margins for one or the other partisancamp. For this to be the case, two conditionsmust be met: (a) elections are close betweenidentiable and distinct rival party blocs, and(b) the number of uncommitted (swing) votersis large enough to potentially tip the outcomeof the elections toward one party bloc or theother (Kitschelt & Wilkinson 2007a, p. 28).

For most scholars, robust competition pro-vides a check against clientelism, although themechanisms they identify vary. Geddes (1991),for example, argues that robust, balanced party

www.annualreviews.org Clientelism 297

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

competition induces politicians to profession-alize the bureaucracy and abandon clientelism(see also Grzymala-Busse 2007, Kitschelt &Wilkinson 2007a). Keefer argues that once onepolitical party develops an electoral advantagevia a shift to a programmatic strategy, the pres-ence of robust competition will compel otherparties to follow (Keefer 2006, Keefer &Vlaicu2008).

Work by others, however, demonstratesthat competition can be a double-edged sword.InLatinAmerica, for example, the high levels ofelectoral competition helped provide the impe-tus for labor-based parties to transform them-selves into clientelist political parties (Levitsky2007).10 In Africa, Lindberg&Morrison (2008)nd that the more competitive the elections ina given district, the higher the percentage ofcitizens who report voting based on clientelistreasoning. More generally, we should expectthat where clientelism is already common,politicians facing robust competition may actu-ally increase the use of such tactics. But whomdo they target?Does competition induce politi-cians to use clientelism to build support amongmarginal or swing voters, or to reward andperhaps encourage higher turnout among partyloyalists (core supporters)?11 The evidence forwhether parties target core or swing voterswith clientelist appeals is mixed. A variety ofstudies have found that parties target marginalor swing voters (see Schady 2000, Stokes 2005,Magaloni 2006, Keefer & Khemani 2009, andindirectly, Golden & Tiwari 2009). But othersnd that parties consistently direct benets totheir core supporterseven where turnout isnot an issue (e.g., under compulsory voting)and where they can reasonably expect thesupport of loyalists even without clientelist

10Nyblade & Reed (2008) and Golden & Tiwari (2009) sim-ilarly nd that robust electoral competition and the accom-panying uncertainty make politicians more likely to engagein criminal or corrupt behavior.11Dunning&Stokes (2010) actually distinguish between coresupporters (those who are a predictable part of a clientelistnetwork but are ideologically heterogeneous) and loyalists(those who will vote for the party on ideological grounds,regardless of whether they receive a clientelist benet).

inducements (see Hiskey 1999, Ansolabehere& Snyder 2002, Stokes 2005, Perez Yarahuan2006, Dunning & Stokes 2010). What is in-creasingly clear is that parties pursue both typesof strategies, although more work is needed tounderstand what determines the particular mixof targeting strategies a party pursues.

In an interesting new study along these lines,Dunning & Stokes (2010) empirically demon-strate that parties target both swing and corevoters. They argue that this mixed strategy is afunction of the internal organization of clien-telist parties. Vote brokers and party leadershave different strategies. Vote brokers prefer totarget loyalist supporters, the cheapest way tobuild a network. Party leaders prefer to targetswing voters in order to increase the probabilityof victory. The result is that party leaders tendto target swing districts while, within thosedistricts, brokers target more loyal voters thanoptimal from the perspective of party leaders.

Regardless of whether the response tocompetition and electoral uncertainty is tryingto mobilize core supporters or trying to swaymarginal/swing voters to their side, clientelismcan be a useful tool for parties/politicians.However, the precise mix of programmatic andclientelist policies politicians end up offeringvoters depends on two factors: the need toappeal to marginal voters, and the level ofdevelopment. The more competition forcespoliticians to appeal to indifferent, marginalvoters, the more costly clientelist strategiesbecome, ceteris paribus. The payment needed tosway an indifferent voter to ones side is likelyto be higher than what is needed to turn outloyalists and those who lean toward the party,and the risk that indifferent voters will fail touphold their end of the clientelist bargain isalso much greater. Thus, because the return onclientelist investment is lower when indifferentvoters need to be targeted, the incentive to pur-sue alternative mobilization strategies shouldincrease with the need to appeal to indifferentvoters (see Kitschelt & Wilkinson 2007a).

Economic development also plays a role byinteracting with competitiveness to heightenthe incentive to switch to programmatic

298 Hicken

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

appeals. As discussed in the next section, athigh levels of development the price of sway-ing voters of all stripes with clientelist goodsincreases, as does the value voters place onprogrammatic goods (Kitschelt & Wilkinson2007a, Lyne 2007; see also Geddes 1991).

CLIENTELISM ANDDEVELOPMENT

Economic Development

Perhaps the most common association drawnin the clientelism literature is between the levelof economic development and the prevalenceof clientelism.The literature reports this asso-ciation both cross-nationally, noting that clien-telism appears to be more prevalent in develop-ing countries, and within nations, noting thatpoorer voters appear to be more susceptible toclientelist offers than richer voters (e.g., Bruscoet al. 2004). We can trace the scholarly roots ofthis association to the modernization and de-velopmentalist schools from the mid-twentiethcentury, but current theorizing views therelationship between economic developmentand clientelism as probabilistic rather than de-terministic. Although the empirical associationbetween poverty/development and clientelismis fairly robust (see Wantchekon 2003, Keefer2006, Kitschelt & Wilkinson 2007a, Remmer2007, Bustikova & Corduneanu-Huci 2009;for opposing views see Schaffer 2004, Speck &Abramo2001), there are a variety of views aboutthe causal mechanisms at work. We can dividethese into two groups: rst, those that em-phasize the effect of development on potentialclients assessment of the value of a clientelistoffer versus other alternatives, and second,those that emphasize the effect of developmenton the capacity of patrons to provide clientelistbenets.12

Development might affect voters assess-ment of the relative value of clientelist benets

12Kitschelt & Wilkinson (2007a, pp. 2428) term these, re-spectively, demand-side and supply-side factors.

in a number of ways. First, as incomes rise, themarginal utility to a recipient of a given mate-rial benet decreases. Because of this diminish-ing marginal utility of income, candidates canget more bang for their clientelist buck by tar-geting the poor (see Dixit & Londregan 1996,Calvo & Murillo 2004). Second, as incomevaries, so do voter preferences over the relativevalue of different types of benets (Kitschelt &Wilkinson 2007a). With greater developmentcomes greater voter involvement in the broadernational economy. Thus, the impact of govern-ment policies on voters economic well-beingalso increases. Things like tax rates, exchangerates, the rate of growth, interest rates, wagerates, etc., become more important to votersas their incomes rise, ceteris paribus. Thus, theopportunity cost of supporting politicians whopromise only clientelist benets increases rela-tive to voting for candidates who (promise to)deliver broader, collective goods and policies.Third, the poor are more risk averse than theirmore wealthy counterparts, and thus the im-mediacy of clientelist exchange is appealing.Better a concrete material benet today thanthe promise of some policy benet tomorrow(see Scott 1977b, Kitschelt 2000, Kitschelt &Wilkinson 2007a, Desposato 2007).13

Development also directly shapes the capac-ity of potential patrons to provide clientelistbenets by affecting the costs and benets ofclientelist strategies. In brief, as incomes risethe cost of clientelism rises, while the benetsdecline, ceteris paribus, thusmaking other strate-gies more appealing. First, development leadsto increased voter mobilityboth physical andoccupational (Ramseyer & Rosenbluth 1993,Shugart & Nielson 1999, Bloom et al. 2001,Kitschelt & Wilkinson 2007a, Kitschelt et al.2010). Greater mobility undermines the tra-ditional patron-client networks through whichclientelist funds can be distributed and makes itdifcult to successfully develop, monitor, and

13Lyne (2007) argues that even wealthy voters should gener-ally prefer the immediate benets available via clientelism tooffers of larger-scale collective goods.

www.annualreviews.org Clientelism 299

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

maintain new clientelist networks. Thus, thecost of clientelism increases (and the incen-tives to employ a clientelist strategy decrease)as development and mobility increase. Second,given the decreasing marginal utility of incomeas incomes rise, the resources needed to sup-port clientelism grow as well. To maintain agiven level of support via clientelism, politi-cians must increase the benets they distribute.This is not necessarily problematic if the re-sources available for clientelism keep pace withthe growth in income. However, politicians arelikely to nd that their access to clientelist re-sources is increasingly constrained as incomesrise. Where clientelist policies entail a transferof resources from middle/upper class voters topoorer voters, one can expect to see resistanceto clientelist policies increase as the cost of suchtransfers increases (Robinson & Verdier 2003,Hicken 2007, Stokes 2007a,b). In light of this,candidates may nd it more appealing to curtailclientelism and switch to alternative strategiesthat produce a better return. In addition, as thecost of clientelism rises, the negative externali-ties for politicians also increase. These negativeexternalities can include reduced economic per-formance (Keefer 2007) and, given the generalshift away from clientelist preferences amongvoters as incomes rise, greater public discon-tent with clientelist practices and an increase inperceived corruption (Kitschelt & Wilkinson2007a, Singer 2009, Kitschelt et al. 2010).

There is still much to be done to parsethe mechanisms connecting economic devel-opment with clientelism. Rather than simplytesting the hypothesis that development is neg-atively related to clientelism, as most studieshave done, future work should try to derivetestable implications of the competing causalstories. For example, if the mechanism thatconnects income to clientelism is mobility, wewould expect the degree of clientelism to differacross populations that have similar incomesbut different levels of physical or employmentmobility. Or, if changes in voter attitudes aretruly an important part of the story, we shouldbe able to observe differences in how voters

across time and across income groups respondto survey questions about clientelism. Workby Brusco et al. (2007) is a promising move inthis direction. The researchers set out to testthe risk-aversion explanation using a surveyof Argentinean voters. The survey revealedno support for the risk-aversion hypothesis;although poor voters were indeed more riskaverse than wealthier voters, the level of riskaversion by itself had no independent effect onthe likelihood of a voter selling her vote.

State Development and Access toPublic Resources

One of the necessary conditions for clientelismto exist is that potential patrons must haveaccess to resources that can be tapped for clien-telist purposes. As Stokes (2007a) points out,these resources need not be public resources,but a supply of plentiful and easily accessiblepublic resources certainly makes clienteliststrategies more viable. This insight lies at theheart of Shefters (1977, 1994) work on the ori-gins of programmatic and clientelist party sys-tems. Shefter argues that the relative timing ofbureaucratization (i.e., the development of a ra-tional administrative system) and democratiza-tion (i.e., the introduction of universalmale suf-frage) determines whether party leaders opt forclientelist or programmatic voter-mobilizationstrategies.14 At the onset of democracy, the

14A related line of argument focuses on the size of thestate and its involvement in the economy as key indepen-dent variables. Specically, the propensity toward clien-telism increases as the state expands and comes to commandmore resources, and as the level of state intervention in theeconomy increases (see Kitschelt et al. 1999, Kitschelt &Wilkinson 2007a, ODwyer 2006). Conversely, as the sizeof the state and its level of involvement in the economyshrink (e.g., via liberalization and privatization), the opportu-nities and incentives for clientelism shrink as well (Kitschelt& Wilkinson 2007a, Scheiner 2007, Kitschelt 2007). How-ever, the empirical evidence is mixed. Levitskys (2007) workdemonstrates under some conditions liberalization in LatinAmerica actually induced some parties to adopt clientelistpolicies. In his review of the literature, Roniger (2004, p. 367)notes the ourishing of clientelism under liberalization andreduced state intervention, stating that clientelism proves to

300 Hicken

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

rst question potential party leaders must faceis whether they can count on access to stateresources. For those parties that have been shutout of power (externally mobilized parties inShefters terminology), patronage is simply nota viable option. Instead, such parties developprogrammatic appeals to mobilize voters andactivists.15 Where these parties subsequentlygain power, we see political competition pro-ceeding largely along programmatic grounds(Piattoni 2001b, p. 19). For those parties orga-nized by entrenched elites who hold positionswithin the prevailing regime (internally mobi-lizedparties), the keyquestion iswhether thoseparties can expect to have access to the resourcesnecessary for clientelism. Whether clientelismis an option for these parties depends on the ex-tent to which the bureaucracy is autonomous orpoliticized. Where there exists an administra-tive system with a high degree of bureaucraticautonomy, the ability of politicians to divertstate resources toward clientelist strategies willbe greatly circumscribed, and programmaticstrategies will thus be more appealing. If, how-ever, the bureaucracy is politicized, then theexploitation of the spoils of government forpartisan purposes is too strong a temptation toresist (Piattoni 2001b, p. 19; see also Anderson1988, Hutchcroft 2000, van de Walle 2003).

The evidence for Shefters theory is mixed.The contributors to Piattonis (2001a) excel-lent edited volume take Shefters theory as thestarting point and examine the development ofclientelist practices in several countries acrossEurope. They nd that although the theorydoes a good job of accounting for some of thevariation across cases (e.g., Sweden, Germany,Greece, Spain, and Italy), it leaves many impor-tant cases unexplained (e.g., England, France,the Netherlands) (Piattoni 2001a, p. 19). In thelatter cases, the incentives for clientelism were

be highly adaptive to changing market logics. . .and capital-istic considerations. . . .15For a similar logic as it relates to party strategies in estab-lished democracies, see Golden & Chang (2001), Samuels(1999), and Cox & Theis (1998).

present, but citizen preferences were such thatthe political parties opted for alternative strate-gies. More generally, Shefters argument can-not account for why transformations away fromclientelism do occur in some cases (as in theUnited Kingdom and United States).

Worth noting are two recent attempts totest elements of Shefters theory. Keefer (2006)echoes a common critique of the theory:Once aparty is in power, what is to keep it from politi-cizing the bureaucracy to turn it into a source ofpatronage? Indeed, such political interferencewith once autonomous bureaucracies is not un-common (Hicken 2001, Baxter et al. 2002).Further, contra Shefter, he argues that wherepoliticians are truly unable to politicize the bu-reaucracy, programmatic parties should be lesslikely to emergethe logic being that voterswill continue to assign credit for good pro-grammatic outcomes to the high-quality bu-reaucracy, inducing parties to switch to alterna-tive mobilization strategies for which they canclaim credit, such as clientelism (Keefer 2006,p. 6). Using a cross-national analysis of morethan 100 countries between 1975 and 2004, hends support for the argument that countrieswith high-quality bureaucracies are less likelyto develop programmatic parties.

In an interesting new study, Bustikova &Corduneanu-Huci (2009) use a new dataset onclientelism created by Kitschelt and his collab-orators (described below) to assess some of thedeterminants of clientelism. They nd that,for developing countries, the level of economicdevelopment is the strongest predictor of thelevel of clientelism. Beyond a certain level ofincome, however, the bureaucracys reputationfor delivering needed public goods, regardlessof which party is in power, is the best predictorof the degree of clientelism. This is consistentwith Shefters argument about the link betweenbureaucratic quality and the emergence of pro-grammatic parties, although the studysmannerof operationalizing bureaucratic quality raisessome questions about endogeneity and thedirection of the relationship. Specically, theauthors use infant mortality as a proxy for

www.annualreviews.org Clientelism 301

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

bureaucratic capacity. Given the strong ev-idence of a link between clientelism and anunderprovision of public goods (discussedbelow), one wonders whether the causal arrowsought to be reversed.

In short, the evidence for Shefters argumentis mixed. But what is clear from these and otherstudies is the importance of institutional lega-cies.16 Not surprisingly, the best predictor ofwhether a given country currently has program-matic parties is whether it has a historical legacyof such parties.17

Consequences

The starting point (stated or unstated) for mostof the literature is that clientelism is inefcientat best and downright dangerous at worst. Itmay be a necessary or at least an unavoid-able evil in many cases, but it is nonethelessregrettable. There is however, an alternativeschool of thought that emphasizes theoreticallysalutary aspects of clientelism. First, it points tothe redistributive and social welfare aspects ofclientelist exchange and argues that our stan-dard for comparison should be not only pro-grammatic political systems but also predatoryor exploitative systems. Yes, clientelism may beless than ideal, but if the alternative is a statethat provides even fewer benets to citizens, es-pecially the poor, then clientelism is not such abad bargain. Clientelism can be one importantmechanism for securing transactional benetsfrom the state. Second, there is also a line ofargument that stresses the benets of localism

16See Grzymala-Busse (2007) for how historical institu-tional legacies interacted with the robustness of democraticcompetition to shape party strategies in postcommunistdemocracies.17In addition to competition, economic development, andaccess to state resources, the literature identies several otherpossible determinants of clientelism. These include culturalnorms (such as norms of reciprocity) (see Lemarchand 1977,Auyero 2000, Putnam 1993), ethnicity (see Chandra 2004,2007;Kitschelt&Wilkinson2007a), andpolitical institutionssuch as regime type, electoral systems, or ballot design anddistribution (see Golden 2003, van de Walle 2003, Roniger2004, Lehoucq & Molina 2002, Brusco et al. 2004).

and specialization.18 Clientelism is a mech-anism for dividing up national public goodsand distributing them to local constituencies.There are costs in terms of lost efciencies andeconomies of scale, but in exchange we get out-comes that are better calibrated to local needsand conditionssome clients will receive cash,others foodstuffs, others local public goods,depending on local demands (Roniger 2004).

Despite the plausibility of some of thesearguments, the empirical reality has generallybeen otherwise. The consensus in the literatureis that clientelism has profound negative impli-cations for the way in which democracy func-tions, citizen attitudes about the quality of theirdemocracy, and the capacity of governments toproduce needed public policies. To begin with,clientelism has the potential to reverse the stan-dard accountability relationship that is centralto democratic theory. By obliging voters to giveup some of their political rights in exchange foraccess to distributive benets, clientelism limitsthe exercise of citizenship rights and ultimatelymay undermine democratic consolidation (Fox1994, p. 153). In clientelist systems, voters maylose the ability to effectively hold politicians andparties accountable for their behavior in ofce,and instead, voters become the ones held ac-countable for their actions by parties and politi-cians (Stokes 2005, Lyne 2007, Kitschelt et al.2010). It is not surprising, then, to see that vot-ers in clientelist systems are, on average, morecynical and disillusioned than their counter-parts in more programmatic systems (Kitscheltet al. 2007b, 2010).

Clientelism also hampers the develop-ment of the political institutions necessaryfor democratic development and consolidation(Graziano 1973). As discussed above, it warpssystems of representation and accountability,consolidates incumbency advantage, and canundermine thepromise of a secret ballot (Stokes2005, Lyne 2007). Clientelism is also associ-ated with the politicization of the bureaucracy

18See Hicken & Simmons 2008 for a summary of these ar-guments as they relate to particularism generally.

302 Hicken

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

and is an impediment to the development of asystem of administrative control and oversight(Shefter 1994,Golden 2003, Keefer 2006, Cruz& Keefer 2010).19 There is also some evidencethat the freedom of information is higher inprogrammatic party systems compared to theirclientelist counterparts (Keefer 2006).

Furthermore, clientelism affects the partysystem. Programmatic systems tend to havelower electoral volatility, lower party-systemfragmentation, and higher levels of party-system institutionalization (Kitschelt et al.2010). In addition, clientelism shapes the be-havior of political parties in power. For exam-ple, Desposato (2007) nds that clientelist par-ties behave differently in the legislature than dotheirmore programmatic counterparts. Speci-cally, parties in themore clientelist state legisla-tures in Brazil are less likely to rely on roll-callvoting and less cohesive when they are in theopposition.

In terms of governance and economic per-formance, clientelism is clearly linked to higherlevels of rent seeking, but this does not neces-sarily translate directly into poor economic per-formance. The chain of variables from clien-telism to economic growth is a long one, andwe know that clientelism, and associated rentseeking, may proceed hand in hand with stronggrowth as long as certain sectors of the econ-omy maintain high levels of efciency and arenot swamped by the costs and inefcienciesassociated with clientelist sectors (Hill 1996,Hutchcroft 1997, Khan & Jomo 2000, Hicken2001, Kitschelt 2010).

Still, clientelism tends to distort incen-tives in predictable ways and is thus associ-ated with certain patterns of governance. Anumber of studies nd a correlation betweenclientelism and the size of the public sec-tor, whether measured in jobs or wage bills(Gimpelson & Treisman 2002, Calvo &

19For suggestive evidence that clientelism may hamper thedevelopment of the rule of law, see Kitschelt et al. 2010.By way of contrast Keefer 2006 nds no support for a linkbetween the presence of programmatic parties and the ruleof law.

Murillo 2004, ODwyer 2006, Grzymala-Busse2008).20 Similarly, clientelism has been linkedto larger public decits and public sector in-efciencies. There is also a stark difference inthe types of goods supplied in clientelist ver-sus programmatic systems. In clientelist sys-tems, nontargeted, national public goods areundersupplied while there is an overprovisionof goods and services targeted to narrow con-stituencies. For example, programmatic sys-tems exhibit lower levels of targeted spending,higher primary school enrollment rates, andmake [please keep] more efcient and effectiveuse of public resources than do clientelist sys-tems (Keefer 2006, 2007; Hicken & Simmons2008).

Finally, numerous studies have found astrong link between clientelism and corruptionor perceptions of corruption. Theoretically,it is not hard to see why such an associationwould exist. As Singer (2009) points out in aninteresting new study, clientelism may drivecorruption through any of three separate causalpathways. First, clientelism or certain formsof clientelist exchange, e.g., vote buying, mayactually be illegal. Second, by underminingthe ability of citizens to hold public ofcialsaccountable, clientelism may foster a cultureof impunity and make it difcult to punishindividuals for corrupt behavior. Third, thedemand for resources to support clientelistexchange may increase the incentives ofpoliticians to raise funds through illicit means.Expectations/assumptions about the associa-tion between clientelism and corruption are sostrong that some propose using indicators ofcorruption as a proxy for the level of clientelism(Persson et al. 2003, Keefer 2007). Empirically,these expectations are borne out in severalrecent studies. For example, Singer (2009)nds that businessmen report higher levels ofcorruption in their interactions with govern-ment ofcials in clientelist systems than they do

20Calvo & Murillo (2004), however, argue that there is atradeoff between public sector wages and the number of pub-lic sector workers.

www.annualreviews.org Clientelism 303

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

in more programmatic systems (see also Keefer2007, Kitschelt 2007, Kitschelt et al. 2010).

CONCLUSION

This most recent wave of scholarly interest inclientelism is quite promising. Although con-ceptual and denitional questions are still de-bated, there has been a greater emphasis in re-cent years on developing and testing theoriesabout the causes and consequences of clien-telism.Traditionally, local or country-level casestudies have been the most common approachto the study of clientelism. The case-study ap-proach has produced works of scholarship thatarenotable for their empirical richness and theircreativity in devising ways to capture clien-telism (see Auyero 2000). But these types ofstudies are limited by the standard problem ofgeneralizability and, in some cases, a lack of at-tention to theory or causal analysis. In addi-tion, their very creativity can make replicabilitya challenge for many of these studies.

The recent work on clientelism continuesits attention to an in-depth understandingof specic cases, but it also addresses theselimits in a variety of ways. First, more andmore of the work is explicitly comparative,whether across countries (e.g., Piattoni 2001a,Kitschelt & Wilkinson 2007c), within coun-tries (e.g., Brusco et al. 2004, Calvo & Murillo2004, Desposato 2007), or across time (e.g.,Hagopian 2009). Second, there is a greateremphasis on developing theories, on derivingtestable hypotheses from those theories, andon evaluating those hypotheses empirically.This shift reects increased attention to themicrofoundations and causal mechanisms thatconnect clientelisms various independent anddependent variables. It is also an indication ofnew tools and methods that allow us to begin toget at these microfoundations and mechanismsin ways that simply were not possible in thepast. The increasing use of public opinionsurveys and eld experiments is particularlynotable in this regard. Also noteworthy is arecently completed effort by a team based atDuke University to gather detailed data on

parties propensity to use programmatic versusclientelist appeals (the International Demo-cratic Accountability Expert Survey, http://www.duke.edu/web/democracy/index.html).The data were collected via an expert surveyof over 1,400 political scientists, sociologists,and journalists. Respondents were asked toanswer more than 40 questions on the nature ofpartisan activities in their country of expertise,covering four subject areas: (a) how parties areorganized, (b) how politicians provide targetedbenets to voters, (c) how politicians can policethe contingency of clientelist exchange, and(d ) how politicians appeal on programmatic is-sues (Kitschelt 2010, pp. 1213). The results ofthis effort, along with several country-level sur-veys of voters, are providing researchers withnew and better quality indicators of clientelismthan have ever before been available.

The development ofmore precise indicatorsfromboth in-country and cross-country surveysis a welcome advance. One of the chief chal-lenges in the empirical study of clientelism isthe difculty in operationalizing and measur-ing the concept.21 Few if any smoking gunscan be found. As a result, we researchers areforced to rely on a stunning variety of (some-times crude) proxies. This is particularly, butnot exclusively, true of large-n empirical work,where the search for a valid measure of clien-telism is especially daunting. Table 1 presentsjust a few examples of themanyways clientelismhas been operationalized in the recent empiricalliterature.

Probably the most commonly used proxyfor clientelism is some measure of the sizeof the public sector, be it the public wagebill or the number of government personnel.Although this may sometimes be the best wecan do, the size of the public sector capturesonly one possible outcome of clientelistexchangepublic employment. As is clearfrom this article, the benets that patronscan offer to clients include much more than a

21See Kitschelt & Wilkinson (2007a, pp. 32329) for a sum-mary of these challenges.

304 Hicken

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

Table 1 Clientelism as a variable in empirical analysis

Operationalization/proxy Sample worksCharacteristics of voter-politician exchange relationship Brusco et al. 2004, 2007Congruence between the preferences of parties and of their votersa Kitschelt et al. 2010Congruence of appeals by party operativesa Kitschelt et al. 2010Demographic characteristics of voters Desposato 2007Level of clientelism in a given unit of analysis Desposato 2007Level of corruption Persson et al. 2003, Keefer 2007Parties efforts to supply clientelist benets Singer 2009, Kitschelt 2010Presence or absence of programmatic parties Keefer 2006, Cruz & Keefer 2010Salience of a left-right partisan dividea Kitschelt et al. 2010Size of public investment/construction budgets Keefer 2007Size of the public sector (wages and/or personnel) Calvo & Murillo 2004, Grzymala-Busse 2008, Remmer

2007, Robinson & Verdier 2003Spending on capital expenditures Remmer 2007, Chhibber & Nooruddin 2004Strength of an economic partisan dividea Kitschelt et al. 2010Voting rationale as reported by voters Lindberg & Morrison 2008

aA measure of programmatic party structuration.

government job. If public employment is partof a package of correlated clientelist benetsthat are regularly offered to voters, then it mayindeed be a useful proxy. If, however, publicemployment and other types of clientelistbenets are substitutes for one another, thenwe need to be cautious about inferring theimportance of clientelism from the wage bill ornumber of government bureaucrats. We alsoneed to be cautious about concluding that lowlevels of programmatic exchange necessarilyequate to high levels of clientelism. It is notuncommon for scholars to operationalizeclientelism as the degree to which parties are(un)programmatic. However, although there isclearly a strong negative relationship betweenclientelist and programmatic exchange, it isnot a one-to-one relationship. Clientelism is acommon alternative strategy to programmaticexchange, but is not the only alternative.22

There of course remains an abundance oftheoretical and empirical work to be done.

22Other alternatives include noncontingent pork-barrel pol-itics, noncontingent constituency service, or appeals that relyon charisma or shared ethnicity.

There is an increasing number of excellentcountry-level surveys, but comparable cross-national data about voter attitudes and behav-iors toward clientelism remain relatively scarce.There is also room for much more work onthe factors that determine the particular mixof strategies that politicians/parties choose toemploy and what voters they choose to target.We have some good ideas about the factors thatmake clientelism more or less likely, and someunderstanding about how parties mix clientelistappeals with other types of appeals, but there ismore to discover. For example, what factors af-fect the mix of swing and core voter targeting?What factors affect the choice over the strategicmix of clientelism and programmatic policies?In addition, within the broad category of clien-telist exchange, patrons can offer clients a vari-ety of benets.What affects the types of benetsthat voters demand, or that politicians chooseto supply? Do certain types of benets tend togo together?

Future research might further exploremechanisms linking poverty/development toclientelism. The literature suggests severalpossible causal mechanisms, but little has beendone to tease out their relative inuence. Other

www.annualreviews.org Clientelism 305

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

interesting work could be done on how clien-telism shapes party organization and how, con-versely, party organization may shape the typeof clientelist strategies a party pursues. In addi-tion, we are still lacking in our understandingof why some states are able to transition fromclientelism to other forms of exchange, while inothers clientelism adapts and endures. Finally,it is worth exploring more carefully the role ofpolitical institutions such as electoral systemsand regime type. At present, there is relativelylittle evidence that such institutions have an

independent effect on the level of clientelism(in part because, in the long run, institutionalchoice may be endogenous to preferencesabout clientelism), but it is quite possible thatinstitutions play an important modifying role.For example, do they shape the form or typeof clientelist exchange politicians choose topursue? Do the effects of clientelism on demo-cratic health and governance vary dependingon the type of institutions in place? Taking in-stitutionsmore seriously could yield interestingnew theoretical and empirical insights.

DISCLOSURE STATEMENT

The author is not aware of any afliations, memberships, funding, or nancial holdings that mightbe perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS

Thanks to Anna Grzymala-Busse for her helpful and challenging comments.

LITERATURE CITED

Anderson B. 1988. Cacique democracy in the Philippines: origins and dreams. New Left Rev. 169:333Ansolabehere S, Snyder JM Jr. 2002. Party Control of Government and the Distribution of Party Expenditures.

Typescript, Mass. Inst. Technol.Auyero J. 2000. Poor Peoples Politics: Peronist Survival Networks and the Legacy of Evita. Durham, NC: Duke

Univ. PressBaxter C, Malik YK, Kennedy CH, Oberst RC. 2002. Government and Politics in South Asia. Boulder, CO:

WestviewBloom DE, Craig PH, Malaney PN. 2001. Study of Rural Asia: Volume 4. The Quality of Life in Rural Asia.

Oxford, UK: Oxford Univ. PressBrusco V, Nazareno M, Stokes. 2004. Vote buying in Argentina. Latin Am. Res. Rev. 39(2):6688Brusco V, Nazareno M, Stokes. 2007. Poverty, risk, and clientelism. Presented at Annu. Meet. Am. Polit. Sci.

Assoc., Aug. 30Sep. 2, ChicagoBustikova L, Corduneanu-Huci C. 2009. The correlates of clientelism: economic development and historical state

capacity. Presented at Annu. Meet. Am. Polit. Sci. Assoc., Sep. 36, TorontoCallahan WA, McCargo D. 1996. Vote-buying in Thailands northeast: the July 1995 general election. Asian

Surv. 36:37692Calvo E, Murillo MV. 2004. Who delivers? Partisan clients in the Argentine electoral market. Am. J. Polit.

Sci. 48(4):74257Chandra K. 2004. Why Ethnic Parties Succeed: Patronage and Ethnic Head Counts in India. Cambridge, UK:

Cambridge Univ. PressChandra K. 2007. Counting heads: a theory of voter and elite behavior in patronage democracies. See Kitschelt

& Wilkinson 2007c, pp. 183239Chhibber P,Nooruddin I. 2004.Do party systems count?The number of parties and government performance

in the Indian states. Comp. Polit. Stud. 37(2):15287

306 Hicken

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

Chubb J. 1982. Patronage, Power, and Poverty in Southern Italy: A Tale of Two Cities. Cambridge,UK:CambridgeUniv. Press

Clapham C, ed. 1982. Private Patronage and Public Power. New York: St. MartinsCox GW, Thies MF. 1998. The cost of intraparty competition: the single, nontransferable vote and money

politics in Japan. Comp. Polit. Stud. 31(3):26791Cruz C, Keefer P. 2010. Programmatic political parties and public sector reform. Presented at Annu. Meet. Am.

Polit. Sci. Assoc., Sep. 25, Washington, DCCurtis G. 1971. Election Campaigning Japanese Style. New York: Columbia Univ. PressDesposato SW. 2007. How does vote buying shape the legislative arena? See Schaffer 2007, pp. 14479Dixit A, Londregan J. 1996. The determinants of success of special interests in redistributive politics. J. Polit.

58:113255Dunning T, Stokes S. 2010. How does the internal structure of political parties shape their distributive strategies?

Presented at workshop Political Parties in the Developing World, Princeton Univ., Apr. 30May 1,Princeton, NJ

Eisenstadt S, Lemarchand R, eds. 1981. Political Clientelism, Patronage and Development. Beverly Hills, CA:Sage

Eisenstadt S, Roniger L. 1984. Patrons, Clients and Friends. Cambridge, UK: Cambridge Univ. PressFox J. 1994. The difcult transition from clientelism to citizenship: lessons from Mexico. World Polit.

46( Jan.):15184Geddes B. 1991. A game theoretical model of reform in Latin American democracies. Am. Polit. Sci. Rev.

85:37192Gellner E, Waterbury J, eds. 1977. Patrons and Clients in Mediterranean Societies. London: DuckworthGimpelson V, Treisman D. 2002. Fiscal games and public employment. World Polit. 54:14583Golden MA. 2003. Electoral connections: the effects of the personal vote on political patronage, bureaucracy

and legislation in postwar Italy. Br. J. Polit. Sci. 33:189212GoldenMA,ChangECC. 2001.Competitive corruption: factional conict andpoliticalmalfeasance in postwar

Italian Christian democracy. World Polit. 53( July):588622Golden MA, Tiwari D. 2009. Criminality and malfeasance among national legislators in contemporary India.

Presented at Annu. Meet. Am. Polit. Sci. Assoc., Toronto, Sep. 36Graziano L. 1973. Patron-client relationships in Southern Italy. Eur. J. Polit. Res. 1(1):334Grzymala-Busse A. 2007. Rebuilding Leviathan: Party Competition and State Exploitation in Post-Communist

Democracies. New York: Cambridge Univ. PressGrzymala-Busse A. 2008. Beyond clientelism: incumbent state capture and state formation. Comp. Polit. Stud.

41(4/5):63873Hagopian F. 2009. From patronage to program: the emergence of party-oriented legislators in Brazil.

42(3):36091Hicken A. 2001. Parties, policy and patronage: governance and growth in Thailand. In Corruption: The Boom

and Bust of East Asia, ed. JEL Campos, pp. 16382. Quezon City, Philippines: Ateneo de Manila PressHicken A. 2007. How do rules and institutions encourage vote buying? See Schaffer 2007, pp. 6889Hicken A. 2009. Building Party Systems in Developing Democracies. Cambridge, UK: Cambridge Univ. PressHicken A, Simmons J. 2008. The personal vote and the efcacy of education spending. Am. J. Polit. Sci.

52(1):10924Hill H. 1996. Indonesias industrial policy and performance: orthodoxy vindicated. Econ. Dev. Cult. Change

45(1):14774Hirschman A. 1970. Exit, Voice, and Loyalty. Cambridge, MA: Harvard Univ. PressHiskey J. 1999. Does democracy matter? Electoral competition and local development in Mexico. PhD diss., Dep.

Polit. Sci., Univ. PittsburghHutchcroft PD. 1997. The politics of privilege: assessing the impact of rents, corruption, and clientelism on

Third World development. Polit. Stud. 45(3):63958Hutchcroft PD. 2000. Colonial masters, national politicos, and provincial lords: central authority and local

autonomy in the American Philippines 19001913. J. Asian Stud. 59(2):277306Jamal AA. 2007. Barriers to Democracy: The Other Side of Social Capital in Palestine and the ArabWorld. Princeton,

NJ: Princeton Univ. Press

www.annualreviews.org Clientelism 307

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

Keefer P. 2006. Programmatic parties: Where do they come from and do they matter? Presented at Annu. Meet.Am. Polit. Sci. Assoc., Aug. 31Sep. 3, Philadelpia

Keefer P. 2007. Clientelism, credibility, and the policy choices of young democracies. Am. J. Polit. Sci.51(4):80421

Keefer P, Khemani S. 2009. When do legislators pass on pork? The role of parties in determining legislatoreffort. Am. Polit. Sci. Rev. 103(1):99112

Keefer P, Vlaicu R. 2008. Democracy, credibility and clientelism. J. Law Econ. Organ. 24(2):371406Kettering S. 1986. Patrons, Clients and Brokers in Seventeenth Century France. Oxford, UK: Oxford Univ. PressKhan MH, Jomo KS, eds. 2000. Rents, Rent-seeking and Economic Development. Cambridge, UK: Cambridge

Univ. PressKitschelt H. 2000. Citizens, politicians, and party cartellization: political representation and state failure in

post-industrial democracies. Eur. J. Polit. Res. 37:14979Kitschelt H. 2007. The demise of clientelism in afuent capitalist democracies. See Kitschelt & Wilkinson

2007c, pp. 298321Kitschelt H. 2010. Citizen-politician linkage strategies in democracies and investments in party organization. Pre-

sented at Workshop on Political Parties in the Developing World, Princeton Univ., Apr. 30May 1,Princeton, NJ

Kitschelt H, Hawkins KA, Luna JP, Rosas G, Zechmeister EJ. 2010. Latin American Party Systems. Cambridge,UK: Cambridge Univ. Press

Kitschelt H, Mansfeldova Z, Markowski R, Toka G. 1999. Post-Communist Party Systems. Competition, Repre-sentation, and Inter-Party Collaboration. Cambridge, UK: Cambridge Univ. Press

Kitschelt H,Wilkinson SI, eds. 2007a. Citizen-politician linkages: an introduction. See Kitschelt &Wilkinson2007c, pp. 149

Kitschelt H, Wilkinson SI, eds. 2007b. A research agenda for the study of citizen-politician linkages anddemocratic accountability. See Kitschelt & Wilkinson 2007c, pp. 32243

Kitschelt H, Wilkinson SI, eds. 2007c. Patrons, Clients, and Policies: Patterns of Democratic Accountability andPolitical Competition. Cambridge, UK: Cambridge Univ. Press

Krishna A. 2007. Politics in the middle: mediating relationships between the citizens and the state in ruralnorth India. See Kitschelt & Wilkinson 2007c, pp. 14158

Lande C. 1977. The dyadic basis of clientelism. See Schmidt et al. 1977, pp. xiiixxxvLehoucq FE, Molina I. 2002. Stufng the Ballot Box: Fraud, Electoral Reform, and Democratization in Costa Rica.

Cambridge, UK: Cambridge Univ. PressLemarchand R. 1977. Political clientelism and ethnicity in tropical Africa: competing solidarities in nation-

building. See Schmidt et al. 1977, pp. 10023Levitsky S. 2007. From populism to clientelism? The transformation of labor-based party linkages in Latin

America. See Kitschelt & Wilkinson 2007c, pp. 33580Lindberg SI, Morrison MKC. 2008. Are African voters really ethnic or clientelistic? Survey evidence from

Ghana. Polit. Sci. Q. 123(1):95122Lyne M. 2007. Rethinking economics and institutions: the voters dilemma and democratic accountability.

See Kitschelt & Wilkinson 2007c, pp. 20626Magaloni B. 2006. Voting for Autocracy: Hegemonic Party Survival and its Demise in Mexico. Cambridge, UK:

Cambridge Univ. PressMagaloni B, Daz-Cayeros A, Estevez F. 2007. Clientelism and portfolio diversication: a model of electoral

investment with applications to Mexico. See Kitschelt & Wilkinson 2007c, pp. 381429Mainwaring S. 1999. Rethinking Party Systems in the ThirdWave of Democratization: The Case of Brazil. Stanford,

CA: Stanford Univ. PressMavrogordatos GTh. 1983. Still-Born Republic: Social Coalitions and Party Strategies, 19221936. Berkeley:

Univ. Calif. PressMedina LF, Stokes S. 2007. Monopoly and monitoring: an approach to political clientelism. See Kitschelt &

Wilkinson 2007c, pp. 15082Moerman M. 1969. A Thai village headman as a synaptic leader. J. Asian Stud. 28(3):53549MunoW.2010.Conceptualizing andmeasuring clientelism. Presented atworkshopNeopatrimonialism inVarious

World Regions, Aug. 23, GIGA, Hamburg

308 Hicken

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

:289

-310

. Dow

nloa

ded

from

ww

w.an

nual

revi

ews.o

rg A

cces

s pro

vide

d by

Uni

vers

idad

e Fe

dera

l do

Rio

de Ja

neiro

on

05/1

8/15

. For

per

sona

l use

onl

y.

PL14CH14-Hicken ARI 18 April 2011 9:31

Nyblade B, Reed SR. 2008. Who cheats? Who loots? Political competition and corruption in Japan, 19471993. Am. J. Polit. Sci. 54(4):92641

ODwyer C. 2006.Runaway State-Building: Patronage Politics and Democratic Development. Baltimore,MD: JohnHopkins Univ. Press

Perez YarahuanG. 2006. Policy Making and electoral politics: three essays on the political determinants of social welfarespending in Mexico, 19882003. PhD diss., Dep. Public Policy, Univ. Chicago

Persson T, Tabellini G, T