Self-Enforcing Clientelism

Chappell Lawson Kenneth F. GreeneAssociate Professor Associate Professor

Massachusetts Institute of Technology University of Texas at Austinclawson@mit.edu kgreene@austin.utexas.edu

Draft

April 30, 2012

Abstract

Recent research on clientelism focuses on deliberate exchanges between mercenary voters

and strategic political brokers. In this “instrumentalist” view, machine politics is only sustainable

where patrons can monitor voters’ actions – a situation that does not apply in many places

known for clientelism. In this paper, we build a different theory of clientelism around the norm of

reciprocity. If exchanges rely on clients’ feelings of obligation to return favors to their patrons,

then clientelism can be self-enforcing and persist despite ballot secrecy. To support this

argument, we draw on ethnographic reports, survey data, and experiments from a variety of

countries, as well as split-sample experiments embedded in two new surveys on Mexico

specifically designed to test our predictions. Our findings have implications for voting behavior,

party organization, and the types of public policies that may diminish clientelism.

1. Introduction

Political scientists have long been interested in relationships in which leaders exchange

selective benefits for political allegiance.1 In developing countries, scholars typically group these

relationships under the rubric of “clientelism”; followers are known as clients, leaders as patrons,

and intermediaries as brokers. In the United States, analysts have referred to these same three

groups as “constituents”, “bosses”, and “precinct captains”; the relationships among them are

known as “machine politics” (Ostrogorski 1910, Gosnell 1937, Allswang 1977, Cox and Kousser

1981, Erie 1988). Despite extensive research in both contexts, however, political scientists have

not produced a consensus on how such relationships are maintained.

Recent theory argues that voters only comply with political brokers’ wishes if they

believe that their vote choices are monitored and that they can thus be sanctioned if they fail to

support the machine. This instrumentalist approach focuses on deliberate exchanges between

voters who seek to extract tangible benefits and strategic political brokers who want to deliver

votes to the politicians they serve as cheaply as possible (Dal Bó 2007; Kitschelt and Wilkinson

2007b; Nichter 2008; Stokes 2005). Such arguments emphasize the degree to which careful

surveillance – or at least the credible threat of such surveillance (Chandra 2007) – ensures that

constituents follow-through on their part of the clientelist bargain.

Yet this instrumentalist approach has not fully resolved the problem of voter compliance,

especially where ballot secrecy is well established. When clientelist exchanges are

asynchronous (i.e. brokers distribute benefits before an election in hopes of generating support

1 In keeping with recent work (e.g., Kitschelt and Wilkinson 2007b, Stokes 2007), we consider

benefits to be “selective” if they are excludable goods that can be distributed in a discretionary

fashion. We conceive of “clientelism” as a relationship between patrons with access to

resources that they trade for political support, and “vote-buying" as one possible manifestation

of clientelism in which benefits are exchanged for votes. 2

on Election Day), clients may “defect” from the agreement after receiving their payoff (Stokes

2005). To diminish the incentive to renege, patrons not only must be able to monitor voters’

choices, either at the time they cast their ballots or after the fact, but they must be able to

sanction individual voters who defect. Without identifying individual defectors, brokers cannot

know which voters to sanction by withdrawing benefits in subsequent elections. Clientelism

would then break down for one of two reasons. If brokers pay supporters and opponents alike,

then the disincentive for voting against the machine disappears. In addition, clientelism may

prove unsustainably costly (Dal Bó 2007). If brokers inadvertently sanction supporters, they

would create spurned voters who would not support the machine in the future under any

standard punishment-path strategy.2 Sanctioning groups of voters would only hasten

clientelism’s demise because a machine that loses once in a constituency would withdraw

benefits from some supporters who, in response, may not renew their support for the machine

(see Levine and Pesendorfer 1995; also see Finan and Schecter 2009).3

Brokers can overcome these problems in some contexts. Some countries use partisan

ballots, effectively eliminating ballot secrecy or permit party operatives in polling places,

conveying the perception of monitoring even when voting is secret (see Kitschelt and Rozenas

2011). In other places, operatives use more clever tactics to violate the secret ballot and

overcome the problem of asynchronous exchange. For instance,in Mexico, brokers are rumored

to have asked voters to photograph their marked ballot using cell phones. In Italy, operatives

2 If voters use “grim trigger”, then they would cease to be clients forever after not receiving a

payoff in any given election cycle (Stokes 2005). If both brokers and voters use “tit-for-tat,” then

a similar outcome would occur.

3 The logical implication is that clientelism based on instrumental calculations alone should be

sustainable only where the machine has an unlimited budget (so that it can buy opponents) or

where it holds monopoly power.3

from the Christian Democratic Party reportedly distributed left shoes to its clients before

elections with the promise of delivering the right shoes if it won (Chubb 1982). In many settings,

however, parties and candidates have limited capacity to monitor voters’ choices, allowing

instrumentally motivated voters to “take the money and run” and ultimately undermining

machine politics.

In recent work, some instrumentalist scholars have recognized this problem and

attempted to grapple with it systematically (Piattoni 2001: 7, Kitschelt and Wilkinson 2007: 8,

Magaloni 2006, Kitschelt and Rozenas 2011). For instance, Stokes (2005, 2007) argues that the

monitoring problem is mitigated by local brokers who are deeply embedded in their communities

and can guess at how their clients vote. Unfortunately, this key claim has not been testes

systematically. However, research in experimental psychology shows that people are

surprisingly bad at detecting when others are lying, even when they are trained to do so (Grohol

1999, Kohnken 1987, Kraut and Poe 1980).4 In addition, clientelism persists in many places

where well-organized political machines do not operate.5 Thus, difficulties in monitoring and

sanctioning voters continue to pose a serious challenge to instrumentalist theories of

clientelism.

Given the theoretical and practical problems with enforcement of the clientelist bargain

through external monitoring, we propose that machine politics can be supported by a

psychological mechanism rooted in norms of reciprocity. Specifically, we argue that the receipt

of gifts, favors, services, or protection creates feelings of obligation among voters, who then

4 Using turnout at the polls or at campaign events does not necessarily improve brokers’

guesses about which individuals vote for the machine.

5 Examples include Nigeria and Benin (Van de Walle 2007: 64), Taiwan (Wang and Kurzman’s

2007: 233), Paraguay (Finan and Schecter 2009: 17), Sao Tomé e Principe (Vicente 2008), and

Peru (González Ocantos et al. 2011a; Seawright 2011).4

spontaneously support their political patrons. Under such circumstances, clientelism is self-

enforcing.

This norms-based approach to clientelism has roots in certain ethnographic studies from

anthropology, sociology, and political science that emphasize the embeddedness of

relationships between patrons and clients and the “moral economy” of exchanges between the

two (Boissevain 1966, Wolf 1969, Powell 1970, Scott 1972, Lemarchand and Legg 1972,

Johnson 1974, Scott 1976, Eisenstadt and Lemarchand 1981, Lomnitz 1982 , Chubb 1982,

Komito 1984, Roniger and Günes-Ayata 1994, Gay 1994, Fox 1994, Auyero 2001, Calvo and

Murillo 2004, Gay 2006, Krishna 2007, Schaffer and Schedler 2007). These largely descriptive

studies have routinely emphasized norms of reciprocity as the basis for clientelist exchanges,

but they lack a plausible psychological micro-foundation that is as conceptually well-grounded

as the instrumentalist underpinnings of principal-agent models of vote-buying.

We show that norms of reciprocity create a separate foundation for machine politics. The

sense of obligation that brokers create through the provision of selective benefits can help

politicians build clientelist networks even in the context of ballot secrecy. One implication is that

clientelism may be much more entrenched than existing analyses would lead us to believe and

will not necessarily disappear rapidly once ballot secrecy is enforced, as long as politicians have

access to discretionary resources. Rather, purging clientelism from political life may require a

normative component – specifically, that citizens reject on principle the exchange of votes for

selective benefits because they feel a greater obligation to vote their conscience, to obey the

law, and to support democratic institutions.

Our argument also has ramifications for party organization. Because machine politics

can persist in the absence of active monitoring, politicians may not need the large-scale, deeply

embedded organizations that would otherwise be required to monitor and sanction voters

(Gyrzmala-Busse 2005; Kitschelt and Wilkinson 2007b; Stokes 2005, 2007). Rather, they can

5

secure election by earning voters’ loyalty, somewhat akin to the way that politicians generate

gratitude through constituency service and pork-barrel politics.

The next section of this paper provides a systematic rationale for the “norms-of-

obligation” interpretation of clientelism. We caution that the instrumentalist and norms-based

approaches contrasted here are not synonyms for “rational choice” and “cultural” approaches.

There is nothing inherently cultural about discharging one’s debts through acts of political

support. The third section discusses observable implications of this approach and compares

them to those of instrumentalist models; this section draws on ethnographic studies, surveys,

and experiments, as well as on two new surveys from Mexico. The concluding section returns to

the implications of our findings for electoral behavior, party organization, and public policy.

2. Reciprocity and Obligation in Clientelist Exchange

Our approach is based on the notion that powerful instincts of reciprocity cause people

to feel indebted to those who provide them with benefits of more than token significance. When

the recipient cannot discharge this debt of obligation materially, patron-client relationships may

emerge. Of course, not all such debts generate clientelism. After detailing the psychological

foundation for reciprocity, we generate more specific hypotheses about the conditions under

which feelings of obligation promote clientelism.

Reciprocity is a fundamental element of human social interaction. Norms of reciprocity

are recognized at a very young age (Harris 1970, Dreman and Greenbaum 1973, Birch and

Billman 1986) and, although cultural differences can influence the expression of these norms

(Herrmann et al. 2008, Gächter and Herrmann 2008), the concept and practice of reciprocity are

essentially universal (Heinrich et al. 2005). This component of human nature appears to be

evolutionarily “hard-wired” (Hammerstein 2003, Gintis et al. 2003, Berg et al. 1995, Ornstein

1980, Trivers 1971, Simmel and Wolff 1950). Recent research in genetics and neuroscience

6

even identifies some of the specific alleles and biological pathways that underlie reciprocal

exchange (McCabe et al. 2001, Sanfey et al. 2003, de Quervain et al. 2004, Cesarini et al.

2008, Fowler et al. 2008).

Instincts of reciprocity can powerfully influence behavior (see Fehr and Fischbacher

2002, Elster 1989: 192-214; Dawes and Thaler 1988: 195). For instance, experimental research

shows that people often treat others fairly even when they face an incentive to do otherwise

(McCabe et al. 1996, McCabe et al. 2003, Berg et al. 1995). Several years ago, two leading

behavioral economists remarked that, “the obligation to reciprocate is so strong that we take it

for granted” (Carmichael and MacLeod 1997: 502). Scholars from various disciplines have now

documented the role of reciprocity in a wide range of social activities, but it has not yet been

used as the basis for understanding political clientelism.6

Instincts of reciprocity lead people to feel indebted to those who provide them with gifts,

services, favors, or protection (Mauss 1990, Sherry 1983). Among individuals of similar status,

such obligations are often discharged by providing a good of comparable worth to the original

“gift”.7 But not all recipients have the wherewithal to pay their patrons back in kind. Patron-client

relationships may emerge when recipients discharge their debt by according the giver greater

social status, esteem, or loyalty (Weinstein et al. 1969, Bienenstock and Bianchi 2004). This

exchange creates a status hierarchy among previous equals (Rao 2001, Aragon 1996, Bartlett

1980).8

In the political sphere, recipients typically reciprocate patronage through outward

6 One potential criticism of this literature is that reciprocity’s role has not been experimentally

tested for large-scale transactions. Fortunately, voting is a low-cost activity that falls well within

the scope of the gains at stake in the experiments on the effect of reciprocity.

7 Debts can also be discharged by maintaining a long-term relationship in which mutual

assistance is taken for granted as in friendship or marriage.7

manifestations of allegiance and shows of solidarity (Forster 1963; Scott 1972, Eisenstadt and

Roniger 1984) – such as voting for a particular candidate, attending rallies, volunteering on a

campaign, contributing their energy to projects sponsored by patrons, or joining protests when

urged to do so (Harik 2004: 81-110, 147-62, Cambanis 2006, Radnitz 2010). Status hierarchy

has long been the core of clientelism as conceptualized by scholars in anthropology and

sociology. As Lomnitz puts it (1988: 47):

Loyalty in unbalanced exchange relations is the basis for political support…The

patron provides security of employment, political protection, and dependability in

unexpected circumstances of need in exchange for loyalty, expressed through

personal commitment to the patron in labor, political support, and ideological

allegiance.

Political clientelism becomes established when it is clear to both sides that recipients

can never repay benefits in kind, turning one group into perpetual recipients of material goods,

services, favors, or protection and the other group into perpetual providers. As Scott

characterizes it, the “patron is in a position to supply unilaterally goods and services which the

potential client and his family need for their survival and well-being” (1972: 93).

A key question concerns the conditions under which feelings of obligation become

activated in service of clientelism, and thus why scholars observe variation in the extent of

machine politics across individuals, communities, and countries (even where politicians have

similar levels of access to discretionary resources). We underscore two important limitations on

the power of obligation to activate clientelist relations.

First, not all gifts automatically activate instincts of reciprocity. Token items (e.g., a pen

bearing a party logo) may be taken as a gesture of introduction or courtesy, rather than as the

“quid” in a quid pro quo. Likewise, items distributed to everyone who happens to be on a street

corner at a particular moment could be interpreted similarly as materials meant to attract

8

attention or to announce a candidate’s presence, rather than as part of an exchange

relationship. The same holds if the benefits distributed are interpreted by recipients as

repayments for past good behavior or as entitlements rather than gifts. Selective benefits

offered by political machines may also be insufficient to generate clientelist support if the

obligation they create is not deemed “worth” a vote (or some other demonstration of adhesion).

The “going rate” for a vote likely varies across voters with different levels of resources.

Second, obligations stemming from reciprocity, do not necessarily propel citizens to

become clients when all factors are taken into account (Klosko 1990). A person may feel an

obligation to her family to accept groceries from a political party in return for her vote, and this

exchange may in turn generate an obligation to vote for that party. However, that new obligation

may conflict with some existing obligation, such as the personal conviction that she should to

vote her conscience or (where selling one’s vote is illegal) the moral obligation to obey the law.

Thus, factors like the intensity of a voter’s attachment to competing parties, the degree to which

she has imbibed civic norms, her respect for the law, and support for representative institutions

could affect the extent and durability of clientelism. Just as instrumentally motivated clients do

not automatically support the machine when they have competing reasons to choose another

9

party,9 voters motivated by feelings of obligation will only become captured clients under specific

circumstances, as we detail below.

3. Observable Implications of the Reciprocity and Instrumentalist Approaches

Both instrumental and normative motivations for clientelism may be present within a

given system, and machine politics will presumably be most durable when both mechanisms

operate. In some cases, both approaches make similar predictions. For instance, both predict

that clientelism would deteriorate when politicians run out of discretionary resources to

distribute. However, the two approaches do yield a number of different predictions, presented in

Table 1, which can be used as the basis for more systematic comparison. This comparison

helps “locate” our norms-based approach in relation to the now-dominant literature on

clientelism and yields testable hypotheses.

[Table 1 about here]

The key difference between the two approaches concerns the importance of monitoring

voters’ actions. If instrumental models are correct, machine politics is limited to contexts in

which patrons can credibly threaten to reward or punish their followers for non-compliance (Dal

Bó 2007; Kitschelt and Wilkinson 2007b; Nichter 2008; Stokes 2005). By contrast, clientelism

based on reciprocity may persist even where voters do not believe that they are monitored.

The two approaches also yield competing predictions about retrospective voting. In the

instrumental view, voters care exclusively about prospective costs and benefits (Kitschelt and

Wilkinson 2007b: 25, 342; Lyne 2007; Stokes 2005). Although their perceptions of what benefits

they may receive in the future are likely informed by what they have received in the past,

instrumentally motivated voters care only about how today’s interactions condition their future

10

stream of benefits. Clients motivated by obligation, however, may demonstrate loyalty to leaders

who did right by them in the past, regardless of what they stand to gain in the future. That is,

their actions today may be motivated by repaying a past favor rather than as part of a

calculation about what they expect to receive in the future. They may thus cast their ballots

based on purely retrospective considerations (see Kinder and Kiewiet 1979, Fiorina 1981, Cain

et al. 1987). Intriguingly, they may even seek to reciprocate perceived injuries and injustices by

voting against politicians who failed to deliver when expected to do so. This phenomenon,

known as “altruistic punishment,” can even lead citizens to turn against patrons when doing so

diminishes the chances of receiving clientelist benefits in the future (Fehr and Gächter 2000,

Fehr and Gächter 2002, de Quervain et al. 2004).

Political attitudes also play a key role in the reciprocity approach to clientelism.

Instrumentalist approaches generally discount citizens’ attitudes because such attitudes do not

affect the costs and benefits of supporting the machine. In the norms-based approach, however,

civic values create conflicting obligations that should make voters less likely to participate in

clientelist arrangements or follow through on a clientelist bargain once it is struck. Civic-minded

voters should thus be more likely to see clientelist exchanges as illegitimate, to reject proffered

benefits, and to perceive a value conflict between clientelist and civic obligations.

Finally, the reciprocity framework predicts that the provision of benefits by politicians

induces feelings of indebtedness among recipients to support the politicians in question. As the

value of a particular benefit rises in the mind of the recipient, he should become a more reliable

client. Instrumentally minded voters clearly also assess the value of a proffered benefit such that

voters evaluate whether a given good is “worth” their vote. To the extent that the two

perspectives differ on this point, they predict differences in how voters react to two benefits of

similar monetary value. For the strictest of instrumentalists, two such benefits should elicit

similar responses, whereas in the reciprocity framework, they may elicit widely varying feelings

11

of obligation. For instance, a needed “just in time” service such as a free visit to a doctor for a

sick child may elicit more obligation than a cash payment that could purchase such a visit.

Instrumentalist arguments are currently better-established in the political science

literature, yet as Brusco and colleagues (2004:81) point out, neither framework has been

subjected to a systematic test. For this reason, we first summarize supporting evidence for the

norms of reciprocity approach from existing studies using a variety of methods. We then present

new data from split-sample experiments embedded in two new surveys from Mexico in which

registered voters were selected at random from seven precincts (2009, N=545) and four

12

different precincts (2010, N=360) in the Federal District of Mexico City and the State of

13

Mexico.10 We view this combination of observational and experimental methods as a step

beyond the observational approach researchers have used to test principal-agent models of

clientelism based on instrumentalist micro-foundations (Stokes 2005; Brusco et al. 2007).

Clientelist Exchanges Generate Obligation

If the provision of selective benefits failed to conjure up feelings of obligation, there

would be little reason to think such feelings motivated clientelist exchange. Consistent with our

argument, ethnographic accounts of clientelism produced over the last 75 years are drenched in

the language of obligation (inter alia, Gosnell 1937, Weinstein et al. 1969, Powell 1970, Scott

1972, Lemarchand and Legg 1972, Lemarchand 1972, Eisenstadt and Lamarchand 1981,

Eisenstadt and Roniger 1984, Erie 1988, Roniger 1990, Fukui and Fukai 1996, Gay 1998,

Auyero 1999, Auyero 2000, Auyero 2001, Levitsky 2003, Gay 2006). Clients regularly report

feeling indebted to the politicians who provide them with benefits and that such obligation

compels them to act accordingly (Gosnell 1937, Erie 1988, Auyero 2002, Gay 1998, Gay 2006,

Wang 2007).

One example of reciprocity from recent fieldwork comes from the so-called “new leaders”

(naya netas) in northern India. Although these village notables cannot monitor voters’ behavior,

they can count on beneficiaries to support them nonetheless. As one naya neta put it:

It is a matter of keeping faith. People can obviously vote as they wish. But most

people remember well who has helped them in times of need. And it is only a

rare person who is faithless (Krishna 2007: 148).

Field reports on electoral politics in other countries, including Benin, Japan, northern Portugal,

and the Philippines have uncovered the same dynamic (Schaffer and Schedler 2007: 21).

If the approach we propose is correct, not only should benefits generate obligations, but

variation in the value of benefits should affect the degree of obligation. Data from existing

14

surveys on Brazil and Mexico suggest that such a relationship does indeed exist. For instance,

respondents in the 2002 Brazilian Election Study, which polled over 2,500 adults nationwide,

were asked whether voter should accept a particular gift and then vote for the party that

provided it in return. In the aggregate, respondents ordered items in a predictable way, reporting

the greatest obligation to support a party that provided medical care for a sick child (61%),

followed by one that arranged for a spot in school for the voter’s child (52%), provided groceries

(46%), and gave the voter a bicycle (30%). (See Supporting Materials.) Likewise, respondents

in the nationally representative Mexico 2006 Panel Study reported that providing important

public services such as water, sewage, or electricity created more of an obligation than offering

a bag of groceries, which in turn was worth more than holding a neighborhood party. Data on

self-reported receipt of gifts from the same survey also show that the more recipients said that

the gifts mattered to them, the greater the obligation they felt to support the party providing it (r

= .51, p < .01, N = 79). (See Supporting Materials.)

Our own surveys provide a more direct test of this hypothesis using a series of split-

sample experiments. Interviewers read third-person vignettes of the following type, in which half

the sample was prompted with the smaller sum and half with the larger sum: “Let’s imagine that

15

a candidate for municipal president offers Gabriel/Gabriela [50/500] pesos11 in exchange for

his/her vote and Gabriel/Gabriela accepts the money. In your opinion, how much obligation

should Gabriel/Gabriela feel to vote for this candidate – a lot, some, a little, or none?” Similar

vignettes were read where the benefit in question was a bicycle, several bags of cement, or

medical treatment for a sick child.

We used hypothetical third-persons to diminish social desirability bias in responses,

16

given that many of the practices discussed were illegal.12 Interviewers matched the gender of

the person in the vignettes to that of the respondent in order to enhance identification with the

subject of the vignette (King 2004). We also used eight versions of the survey to prevent

answers to one question from affecting answers to another (“contamination bias”).

The results presented in Table 2 show that respondents, in the aggregate, had a clear

ordering of the degree of obligation that a citizen should feel when he or she accepts a benefit in

exchange for political support. (The use of split-sample survey experiments prevents us from

examining individual preference orders over the whole range of items in Table 1.) Small

monetary payments induced the least amount of obligation to support a candidate, with

approximately one quarter of the sample reporting some obligation. Larger monetary payments

generated more in the split-sample experiments, but less than other gifts. A gift of a bicycle

induced obligation among more than one-third of respondents, and a bag of cement (crucial to

people living in poorly constructed homes) raised that proportion to almost two-fifths. As we

would expect, the provision of a doctor’s visit for a sick child was the most powerful inducement,

with nearly half of respondents saying that such an act should create a sense of obligation in the

voter to support the politician providing it. Importantly, both the provision of cement and a

doctor’s visit were valued more highly by respondents than receiving 500 pesos, even though

this sum would typically be more than enough to purchase either benefit. (These differences

were statistically significant at the .01 level.)

[Table 2 about here]

Evidence from a variety of research methods and across several countries thus shows

that clientelist benefits generate feelings of obligation among voters. But do feelings of

obligation actually lead people to choose candidates on the basis of clientelist appeals? In our

2009 survey in Mexico, we asked half the respondents whether they would vote for the party

with which they sympathized or the party that resolved an important legal issue for them in the

17

past.13 Only 30.2% inclined toward the party with which they sympathized whereas 54.4% chose

the party that gave them a benefit in the past. (Another 15.4% were uncertain which party to

choose.) In other words, obligation stemming from past receipt of benefits weighed more heavily

on voters than their partisan sympathies. Other research shows that party affiliation powerfully

influences vote choices in Mexico (Lawson and McCann 2003, Domínguez and Lawson 2003,

Domínguez et al. 2008) and that one component of partisanship concerns issues preferences

that underlie prospective programmatic voting (Greene 2007).

As another test of the relationship between obligation and voting, we constructed an

index of political obligation by adding responses from three questions that were asked of all

respondents in our 2010 survey: how much obligation should Gabriel/Gabriela feel to vote for a

party that gave him/her a) 500 pesos, b) a paid doctor’s visit for his/her sick child, and c) a

rooftop water tank in exchange for his/her vote. A separate item on the survey asked

respondents whether a voter who lives in an area that suffers from water shortages should

choose a candidate that offers him/her a week’s worth of water before the election in exchange

for his/her vote or one who offers nothing before the election but promises to improve the water

system in the area if he wins. This question thus asks voters to choose between a pre-electoral

selective benefit that we view as a clientelist good and a post-electoral community benefit. In

logistic regression models where preference over candidates offering one or the other type of

benefit is the dependent variable, the coefficient on the index of political obligation is correctly

signed and significant at the .05 level. (See Table 3.) The magnitude of the effect remains

virtually unchanged when we control for other variables that could plausibly affect orientations

toward clientelism, including family wealth as measured by an inventory of common household

items (Kitschelt and Wilkinson 2007; Stokes 2005), education, feelings of political efficacy (as

measured by a standard question about whether or not respondents think that politics is too

complicated for them to understand), and feeling thermometer ratings of the main political

18

parties. (We return to the effect of civic values, also included in the model, in a subsequent

section.) A simulation using the full model (“Model 4” in Table 3) shows that an otherwise

average individual who feels the highest level of political obligation is 12% more likely to choose

the clientelist candidate. This change more than doubles the likelihood that an individual would

19

support the machine, from 10% to 22%, all else equal.14

[Table 3 about here]

Evidence from a variety of sources thus supports the notion that the (1) the provision of

selective benefits produces feelings of obligation, (2) subjectively more valuable benefits

produce more obligation, and (3) feelings of obligation are associated with clientelist voting.

Clientelism Persists Despite Ballot Secrecy

Instrumental interpretations of clientelism rest on brokers’ ability to circumvent the secret

ballot. In contrast, our norms-based approach implies that voters spontaneously support the

machine, even where ballot secrecy is secure. Ample evidence using several research methods

supports this view.

Field research shows that clientelism persists without policing of voters’ actions in many

contexts. For instance, Van de Walle (2007: 64) points out that individualized monitoring is

implausible in countries like Nigeria and Benin, which are not generally regarded as strangers to

clientelism. Wang and Kurzman’s (2007: 233) detailed analysis of vote buying in a 1993 contest

for county executive in Taiwan reaches the same conclusion: although the ballot was secret,

operatives from the Kuomintang documented the purchase of 14,090 votes by relying on

networks of trust and obligation. In a number of other countries that lack structured and

enduring political parties that could act as clientelist machines, analysts routinely report high

levels of clientelism. Peru has been referred to as a case of party system collapse (Seawright

2011; Morgan 2011), yet in the 2010 elections, 24.5% of voters reported involvement in

clientelist exchanges (González Ocantos et al. 2011a). More generally, the degree of party

institutionalization across Latin American countries is not related to voters’ reported involvement

in clientelism. Paraguay, Colombia, Brazil, Bolivia, and Guatemala show lower than average

20

levels of party structuration with higher than average levels of clientelism.15

Nationally representative survey data also show that clientelism extends much further

than monitoring. For instance, in São Tomé e Príncipe, 90% of subjects said that brokers

attempted to buy their political support in parliamentary elections, yet only 14% reported any

attempt at monitoring or sanctioning (Vicente 2008). In a similar vein, survey data from Brazil

indicate that many citizens would be willing to carry through on vote-selling transactions, even

when the questions asked do not imply any policing of voters’ behavior (see Supporting

Materials).

In our surveys, Mexican voters also perceived that the parties had a relatively limited

capacity (or desire) to monitor votes: just 25.5% said that they saw party representatives inside

polling places trying to determine who voters chose and only 12.1% reported seeing party

representatives threaten to sanction voters. Nevertheless, 71.4% believed that parties regularly

or sometimes try to buy votes in their neighborhood, and 69.3% think that people in their

neighborhood sell their vote (50.4% say that it occurred with frequency and another 18.9% that

it happened sometimes). Thus, far more clientelism occurs than the parties’ monitoring

capabilities would seem to support.

In one particularly clever analysis of heavily clientelist Paraguay, Finan and Schecter

(2009: 17) show that individuals who exhibit greater reciprocity in experimental trust games

devoid of any political content and with no implication of monitoring, were more likely to report

involvement in clientelist exchanges (as measured in separate surveys) than those who evinced

less reciprocity. As Vicente concludes, such findings support “the idea that self-enforcement

may be the main mechanism by which vote buying works” in settings where such behavior is

pandemic (2008: 21).

Voters who believe their own choices are monitored could conceivably underreport

monitoring or over-report obligation in opinion polls. If so, then voters who believe the vote is not

21

secret should report higher levels of obligation, indicating that our findings are an artifact of

voter fear. As a test, we compared the obligation induced by the items that appear in Table 2

with perceptions of monitoring. Respondents who suspected that voting was monitored reported

slightly greater obligation for three of the five items; however, these differences were minor and

not statistically significant. We also tested whether perceptions of ballot secrecy affected

respondents’ beliefs that a voter should choose a party offering discrete material benefits over

one offering a broader community benefit. Those who believed that brokers can monitor their

vote choices were no more likely to choose the clientelist party. In other words, variation in

reported obligation is not a product of beliefs about whether parties can monitor voters’

behavior.

Obligation Produces Purely “Retrospective Voting”

In instrumentalist models of clientelism, voters care about the future payoffs they will

receive from a clientelist party. If voters consider past performance, it is only as an indicator of

the likely future stream of clientelist goods that they may receive by supporting the machine

(Cox and McCubbins 1986; Dixit and Londregan 1986; Lindbeck and Weibull 1986; Kitschelt

and Wilkinson 2007b: 25, 342; Lyne 2007, Stokes 2005). In our norms-based approach, by

contrast, voters may make their choices based on purely retrospective considerations because

they may feel an obligation to support politicians who have done right by them in the past,

regardless of what they stand to gain in the future.

The link between the provision of favors or services and retrospective voting is

commonplace in developed countries under the name “constituency service”. American

congressmen devote enormous attention to this activity – between one quarter and one third of

their time – and their efforts appear to pay handsome dividends (Fenno 1978; Fiorina 1981,

Yiannakis 1981, Cain et al. 1987, Fiorina and Rohde 1989). The dividing line between such

22

constituency service and clientelism is not always clear. For instance, Japanese politicians

maintain massive personal support networks (koenkai) that dole out cash gifts at funerals and

furnish constituents with material benefits in times of need (Curtis 1971, Fukui and Fukai 1996,

Scheiner 2007). American political machines, as well as constituency service efforts in Ireland

(Komito 1984), Italy (Rossetti 1994, Golden 2003), and Mexico (Hilgers 2005) similarly blur the

distinction between these activities. Both types of retail politics rely on voters’ obligation to pay

back past benefits with political support.

Although systematic investigation on this sort of retrospective voting has been limited,

the survey data that does exist demonstrates the behavior among voters that we predict. For

instance, in a 2002 survey of voters in Argentina conducted by Brusco et al. (2004) and used as

evidence in favor of the instrumentalist approach (Stokes 2005), respondents were asked

whether they had received a gift from a local representative of a party or organization during the

campaign. If they responded “yes,” they were asked whether the gift affected their vote.

Respondents who reported being influenced were then asked how the gift affected them in an

open-ended question (with verbatim responses recorded). Among those who reported being

influenced, 14% said that they felt an obligation to vote for the party that gave them the gift; 21%

said they changed their vote because the party “helped” them; 17% said they normally voted for

parties who gave them things; and others reported vague positive sentiments based on

retrospective rather than prospective considerations. None reported using prospective criteria.

Other data cast even more doubt on voters’ use of prospective criteria. If clients think

prospectively, then presumably they would only change their vote intentions to candidates that

they think will actually win election and thus be in a position to offer benefits in future elections.

Testing this hypothesis requires panel survey data that allows observation of respondents’ vote

intentions over time as well as whether they receive a clientelist payoff, and their assessments

about each candidate’s likelihood of winning. We only found one survey that meets these

23

criteria: the Mexico 2000 Panel Study.

In Mexico’s 2000 elections, 24% of voters polled reported that they received some sort

of a payoff from one of the parties (Greene 2007: 215) and these payoffs affected vote choices.

Those who received payoffs in the crucial period between early June and the July 6 election

were more likely to switch their vote intention to a final vote choice for the candidate who doled

out the benefit. Models that we report in the appendix show that independent voters who

received a gift from the PRI were 6.4% more likely to vote for that party’s candidate (Labastida

7-13.4), Fox 13.6-26.8, CCS 17.1-18.1.

Among voters who received a payoff, 30.9% changed their vote intention to the

candidate that gave them the benefit, even though they thought that this candidate was unlikely

to win the presidency. The most striking finding concerns the candidacy of leftist Cuauhtémoc

Cárdenas. As his ill-fated third run for the presidency wore on, voters abandoned him in droves

and belief that he would win plummeted. Yet among voters who reported receiving a payoff from

Cárdenas’s party in the last month of the campaign season and said that they voted for him in

the post-election wave of the panel survey, fully 47.1% thought that he was unlikely to win the

presidency when polled less than a month prior to the election. It is highly improbable that

voters who received payoffs from Cárdenas’s party came to believe that he could win during the

final month of his ill-fated and, by then, flailing campaign. Not only were his prospects dim and

getting dimmer in the national media, but most voters (60% to be exact) who received payoffs

by the end of April believed Cárdenas was less likely to win when re-interviewed at the

beginning of June.

These striking findings cannot be accommodated within the instrumentalist framework.

Voters motivated by future payoff would be foolish to choose a candidate that they think cannot

win office, and therefore cannot leverage the spoils of office to make clientelist payments in

future elections. These findings instead imply that voters are motivated by non-prospective

24

criteria.

If many clients do not make their vote choices with an eye toward future payoffs, then

what motivates their choices? Our 2009 survey investigated this issue further. Half of the

sample was asked how much obligation a hypothetical third person would feel to support a party

that promised to deliver a particular good or service (e.g., a medical center) in the future; the

other half of the sample was asked how much of an obligation the person would feel to support

a party that had provided a good or service in the past. If voters ignore or discount past benefits,

then they would presumably feel much less inclination to vote for parties that did right by them

previously than they would for parties that promise desirable benefits in the future. Yet the data

show that this is not the case; respondents reported no greater inclination to support a party that

promised to build a medical clinic in the neighborhood if it won the election there than one that

reminded them it had built such a clinic in the past.

This null result supports our argument, but it is possible that it emerges from the data

because voters do not believe that candidate’s promises are credible. The structure of our 2010

survey allows us to investigate this issue further. As above, respondents were asked how much

obligation a hypothetical person should feel to vote for a party that promised a medical center in

their community if it won and, separately, the obligation created when a party reminded that it

25

improved medical services in the community when last in office.16 Immediately preceding these

questions, half of the sample was primed with a question that featured a party that had made

good on a prior promise. Even these respondents reported more obligation for the party that

reminded of a past benefit than one that promised a future one.

If voters are motivated by purely retrospective considerations, then they may also

engage in another form of retrospective voting by repaying past insults or injuries. Because

clientelism rests on voters’ responses to past favors, failure by politicians to deliver selective

benefits will erode the “moral economy” of machine politics, provoking not only detachment from

clientelist parties but perhaps even a sense of moral indignation among those who were part of

the clientelist network. Erstwhile clients who do feel cheated may then turn against the machine,

even if they are still more likely to receive a future payoff from the party they previously

26

supported than from an opposing party.17

To test for the possibility that voters practice such “altruistic punishment” (Arnold 2001,

Fehr and Gächter 2002), we compare responses to two separate items from our 2010 survey.

The first asked whether a hypothetical voter should feel obliged to vote for a party from which

she normally receives foodstuffs in exchange for support but which did not deliver anything this

year. The second asked respondents whether they agreed with the statement “when someone

takes advantage of me, I get them back” (which was asked without reference to politics). Those

who strongly agreed with the second statement – thus demonstrating a willingness to practice

altruistic punishment in general – were much more likely to think that the spurned voter should

choose a rival party rather than abstain after benefits were withdrawn (chi-squared = 4.14 for

three degrees of freedom, p < .05).

Civic Attitudes Discourage Clientelism

The extent of political clientelism clearly varies across different political contexts and a

viable theory of clientelism must account for such differences. As noted above, even if feelings

of obligation are activated, other feelings might diminish potential clients’ likelihood of following

through on the bargain. We specifically focus on civic attitudes that prompt citizens to see

clientelist exchanges as illegitimate. We present evidence from experiments that prime civic and

anti-civic primes attitudes as well as mass surveys that measure the likelihood that citizens with

different levels of civic values will enter into clientelist exchanges.

Clever field experiments show that civic attitudes influence the effectiveness of clientelist

appeals across sub-national units. For instance, Vicente (2008) demonstrated that exposure to

anti-vote-buying literature in São Tomé e Príncipe increased perceptions that other people in

the neighborhood were voting their conscience and reduced the extent of vote buying. In the

same vein, Wantchekon (2003: fn 20) showed that Beninese villages that were randomly

27

selected to receive nationally oriented, “programmatic” appeals were more critical of vote-buying

than villages that received parochial (regionally-oriented), clientelist appeals.

Surveys that include adequate questions on civic attitudes and clientelism suggest a

strong relationship between the two across individuals. For instance, in the 2002 Brazilian

Election Study, respondents who more strongly favored democracy as a system of government

were significantly more likely to believe that voters should not accept selective benefits in return

for their vote (p < .01). Likewise, respondents in the Mexico 2006 Panel Study who expressed

support for democracy as a system of government were more likely to reject the notion that a

28

hypothetical third person should be willing to trade his vote for groceries (p <.01).18 In both

cases, these relationships survive controls for potential confounding factors like income,

education and age. (See Supporting Materials for details.)

Our 2009 Mexico survey employed a split-sample experiment to test hypotheses about

civic attitudes more directly. Half of respondents were asked whether they believed that it is

important to live in a democracy (the civic prime) and the other half were asked whether they

agreed that people did not get ahead unless they took advantage of others (the anti-civic prime).

All respondents were then asked whether it would be acceptable or unacceptable for a person

to give his or her vote in exchange for a job for a relative, followed by the question that the other

half of the sample had received first. Thus, all respondents received the same questions, just in

a different order. Despite the subtle nature of the experimental manipulation, differences

between the two groups were large and significant. Whereas 36% of respondents who received

the anti-civic prime first were willing to accept the clientelist exchange, only 25% of those who

answered the question on democracy first were similarly inclined (p < .05).

As a further test, we examined both the partial effect of civic attitudes on clientelist voting

and the joint effect of civic attitudes and the index of political obligation presented above on

voting for a clientelist party. These results appear in Table 3 (above). We measured civic

attitudes with a simple additive index of responses to items on how disappointed the respondent

would be in a friend who did not pay his household electric bill, stole a soft drink from a store,

29

did not pay bus fare, or earned money by selling marijuana.19 In logistic regression models that

also control for feelings of political obligation and efficacy, household income, education, and

partisan identification, the coefficient on the index of civic attitudes was correctly signed and

always significant at the .1 level or better. In simulations based on this model, a voter who is

typical in every way but highly civically minded would be 16% less likely to cast her lot with a

clientelist party than a non-civic voter, cutting in half support for the machine. A voter who

reports the highest levels of political obligation to patrons and the lowest levels of civic

obligation would be more than 33% more likely to support the machine than voters who are

30

least civic-minded and most clientelist in orientation.20 Thus, whereas feelings of obligation

encourage voters to support a clientelist party, feelings of civic duty discourage them from doing

so.

4. Conclusion: Durable and Self-Enforcing Clientelism

Scholars in the instrumentalist tradition have occasionally acknowledged that clientelist

linkages can involve strong feelings of affect and duty. As Kitschelt and Wilkinson write:

Continued interaction and exchange between patrons and clients over time – for

example, at local celebrations – may eventually make regular monitoring of

voting unnecessary because…the interaction may be sufficient to induce cultural

expectations of reciprocity inherent in any gift-giving situation (2007: 15).

However, instrumentalist accounts treat such examples as isolated exceptions or aberrations,

rather than as manifestations of a separate, solid basis for clientelism. We argue that the

obligation to reciprocate is the basis for a theoretically and empirically well-supported alternative

mechanism for machine politics in a range of settings.

We do not argue that instrumental calculation is unimportant for clientelism; the credible

threat of sanctions and rewards is surely a fundamental feature of machine politics in many

places. Rather, we suspect that there is substantial variation in the foundation of clientelism

across countries (or regions within the same country). In other words, there may be places

where the absence of spontaneous support necessitates close monitoring of voters by the

machine, places where politicians can rely on norms of obligation to garner votes, and places

where brokers find recourse to both mechanisms.

The role of obligation has broad ramifications for relationships between candidates and

voters. Although we have focused on voters, our argument need not be limited to the base of

the machine. Precinct captains, too, may be motivated by feelings of obligation to the parties

31

and politicians on whose behalf they operate and thus require less monitoring from their bosses.

It is even possible that some types of patrons – e.g., local notables or religious figures who

control discretionary resources – may take obligations to their clients seriously. Ironically, such a

moral sense on the part of intermediaries and patrons would actually strengthen clientelism,

because brokers motivated in this fashion would be less likely to steal the funds they were to

distribute (and thus have more resources to distribute) and promises made by politicians who

were perceived to care about their constituents might be viewed as more credible (and thus

more likely to inspire compliance on Election Day).

Another implication of our findings is that the distinction between clientelist exchange

and constituency service is not as clear-cut as observers might assume. Both types of

interactions between politicians and constituents create a sense of obligation toward politicians

who provide selective benefits. The electoral rewards of such personal exchanges make retail

politics a durable element of all democracies.

Self-enforcing clientelism also affects party structures. If voters spontaneously comply

out of obligation, then clientelism may not require the highly organized and deeply rooted

political machines that most analysts now argue is necessary (Kitschelt 2000; Kitschelt and

Wilkinson 2007b; Gyrzmala-Busse 2005). For instance, Stokes (2005: 317) argues that “The

typical political machine (or clientelist party) is bottom-heavy, decentralized, and relies on an

army of grassroots militants.” Such elaborate structures may be useful to would-be patrons, but

they may not be necessary. Personalist politicians who distribute benefits may be able to build

clientelist relationships without building full-fledged monitoring apparatuses. This possibility may

help explain the persistence of clientelist politics in democracies around the globe in a period

when economic austerity and media-centered campaigning has diminished party density (Mair

and Van Biezen 2001), transformed mass parties into “electoral-professional” ones with far

fewer local activists (Panebianco 1988), and undermined traditional political machines in many

32

countries (Levitsky 2003; Greene 2007; Weyland 1996).

Finally, our conclusions have important implications for efforts to curb machine politics.

One obvious strategy – with which instrumentalist approaches to clientelism would presumably

agree – is to eliminate the discretionary resources available to political parties and candidates.

Not only would restricting the flow of selective benefits prevent politicians from generating new

obligations, the withdrawal of resources could also lead voters who had come to rely on such

selective benefits to rebel against the party that suddenly failed to hold up its end of the

traditional bargain between patrons and clients.

As long as such resources remain available, public policies designed to reinforce the

secret ballot may prove insufficient to break up political machines. Rather, efforts to instill civic

values also have a role to play. Where norms of obligation underwrite clientelism, machine

politics faces a serious challenge if citizens come to view clientelist transactions as illegitimate,

feel obligated to vote their conscience, or come to view their patrons’ largesse as a right rather

than an obligation that must be repaid with political loyalty (Rossetti 1994). These shifts in

mindset constitute a key part of the “difficult transition from clientelism to citizenship” (Fox 1994,

Gay 2006) – a transition that is difficult precisely because it requires not only institutional reform

but also attitudinal change.

33

Table 1. Observable Implications of the Reciprocity and Instrumentalist Frameworks

Issue Reciprocity Predictions Instrumentalist PredictionsFeelings of obligation Citizens feel an obligation to

support politicians that provide them with selective benefits

Feelings of obligation should not affect vote choices

8 Where status hierarchies already exist, exchanges between social superiors and subordinates reinforce them (Aragon 1996: 49-50,

Weiner 1980). Gifts from followers to leaders (tribute) help to secure protection or assistance, whereas gifts from leaders to followers

cultivate gratitude and sustain bonds of loyalty.

9 Scholars working in the instrumentalist tradition have long recognized the power of partisan loyalty and policy proximity to moderate

the effect of payoffs. For instance, opponents of the machine are considered too costly to buy (Stokes 2005) and supporters are

predicted to respond more positively to payoffs (Cox and McCubbins 1993).

10 The Supporting Materials provide information on site selection and sampling and show that our procedures effectively randomized

the key treatments across respondents.

11 The exchange rate at the time was approximately 12 pesos per dollar.

12 When asked directly whether they would accept a payoff in exchange for their vote, just 6% of respondents answered in the

affirmative. Yet 32.8% said that they had personally received a good or service from a party representative in a recent election. The

strong presence of social desirability bias thus made the use of hypothetical third-persons important.

13 The other half of the sample was asked the same question, except that voters were told they did not sympathize with the party of

either candidate; we use only the first half of the sample here. Not surprisingly, more respondents in the other half of the sample 34

Greater benefits cause greater feelings of obligation

Monitoring Clientelism can exist despite ballot secrecy

Surveillance/lack of ballot secrecy required for clientelism to persist

Retrospective clientelist voting

Purely retrospective voting based on previous provision of benefits

Retrospective considerations only matter as an indicator of

were even more inclined to support the politicians that had given them a specific benefit.

14 Political obligation as measured by our index is also associated with other aspects of clientelist voting. The 10% of respondents

who reported asking a politician for a favor evinced more obligation than those who had never asked for a favor (p < .01). The 6% of

the sample that admitted willingness to exchange their vote for a payoff also felt more obligation than those unwilling to enter into a

clientelist exchange (p < .1).

15 Party institutionalization data come from Jones (2007: 83). Public opinion data come from Americas Barometer 2010. When

unobtrusive measures are used to tap clientelism, rates are substantially higher (González Ocantos et al. 2011b).

16 These questions were asked consecutively rather than as a split sample.

17 Instrumentalist models that employ a standard punishment path strategy such as grim trigger (Stokes 2005) yield similar

hypotheses; however, instrumentally motivated (and thus future-oriented) voters should not rationally turn against patrons who spurn

them in any one election cycle if such a party might still provide them with benefits in the future.

18 A separate survey in Argentina (see Stokes 2005) shows that the correlation between clientelism and civic attitudes does not reach

statistical significance. This null result is the only finding that does not support our argument from any available survey that contains

the requisite questions; it may be a product of the small number of respondents who believed clientelism to be legitimate 35

Possible “altruistic” punishment for cessation of benefits

future payoffs

Civic attitudes Civic values undermine clientelism

Attitudes about clientelism and civic life do not affect clientelist exchanges

(approximately 13%).

19This battery has been used in a number of other surveys sponsored by the Latin American Public Opinion Project (see

http://www.vanderbilt.edu/lapop/).

20 Across all models, first differences for political obligation and the joint effect of political obligation and civic duty are statistically

significant at the 95% level. The effect of civic duty alone is significant at the 90% level.36

Table 2. Degree of Felt Obligation for Benefits Offered, Mexico 2009

Benefit offered Percent who felt “a lot” or “some” obligation to support

politician providing giftMean

obligation

Percent of respondents whothought it correct to accept

the gift50 pesos 24.6 0.78 23.9500 pesos 28.2 0.89 24.9

Bicycle 33.8 1.00 30.2

Cement 39.7 1.15 32.4

Doctor visit 49.9 1.48 71.8

Note: The mean value is calculated on a scale of 0 to 3 where 0 is no obligation, 1 is little, 2 is some, and 3 is a lot. The differences between adjacent benefits in the table are not statistically significant except cement versus medical treatment for which p<.01. Differences between all other paired comparisons are statistically significant at the .01 level.

37

Table 3. Logistic Regression Models of Support for a Clientelist Party

Variable Model 1 Model 2 Model 3 Model 4B SE Sig B SE Sig B SE Sig B SE Sig

Political Obligation Index 0.10 0.04 ** 0.09 0.04 ** 0.10 0.05 ** 0.10 0.05 **Civic Attitudes Index -0.09 0.05 * -0.08 0.05 * -0.10 0.05 *Political efficacy -0.13 0.33 -0.15 0.34 -0.38 0.36SES -0.07 0.13 -0.01 0.13Education 0.10 0.08 0.14 0.08PAN feeling thermometer -0.08 0.07PRD feeling thermometer -0.03 0.06PRI feeling thermometer 0.05 0.06Constant -2.07 0.25 *** -1.12 0.55 ** -1.48 0.81 * -1.50 0.92 *Number of cases 314 295 294 284

Note: Dependent Variable is coded as 1 for voting for a clientelist party and 0 otherwise.* p<.1, ** p<.05, *** p<.01, two tailed tests.

38

References

Allswang, John M. 1977. Bosses, Machines, and Urban Voters: An American Symbiosis. Port

Washington, NY: Kennikat.

Aragon, Lorraine V. 1996. “Twisting the Gift: Translating Precolonial into Colonial Exchanges in

Central Sulawesi, Indonesia”, American Ethnologist, 23 (1): 43-60.

Auyero, Javier. 1999. “‘From the client’s point(s) of view’: How poor people perceive and

evaluate political clientelism”, Theory and Society, 28 (2): 297-334.

Auyero, Javier. 2000. “The Logic of Clientelism in Argentina: An Ethnographic Account”, Latin

American Research Review, 35 (3): 55-82.

Auyero, Javier. 2001. Poor People’s Politics: Peronist Survival Networks and the Legacy of

Evita. Durham: North Carolina: Duke University Press.

Bartlett, Peggy F. 1980. “Reciprocity and the San Juan Fiesta”, Journal of Anthropological

Research, 36 (1): 116-130.

Berg, Joyce, John Dickhaut, and Kevin McCabe. 1995. “Trust, Reciprocity, and Social History”,

Games and Economic Behavior, 10 (1): 122-142.

Bienenstock, Elisa Jayne and Alison J. Bianchi. 2004. “Activating Performance Expectations

and Status Differences through Gift Exchange: Experimental Results”, Social

Psychology Quarterly, 67 (3): 310-18.

Birch, Leann Lipps and Jane Billman. 1986. “Preschool Children's Food Sharing with Friends

and Acquaintances”, Child Development, 57 (2): 387-395.

Boissevain, Jeremy. 1966. “Patronage in Sicily”, Man, 1 (1): 18-33.

Brusco, Valeria, Marcelo Nazareno, and Susan Stokes. 2004. “Vote Buying in Argentina”, Latin

American Research Review, 39(2):66-88.

Cain, Bruce, John Ferejohn, and Morris P. Fiorina. 1987. The Personal Vote: Constituency

Service and Electoral Independence. Cambridge: Harvard University Press.

39

Calvo, Ernesto and María Victoria Murillo. 2004. “Who Delivers? Partisan Clients in the

Argentine Electoral Market”, American Journal of Political Science, 48(4):742-757.

Cambanis, Thanassis. 2006. “With Speed, Hezbollah Picks up the Shovel: Group’s Engineers,

Funds Pour into War Torn Lebanon”, Boston Globe, August 19, 2006.

Caplow, Theodore. 1984. “Rule Enforcement without Visible Means: Christmas Gift Giving in

Middletown”, American Journal of Sociology, 89 (6): 1306-23.

Carmichael, H. Lorne and W. Bentley MacLeod. 1997. “Gift Giving and the Evolution of

Cooperation”, International Economic Review, 38 (3): 485-509.

Centeno, Miguel Angel. 1997. Democracy within Reason: Technocratic Revolution in Mexico.

2nd Edition. University Park, PA: Penn State Press.

Cesarini, David, Christopher T. Dawes, James H. Fowler, Magnus Johannesson, Paul

Lichtenstein, and Björn Wallace. 2008. “Heritability of Cooperative Behavior in a Trust

Game”, Proceedings of the National Academy of Sciences, 10.1073/pnas.0710069105

(published online 3 March).

Chanrda, Kanchan. 2007. “Counting Heads: A Theory of Voter and Elite Behavior in Patronage

Democracies” in Patrons, Clients, and Policies: Patterns of Democratic Accountability

and Political Competition. New York: Cambridge University Press, pp. 80-109.

Chubb, Judith. 1982. Patronage, Power, and Poverty in Southern Italy. Cambridge, England:

Cambridge University Press.

Clapham, Christopher, ed. 1982. Private Patronage and Public Power : Political Clientelism in

the Modern State. London : Frances Pinter.

Cornelius, Wayne. 1975. Politics and the Migrant Poor in Mexico City. Stanford: Stanford

University Press.

Cox, G. W. and J. M. Kousser. 1981. “Turnout and Rural Corruption: New York as a Test Case”,

American Journal of Political Science 25 (4): 646-63.

40

Dal Bó, Ernesto. 2007. “Bribing Voters”, American Journal of Political Science, 51 (4), 789–803.

de Quervain, Dominique J.-F., Urs Fischbacher, Valerie Treyer, Melanie Schellhammer, Ulrich

Schnyder, Alfred Buck, and Ernst Fehr. 2004. “The Neural Basis of Altruistic

Punishment”, Science, 305 (5688): 1254-58.

Jorge I. Domínguez, Chappell Lawson, and Alejandro Moreno, eds. 2008. Consolidating

Mexico’s Democracy: The 2006 Presidential Campaign in Comparative Perspective.

Baltimore: Johns Hopkins University Press.

Domínguez, Jorge I. and Chappell Lawson, eds. 2003. Mexico’s Pivotal Democratic Election:

Candidates, Voters, and the Presidential Campaign of 2000. Stanford and La Jolla, CA:

Stanford University Press and Center for U.S.-Mexican Studies, University of California

at San Diego).

Dreman, S. B. and Charles W. Greenbaum. 1973. “Altruism or Reciprocity? Sharing Behavior in

Israeli Kindergarten Children”, Child Development, 44 (1): 61-8

Eisenstadt, S. N. and R. Lamarchand, eds. 1981. Political Clientelism, Patronage, and

Development. Beverly Hills, CA; Sage.

Eisenstadt, S. N. and Luis Roniger. 1984. Patrons, Clients, and Friends: Interpersonal Relations

and the Structure of Trust in Society. New York: Cambridge University Press.

Elster, Jon. 1989. The Cement of Society: A Study of Social Order. New York: Cambridge

University Press.

Erie, Steven P. 1988. Rainbow’s End: Irish-Americans and the Dilemmas of Urban Machine

Politics. Berkeley, CA: University of California Press.

Fehr, Ernst and Urs Fischbacher. 2002. “Why Social Preferences Matter: The Impact of Non-

Selfish Motives on Competition, Cooperation, and Incentives”, The Economic Journal,

112 (478): C1-C33.

Fehr, Ernst and Simon Gächter. 2000. “Fairness and Retaliation: The Economics of

41

Reciprocity”, Journal of Economic Perspectives, 14 (3): 159-81.

Fehr, Ernst and Simon Gächter. 2002. “Alturistic Punishment in Humans”, Nature, 415: 137-40.

Fenno, Jr., Rochard. 1978. Home Style: House Members in their Districts. 1978. New York:

Little, Brown.

Finan, Federico and Laura Schechter. 2009. “Vote-Buying and Reciprocity”. Typescript,

accessed March 9, 2009, http://www.econ.ucla.edu/ffinan/Finan_Votebuying.pdf,.

Fiorina, Morris P. 1981. Retrospective Voting in American National Elections. New Haven: Yale

University Press.

Fiorina, Morris P. and David Rohde, eds. 1989. Home Style and Washington Work. Ann Arbor,

MI: University of Michigan Press.

Fowler, James H., Laura A Baker, and Christopher T. Dawes. 2008. “Genetic Variation in

Political Participation”, Scientific American, in process.

Fox, Jonathan. 1994. “The Difficult Transition from Clientelism to Citizenship: Lessons from

Mexico”, World Politics, 46: 151-84.

Fukui, Haruhiro and Shigeko N. Fukai. 1996. “Pork Barrel Politics, Networks, and Local

Economic Development in Contemporary Japan”, Asian Survey, 36 (3): 268-286.

Gächter, Simon and Benedikt Herrmann. 2008. "Reciprocity, culture, and human cooperation:

Previous insights and a new cross-cultural experiment," Discussion Papers 2008-14,

The Centre for Decision Research and Experimental Economics, School of Economics,

University of Nottingham.

Gay, Robert. 1994. Popular Organization and Democracy in Rio de Janeiro. Philadelphia:

Temple University Press.

Gay, Robert. 1998. “Rethinking Clientelism: Demands, Discourses, and Practices in

Contemporary Brazil” European Review of Latin American and Caribbean Studies 65: 7-

24.

42

Gay, Robert. 2006. “The Even More Difficult Transition from Clientelism to Citizenship: Lessons

from Brazil”, in Patricia Fernández-Kelly and Jon Shefner, eds., Out of the Shadows:

Political Action and the Informal Economy in Latin America. State College, PA:

Pennsylvania State University Press, pp. 195-217.

Gintis, Herbert, Samuel Bowles, Robert Boyd, and Ernest Fehr. 2003. “Explaining Altruistic

Behavior in Humans”, Evolution and Human Behavior, 24 (3): 153-72.

Golden, Miriam A. 2003. “Electoral Connections: The Effects of the Personal Vote on Political

Patronage, Bureaucracy and Legislation in Postwar Italy”, British Journal of Political

Science, 33: 189-212.

Gosnell, Harold F. 1937. Machine Politics: The Chicago Model. Chicago: University of Chicago

Press.

Greene, Kenneth F. 2007. Why Dominant Parties Lose: Mexico’s Democratization in

Comparative Perspective. New York: Cambridge University Press.

Grohol, J.M. (Feb 1999). Detecting deception: A quick review of the psychological research.

[Online]. http://psychcentral.com/archives/deception.htm.

Hammerstein, Peter, ed. 2003. Genetic and Cultural Evolution of Cooperation. Cambridge, MA:

M.I.T. Press.

Harik, Judith Palmer. 2004. Hezbollah: The Changing Face of Terrorism. New York: I. B. Tauris.

Harris, Mary B. 1970. “Reciprocity and Generosity: Some Determinants of Sharing in Children”,

Child Development, 41 (2): 313-328.

Heinrich, Joseph, et al. 2005. “Economic Man in Cross-Cultural Perspective: Behavioral

Experiments in 15 Small-Scale Societies”, Behavioral and Brain Sciences, 28; 795-815.

Herrmann, Benedikt, Christian Thöni, and Simon Gächter. 2008. “Antisocial Punishment Across

Societies”, Science, 319 (5868): 1362-7.

Johnson, Colleen Leahy. 1974. “Gift Giving and Reciprocity among the Japanese Americans in

43

Honolulu”, American Ethnologist, 1 (2): 295-308.

Kinder, Donald R., and D. Roderick Kiewiet. 1979. “Economic Discontent and Political Behavior:

The Role of Personal Grievances and Collective Judgments in Congressional Voting”,

American Journal of Political Science, 23 (2): 495-527.

King, Gary, Christopher Murray, Joshua Salomon, and Ajay Tandon. 2004. “Enhancing the

Validity and Cross-Cultural Comparability of Measurement in Survey Research”

American Political Science Review 98, 1: 191-207.

Kitschelt, Herbert and Arturas Rozenas. 2011. “Contingent exchange and Contractual

Opportunism: Making Clientelistic Relationships Work”. Manuscript presented at the

“Workshop on democratic Accountability Strategies”, May 18-19, Duke University,

Durham, NC. Accessed online at

http://duke.edu/~kkk4/2011_clientelism/Rozenas_2_4_Contingent_Exchange.pdf,

August 17, 2011.

Kitschelt, Herbert and Steven Wilkinson, eds. 2007a. Patrons, Clients, and Policies: Patterns of

Democratic Accountability and Political Competition. New York: Cambridge University

Press.

Kitschelt, Herbert and Steven I. Wilkinson. 2007b. “Citizen-politician linkages: an introduction,”

in Herbert Kitschelt and Steven I. Wilkinson, eds., Patrons, Clients, and Policies.

Cambridge: Cambridge University Press, pp. 1-49.

Klosko, George. 1990. “The Moral Force of Political Obligations”, American Political Science

Review, 84 (4): 1235-50.

Kohnken, G. (1987). Training police officers to detect deceptive eye witness statements: Does it

work? Social Behavior, 2, 1-17.

Komito, Lee. 1984. “Irish Clientelism: A Reappraisal” Economic and Social Review 15 (3):173-

94.

44

Kraut, R.E., & Poe, D. (1980). Behavioral roots of person perception: The deception judgments

of customs inspectors and laymen. Journal of Personality and Social Psychology, 39,

784-798.

Krishna, A. 2007. “Politics in the Middle: Mediating Relationships between the Citizens and the

State in Rural North India,” in Herbert Kitschelt and Steven I. Wilkinson, eds., Patrons,

Clients, and Policies. Cambridge: Cambridge University Press. pp. 141-158.

Latinobarómetro. 2007. Informe Latinobarómetro (November). Accessed March 9, 2009

http://www.latinobarometro.org/.

Lawson, Chappell and James McCann. 2003. “An Electorate Adrift? Public Opinion and the

Quality of Democracy in Mexico”, Latin American Research Review, 38 (3): 60-81.

Lawson, Chappell, et al. 2007. The Mexico 2006 Panel Study.

http://web.mit.edu/polisci/research/mexico06.

Lemarchand, René. 1972. “Political Clientelism and Ethnicity in Tropical Africa: Competing

Solidarities in Nation-Building”, American Political Science Review, 66 (2): 68-90.

Lemarchand, René and Keith Legg. 1972. “Political Clientelism and Development: A Preliminary

Analysis”, Comparative Politics, 4 (2): 149-178.

Levitsky, Steven. 2003. “From Labor Politics to Machine Politics: The Transformation of Party-

Union Linkages in Argentine Peronism, 1983-1999”, Latin American Research Review,

38 (3): 3-36.

Lomnitz, Larissa Adler. 1982. “Horizontal and Vertical Relations and the Structure of Urban

Mexico”, Latin American Research Review, 16 (2): 51-74.

Lomnitz, Larissa Adler. 1988. “Informal Exchange Networks in Formal Systems: A Theoretical

Model”, American Anthropologist, 90: 42-55.

Mair, Peter and Ingrid Van Biezen. 2001. “Party Membership in Twenty European Democracies,

1980-2000” Party Politics 7 (1), 5-21.

45

Mauss, M. 1990. The Gift. New York: W. W. Norton.

McCabe, Kevin A., Daniel Houser, Lee Ryan, Vernon Smith, and Theodore Trouard. 2001. “A

Functional Imaging Study of Cooperation in Two-Person Reciprocal Exchange,”

Proceedings of the National Academy of Sciences, 98: 11832-11835.

McCabe, Kevin A., Mary L. Rigdon, and Vernon L. Smith. 1996. “Game theory and reciprocity in

some extensive form experimental games”, Proceedings of the National Academy of

Sciences, 93(23): 13421–13428.

McCabe, Kevin A., Mary L. Rigdon, and Vernon L. Smith. 2003. “Positive reciprocity and

intentions in trust games” Journal of Economic Behavior and Organization 52 (2):267-75.

Nichter, Simeon. 2008. “Vote Buying or Turnout Buying? Machine Politics and the Secret

Ballot”, American Political Science Review 102 (February): 19-31.

Ostrogorski, Moisei. 1910. Democracy and the Party System in the United States: A Study in

Extra-Constitutional Government. New York: Macmillan.

Panebianco, Angelo. 1988. Political Parties: Organization and Power. Cambridge: Cambridge

University Press.

Pfeiffer, Silke, Frederic C. Schaffer, Claudio Weber Abramo, and Leslie Busby. 2004. “Vote

buying” in Global Corruption Report 2004. London: Transparency International. Ch. 5.

Powell, John Duncan. 1970. “Peasant Society and Clientelist Politics”, American Political

Science Review, 64(2): 411-425.

Radnitz, Scott. 2010. Weapons of the Wealthy: Predatory regimes and Elite-Led Protests in

Central Asia. Ithaca, NY: Cornell University Press.

Rao, Vijayendra. 2001. “Poverty and Public Celebration in Rural India”, Annals of the American

Academy of Political and Social Science. Culture and Development: International

Perspectives, 573: 85-104

Roniger, Luis. 1990. Hierarchy and Trust in Modern Mexico and Brazil. New York: Praeger.

46

Roniger, Luis and A. Günes-Ayata, eds. 1994. Democracy, Clientelism, and Civil Society.

Boulder, Colorado: Lynne Rienner.

Rossetti, Carlo. 1994. “Constitutionalism and Clientelism in Italy”, in Luis Roniger and A. Günes-

Ayata, eds., Democracy, Clientelism, and Civil Society. Boulder, CO: Lynne Rienner.

Sanfey, Alan G., James K. Rilling, Jessica A. Aronson, Leigh E. Nystrom, and Jonathan D.

Cohen. 2003. “The neural basis of economic decision making in the ultimatum game”,

Science, 300: 1755-1757.

Schaffer, Frederic and Andreas Schedler. 2007. “What is Vote Buying?” in F. Schaffer, ed.

Elections for Sale: The Causes and Consequences of Vote Buying. Boulder, CO: Lynne

Rienner, pp. 17-30.

Schaffer, Frederic, ed. 2007. Elections for Sale: The Causes and Consequences of Vote

Buying. Boulder, CO: Lynne Rienner.

Scheiner, E. 2007. “Clientelism in Japan: The Importance and Limits of Institutional

Explanations,” in Herbert Kitschelt and Steven I. Wilkinson, eds., Patrons, Clients, and

Policies. Cambridge: Cambridge University Press, pp. 276-297.

Scott, James C. 1972. “Patron-Client Politics and Political Change in Southeast Asia,” American

Political Science Review, 66: 91-113.

Scott, James C. 1976. The Moral Economy of the Peasant; Rebellion and Subsidence in

Southeast Asia. New Haven: Yale University Press.

Simmel, Georg and Kurt H. Wolff. 1950. The Sociology of Georg Simmel. Glencoe: Free Press.

Stokes, Susan 2005. “Perverse Accountability: A Formal Model of Machine Politics with

Evidence from Argentina” American Political Science Review, 99(3):315-325.

Stokes, Susan. 2007. “Political Clientelism.” In C. Boix and S. Stokes (ed.) Oxford Handbook of

Comparative Politics. Oxford: Oxford University Press.

Trivers, Robert L. 1971. “The Evolution of Reciprocal Altruism”, Quarterly Review of Biology, 46

47

(1): 35-57.

Wang, Chin-Shou. 2007. “Dilemmas of Electoral Clientelism: Taiwan, 1993”, International

Political Science Review, 28(2): 225-245.

Wantchekon, Leonard. 2003. “Clientelism and Voting Behavior: Evidence from a Field

Experiment in Benin”, World Politics, 55 (3): 399-422.

Weinstein, Eugene A., William L. DeVaughan, and Mary Glenn Wiley. 1969. “Obligation and the

Flow of Deference in Exchange”, Sociometry, 32 (1): 1-12.

Yan, Yunxiang. 1996. The Flow of Gifts: Reciprocity and Social Networks in a Chinese Village.

Stanford: Stanford University Press.

Yiannakis, Diane Evans. 1981. “The Grateful Electorate: Casework and Congressional

Elections”, American Journal of Political Science, 25 (3): 568-80.

Van de Walle, N. 2007. “Meet the new boss, same as the old boss? The evolution of political

clientelism in Africa,” in H. Kitschelt and S. Wilkinson, eds., Patrons, Clients, and

Policies. Cambridge: Cambridge University Press, pp. 50-67.

Vicente, Pedro C. 2008. “Is Vote Buying Effective? Evidence from a Field Experiment in West

Africa”. Unpublished manuscript, accessed March 4, 2009

http://faculty.ucr.edu/~jorgea/econ261/africa.pdf.

Vicente, Pedro and Leonard Wantchekon. 2009. “Clientelism and Vote Buying: Lessons From

Field Experiments in African Elections” Oxford Review of Economic Policy 25, 2: 292–

305.

48

Supporting Materials

In these supporting materials, we present details on the Mexico 2009 and Mexico 2010

Clientelism surveys as well as supplemental analyses that are referred to in the main text.

1. Mexico Clientelism surveys

2009 Mexico Clientelism Survey

The 2009 survey consists of 545 completed interviews with Spanish-speaking Mexican adults,

in six purposively selected sites in the Federal District (Mexico City) and Mexico State (including

areas within the greater Mexico City metropolitan area). Areas were selected for high levels of

electoral competition between different parties (PRI-PAN, PAN-PRD, PRD-PRI, and PRI-PAN-

PRD). Interviews were conducted during September 5-11, 2009 and September 19-21, 2009 by

the firm Data Opinión Pública y Mercados. Eighty percent of interviews were supervised (70% at

the time of the interview, 10% afterward).

2010 Mexico Clientelism Survey

The 2010 survey consisted of 352 completed interviews with adult, Spanish-speaking Mexicans

at four purposively chosen points in the Mexico City metropolitan area. Areas were selected for

high levels of electoral competition between different parties (PRI-PAN, PAN-PRD, PRD-PRI),

with a focus on lower-middle class or poor neighborhoods where some vote-buying is likely.

Interviews were conducted during August 2010 by the firm Data Opinión Pública y Mercados.

Items in the 2010 survey were similar to those in 2009, with the exception that certain

experimental questions that failed to generate an adequate response (e.g., a “blank ballot

question asking respondents to name the price they would charge for their vote) or were judged

confusing by interviewers in 2009 were excluded. As in 2009, 80% of interviews were

supervised (75% at the time and 5% afterwards); to avoid frightening residents in high-crime

49

zones, in some cases the supervisor was stationed in site of the interview but was not physically

next to the interviewer during the interview.

Typical polling sites for both are shown below:

50

2. Comparison of treatment groups on variables of interest

Because the surveys were paper-and-pencil, automated random assignment of

interviewees to treatment groups was not possible. Randomization was accomplished by

interspersing the survey forms that were then used by interviewers. Because a small number of

interviews were not completed and because interviewers could have inadvertently reordered the

questionnaires, we performed the following ex post checks to verify that randomization “worked”

on the spit-sample items referred to in the text (all from the 2009 survey).

Table A1. Treatment versus Control Groups I

Candidate offers Gabriel/a 50 pesos vs. 500 pesos for his/her voteVariable Group A Group B

Mean Std Dev Mean Std DevAge 44.7 16.3 43.8 17.1Self-reported insufficiency of household income 2.83 .89 2.71 .82Percent female 57% -- 58% --Skin color (interviewer-coded); higher = darker 2.43 .78 2.42 .80Education 5.28 2.44 5.06 2.36Social class (interviewer-coded, inverse) 3.90 .85 3.94 .87Percent with refrigerator in home 38% 34%Percent voted in July 2009 midterm elections 71% 71%Efficacy (% saying politics not too complicated) 53% 53%Believes vote is secret 76% 79%Attends church once per week or more .39 .49 .43 .50Note: N = 545. No differences significant at p < .1 except income (p = .09).

In this split-sample experiment, Group B (in which Gabriel received the 500 peso offer) scored

slightly higher self-reported insufficiency scale (meaning they felt more financially squeezed). If

income affects willingness to accept a clientelist payoff, this disparity in income would tend to

increase the differences between the two group (however slightly). Group B did not score

significantly lower on measures of social class or household item ownership. Because income

insufficiency was asked later in the questionnaire and was subjectively self-reported, a

treatment effect on income (i.e., asking about a higher peso amount subsequently affected

respondents’ views of their own income insufficiency) cannot be ruled out. 51

Table A2. Treatment versus Control Groups II

Retrospective voting: Will construct medical clinic vs. did construct medical clinicVariable Group A Group B

Mean Std Dev Mean Std DevAge 45 16.8 43 16.6Self-reported insufficiency of household income 2.75 .82 2.79 .89Percent female 58% 58%Skin color (interviewer-coded); higher = darker 2.41 .78 2.44 .80Education 5.01 2.37 5.33 2.43Social class (interviewer-coded) 3.89 .87 3.9 .84Percent with refrigerator in home 37% 32%Percent voted in July 2009 midterm elections 71% 71%Efficacy (% saying politics not too complicated) 52% 54%Believes vote is secret 79% 76%Attends church once per week or more 42% 40%Note: N = 545. No differences significant at p < .1.

Table A3. Treatment versus Control Groups III

Civic values experiment: receives civic prime first vs. anti-civic prime firstVariable Group A Group B

Mean Std Dev Mean Std DevAge 45.1 16.5 43.4 16.9Self-reported insufficiency of household income 2.83 .86 2.71 .85Percent female 57% 59%Skin color (interviewer-coded); higher = darker 2.45 .76 2.40 .82Education 5.03 2.40 5.31 2.40Social class (interviewer-coded) 3.92 .83 3.92 .88Percent with refrigerator in home 35% 34%Percent voted in July 2009 midterm elections 72% 70%Efficacy (% saying politics not too complicated) 51% 55%Believes vote is secret 79% 77%Attends church once per week or more 42% 42%How disappointed R would be in friend who sold his vote (inverse scale) 1.53 1.45 1.58 1.54

Percent who would be very disappointed in a friend who sold his vote 79% 81%

Note: N = 545. No differences significant at p < .1.

52

3. Supplemental Analyses

Obligation

On page 16 in the main text, we argued that “survey data from Brazil show that citizens

are willing to carry through on vote-selling transactions, even when the questions asked do not

imply any policing of voters’ behavior.” In the 2002 Brazil Election Study, the relevant question

read:

I am going to read you various scenarios and would like you to say what a person SHOULD do something and what a person WOULD do.

i. A candidate offers a handicapped person a wheelchaira. Accept the wheelchair and vote for the candidate, ORb. Not accept the wheelchair and vote for another candidate

ii. A candidate offers a bag of groceries to a very poor and hungry family…

iii. A mother cannot get a place for her child in school. A candidate obtains a place for the child…

iv. A candidate offers a mother with a sick son money for his medical treatment…

v. A candidate offers a truckload of bricks for several families to finish building their homes…

vi. A candidate offers to remodel a soccer field for a group of friends who get together to play soccer each week…

vii. A candidate offers a bicycle for a child…

Respondents were asked whether the person should (or would) accept the gift and vote for the

politician in question. They were thus not given the option of accepting the gift and not

supporting the party who gave it, nor were they allowed to assume that the politician might

renege on giving the gift if the prospective voter accepted. The results are reported in Table A1.

53

Table A4: Feelings of Obligation in Brazil

Hypothetical recipient is..

Benefit offered is…

Should accept the giftand vote for candidate

Would accept the giftand vote for candidate

Yes NoDon’t know Yes No

Don’t know

Handicapped person

Wheelchair 40% 54% 6% 76% 17% 7%

Poor family Groceries 47% 48% 5% 81% 13% 6%

Mother Spot in school for her child 52% 43% 5% 82% 13% 5%

Mother Medical care for her child 61% 35% 4% 86% 10% 5%

Family Truckload of bricks for homes 46% 50% 5% 80% 15% 6%

Group of friends

Refurbish a soccer field 39% 56% 5% -- -- --

Parents Bicycle for their child 30% 65% 5% -- -- --

In Table A4, all differences are significant at the 1% level for whether recipients should accept

the gift and vote for the candidate, except the differences between “Groceries” and “Truckload

of bricks” and “Wheelchair” and “Refurbish a soccer field”, which are significant at the 5% level.

For whether recipients would accept the gift, all differences are significant at the 1% level except

the difference between “Groceries” and “Spot in school” (p = .97). All differences between

“would” and “should” answers are significant for each question at the 1% level. (N = 2,513.)

A similar battery of questions, mentioned on p. 11 of the text, was asked in the

Mexico 2006 Panel Study (Lawson et al. 2007). The question (from Wave 3 of the panel

survey) read:

(INTERVIEWER: USE “Gabriel” IF MALE INTERVIEWEE AND “Gabriela” IF FEMALE INTERVIEWEE) Let’s imagine that there is a person named Gabriel/Gabriela who is a person like you and lives in a community like yours. A representative of a political party ... (READ OUT SCENARIOS) In your opinion, how much of an obligation should Gabriel/Gabriela feel to vote for this political party – a lot, some, a little, or none?

a. throws a party for Gabriel(a)’s community

54

b. gives money to Gabriel(a) to buy a week’s groceries c. promises to build a sewage system in Gabriel(a)’s community if he wins

the election in that neighborhood

Because the panel structure of the sample was complex, and some questions were only

included in some waves, we report separately the results for different subsamples.

Table A5: Feelings of Obligation in MexicoCross-section that accompanied Wave 2 (N = 305)Obligation induced by… A lot Some Little None Don’t know MeanNeighborhood party 23.6 17.4 10.8 40.3 7.9 1.3Money for groceries 28.5 16.7 8.5 39.3 6.9 1.4Paving streets (retrospective)

28.9 18.7 12.1 33.4 6.91.5

Sewage (prospective) 31.8 18.7 13.1 29.5 6.9 1.6Employment (prospective) 32.8 19.9 9.2 32.5 6.2 1.6

Panel Wave 3, all respondents including Mexico City and rural oversamples (N = 1,594)Obligation induced by… A lot Some Little None Don’t know MeanNeighborhood party 17.5 14.7 10.9 50.8 6.1 1.0Money for groceries 24.0 12.6 8.6 48.7 6.1 1.1Sewage (prospective) 25.5 18.4 10.8 39.6 5.7 1.3

Panel Wave 3, national sample only (N = 1,067)Obligation induced by… A lot Some Little None Don’t know MeanNeighborhood party 17.4 15.1 11.5 51.7 4.2 1.0Money for groceries 24.1 13.0 8.9 49.1 4.9 1.1Sewage (prospective) 25.5 19.5 11.0 39.9 4.1 1.3

In Table A5, means are a weighted average of the responses, excluding “don’t know” answers,

where “a lot” = 3 and “none” = 0. Differences in means within each sample in the third wave are

statistically significant at the 1% level. For the Cross-Section in Wave 2 (first set of rows), all

differences are significant at the 5% level or better and three are not significant (“Neighborhood

party” and “Money for groceries”, “Money for groceries” and “Paving streets”, and “Sewage” and

“Employment”).

Aside from the surveys mentioned in the text, we are unaware of any surveys from other

countries that contain items which could be used for this sort of analysis.

In p. 11 of the text, we report that: “ the more recipients said that the gifts mattered to

them, the greater the obligation they felt to support the party providing it (r = .51, p < .01, N = 55

79).” Data are taken from the third wave of the Mexico 2006 Panel Study, in which the following

series of questions was included:

Over the last weeks, has a representative of a political party or candidate given you a

gift, money, food, subsidy or any other type of help?

Which party or candidate was it? (INTERVIEWER: AFTER EACH, ASK) What did

they give you? (INTERVIEWER: MARK ALL MENTIONS)

Would you say this gift or assistance from the party or candidate mattered a lot to

you, mattered somewhat to you, or didn’t matter to you?

In appreciation of the gift or assistance, how obliged would you say you felt to vote for

the candidate who gave it -- very, somewhat, a little, or not at all?

The full results from the panel, which includes the national sample plus the rural and Mexico

City oversamples (N = 1594), are shown in Table A6 (below). Rows do not necessarily sum

to the total because four respondents received more than one type of gift from a party.

Table A6: Gifts from parties and candidates in Mexico 2006 Panel StudyNumber of

respondents receiving

gifts

Number saying gift mattered “some” or “a

lot”

Number feeling “some” or “a lot” of an obligation

to support the party giving the gift

Money 6 2 1Meals (alimentos) 1 0 0Bag of groceries (despensa) 25 13 5Token gift (obsequio) 43 7 5Other good or service 10 3 2Don’t remember / Didn’t answer 3 3 1Total number of respondents receiving at least one gift 84 28 14

Effect of civic values

On p. 20-21 of the text, we argued that respondents in Brazil “who more strongly favored

democracy as a system of government were significantly more likely to believe that voters

should not accept selective benefits in return for their vote.” To test this claim, we used data

56

from the above items for Brazil (from the 2002 Brazil Election Study); the dependent variable is

an additive index of responses to whether people should accept the following benefits in

exchange for their vote; it is measured on an eight-point scale.

The statistical model includes a series of explanatory variables: education; age; income

as measured by the log of the quantity total monthly household income divided by the number of

people in the household (the scale ranges approximately from .6 to 10); opinion of democracy

as measured by how much the respondent agreed or disagreed with the following statement (on

a five-point scale): “Democracy may have its problems, but it is better than other forms of

government”; and an index of “Rouba-mais-faz” which is an additive sum of responses to a

series of items on how much corruption the respondent appeared willing to accept in public life

(Q105a-k); it ranged from 1 to 45.

The second column of Table A7 shows the results of OLS regressions with p-values in

parentheses. The results hold when the “Rouba-mais-faz” index is excluded. [Data not shown.]

Table A7: Voters’ Orientations toward Clientelism in Brazil and MexicoVariable Brazil 2002 Mexico 2006Education level .24

(.01).11

(.00)Income .18

(.01).01

(.45)Age range -.13

(.01)-.04(.05)

Support for democracy .46(.00)

.20(.00)

“Rouba-mais-faz” index .07(.00)

--

N 1,882 1,326R-squared .13 .08

Second, we argued that “respondents in the Mexico 2006 Panel Study who

expressed support for democracy as a system of government were more likely to reject

the notion that a third person (“Gabriel”) should be willing to trade his vote for groceries.”

To test this claim, we used data from the Mexico 2006 Panel Study (Lawson et al., 2007)

57

described above. Specifically, we constructed an index using a four-point scale of

responses to the question about how much obligation a third person should feel to vote

for a party that had provided him or her with groceries for the week. That is, only

answers to the second item (b) in the battery were used; however, the results are similar

for an additive index of responses to all three questions. [Results not shown.]

The explanatory variables are similar to those used in the Brazil model and include

education, age, income (measured by in a ten-point scale),21 and opinion of democracy as

measured by how much respondents agreed or disagreed with the statement: “For me it is very

important to live in a democracy” (also on a five-point scale). The third column of Table A4

shows the results of OLS regressions with p-values in parentheses.

Aside from the surveys mentioned in the text, we are unaware of any surveys from any

other countries that contain items which could be used for this sort of analysis.

21 The non-effect of income in Mexico persists in a range of specifications and with different

proxies for living standards. [Data not shown.]58