A declining playing field for the patrons?

A declining playing field for the

patrons?

The presence of social capital as a predictor for clientelism in Lati n-America

Radboud University

Supervisor: Dr. Saskia P. Ruth-Lovell Word count: 15.429

Abstract

Some Latin-American countries are perfect examples of countries that transitioned to democracy and encountered clientelism. Research indicates that clientelism consists of actions taken by both patrons and clients on the basis of personal, economic and societal conditions. An often-heard counter to clientelism is social capital which, in contrast to clientelism, builds horizontal relations among people. This study aims to determine how social capital can serve as a predictor for the absence of clientelism. Building on existing work in the field of clientelism, it asks: To what extent has social capital been a predictor for the absence of clientelism in Latin-American countries? In this research social capital has been defined through civil society participation and clientelism has been defined through vote buying, the difference between allocating particular or public goods by patrons and the linkages that parties have to the people. The methods that are used are a linear mixed model to address the time-series cross sectional nature of the data. The data that is used is longitudinal data of 18 different countries measured over time on several variables. Analysis of several models indicates that social capital does have an impact on the absence of clientelism. However, the results for the three dependent variables differ from each other in their predictability.

Contents

1. Introduction...5

2.Theoretical Framework...9

2.1 Clientelism...9

2.2 Five criteria of clientelism...9

2.2.1 The dyadic relationship...10

2.2.2 The asymmetry of the relationship...10

2.2.3 Personal and enduring relationships...11

2.2.4 Reciprocity...11

2.2.5 Voluntary engagement...12

2.3 Opportunities and challenges for clientelism...13

2.4 Social capital...15

2.4.1 The social structure...16

2.4.2 Opportunity, motivation and ability...17

2.4.3 Benefits of social capital...17

2.4.4 Risks of social capital...18

2.4.5 The contingencies and value of social capital...19

2.4.6 Conclusion...19

2.5 Clientelism and social capital...21

2.6 What fosters what?...23

2.7 Conclusion...27 3. Research Method...28 3.1 Expert surveys...28 3.2 Case selection...29 3.2.1 Dependent variables...30 3.2.2 Independent variable...32 3.2.3 Control variables...33 3.3 Estimation...34 4 Analysis...36

4.1 The null model...36

4.2 Adding independent variables to the model...38

4.3 Adding the time variable to the model...38

4.4 Analyzing the output...38

5 Conclusion...40

1. Introduction

Clientelism is a concept that has been around for many centuries and has taken multiple forms in its evolution over time (Kettering, 1988). The concept involves both a patron and a client in which the patron makes an offer to the client in return for political support. The support to the patron often exists of a vote for that patron. For the patron it thus is clear-cut; you either support me or you don’t. For the client you can think of multiple rewards for giving political support. These rewards range from food, shelter, clothes or money to for instance protection. What people get in return is often not of interest to scholars, but the underlying circumstances are.

Many scholars have argued that clientelism takes place because patrons can offer their clients resources for survival. Hilgers (2009) defined these clients as one of the three types of clients in her research. The second category of people are able to move beyond these utilitarian relationships and will therefore start to doubt and neglect clientelism as it prevents them from making their own political decisions. A third group is described as a group of people which find that clientelism can help to build communities and teach people democratic values as people that have never engaged in democracy are made aware of the political process.

Thus, what is it that makes ordinary people choose to engage in this clientelist behaviour? What are the predictors and non-predictors of why certain nations encounter more clientelism than other nations? Are these predictors solely based on e.g. people’s living standards, their opportunities or their interest in politics? Or do we encounter more clientelism because of general problems in society, (political) institutions or civil organization?

Simeon Nichter (2018) describes a range of challenges to electoral clientelism which are possible explanations for the absence of clientelism in a country. The institutional challenges are ballot secrecy, compulsory voting and anti-clientelism legislation. Another challenge to clientelism is the shift in focus of partisan strategies to a more program-oriented approach instead of patronage activities. As Nichter finds that electoral clientelism is in fact challenged by these changes, relational clientelism is rather elusive to these challenges. Relational clientelism is a type of clientelism that moves beyond campaign periods and is therefore better able to withstand the challenges that electoral clientelism faces. However, it does face other challenges too such as economic development; a clientelist relation mainly remains attractive to people to the point that they can provide for themselves. Nevertheless, economic development is mainly a problem for electoral clientelism since letting people buy your vote becomes less attractive when finding yourself in a better economic position (Nichter, 2018).

The economic argument is an interesting argument which can be described as a rise of the opportunity costs. The idea of this is that people gain more opportunities to choose their own path as their dependency on a patron for economic survival falls away. Thus, the opportunity costs for the patron to recruit a person into a clientelist relationship increase. Opportunity costs can also be heightened by a substantial raise in awareness of the presence of clientelism as well as through awareness of the negative consequences of clientelism (Weitz-Shapiro, 2012). Awareness and opportunity costs also link back to the second category of people described by Hilgers (2009) who doubt clientelism as they are able to move beyond utilitarian relationships.

A more structural approach to whether or not clientelism takes place in certain societies can be found in Robert Putnam’s (1993) ‘Making Democracy Work’. This piece on Italy holds the argument that a strong history of civic organization and a strong civil society are predictors of whether or not clientelist networks get a foothold. Putnam (1993) argues that there is a trade-off between a strong civil society and clientelist networks which reflects in the differences between the North and South of Italy. The North of Italy has a history of civic organization and had therefore a lot of social capital which ultimately led to strong democratic institutions and the absence of clientelism. The South of Italy has been coping with clientelism which made the emergence of civil society almost impossible. The trade-off argument is strongly in line with the idea that clientelist networks are most likely to be closed, nonparticipatory networks which are based on the individual ties to a leader and therefore do not strengthen civil society (Abers, 1998).

Putnam has not been the only scholar that focused its research on why clientelist relations get a foothold in societies. Many scholars have focused their attention on underlying factors that do or do not contribute to clientelism, whereas others stick to the sole notion of clientelism as a utilitarian perspective on the relation between patron and client. Something very striking from the known research on the relationship between clientelism and social capital is the fact that most arguments come from small-scale research done in regions, cities, municipalities or even communities. What the literature is clearly missing is a more general argument which comes from a national perspective. The absence of a research design of aggregated numbers on social capital and its trade-off with clientelism on the national level provides for many research opportunities.

The aim of this research is therefore to look at social capital on the national level as a predictor for the absence of clientelism to find out whether or not we can determine a relation between the two concepts. In order to do so and to find a general conclusion on the possible relation we must focus on multiple units of analysis. This is necessary since a potential relation in one country might not be generalizable to a wider range of countries. What seems to be of high importance to understand this

relation is to investigate this matter in countries that have different backgrounds and levels of social capital. Latin America seems to be a region that has both highly developed countries as well as less developed countries (UNDP, 2019). The importance of this is to be found in Putnam’s (1993) argument that regions with a high level of social capital because of a long-existing civil society are more likely to develop a society that is less vulnerable to clientelism than societies that have a low level of social capital.

Another important thing in this research is that it must be done over time. As Nichter (2018) already stated, vote buying is the most common and the most studied form of clientelism. However, vote buying almost only occurs in times of elections. Taking one time point for each country may only tell us something about social capital and clientelism as two concepts in a country at one time point. What is interesting to find out is the possible effect of social capital on the absence of clientelism over time.

This type of research is relevant for the field since there is a lot of debate on what fosters clientelism. Also, clientelism is sometimes argued to work the other way around so that it fosters social capital (Gay, 1990). What is also very relevant for this research is that it tries to overcome the level on which most of the research on clientelism is done. Most researchers conclude on the basis of case studies in municipalities, cities, or even neighbourhoods. Also, these studies are often done in short periods of time. This research tries to come up with a more general explanation for the presence or absence of clientelism. Another reason why this research is relevant is that clientelism is something that we view as a negative concept; it namely counters democratic values. Therefore, coming up with a general conclusion about what predicts the presence and absence of clientelism could be of enormous societal relevance.

The research question that follows from the thoughts above is the following:

To what extent has social capital been a predictor for the absence of clientelism in Latin-American countries?

In order to find an answer to the research question this thesis will first provide a general overview of the concepts of clientelism and social capital. The concept of clientelism will be discussed along the lines of the five criteria of clientelism by Muno (2010). I have tried to discuss the spectrum of theories in the widest form possible by linking authors to Muno’s criteria. After that, the concept of social capital will be discussed. This is done along parts of a conceptual model of the concept presented by Adler & Kwon (2002)

In the section that follows the discussion of the concepts themselves, the concepts will be linked to each other in order to exemplify that social capital supposedly does not foster clientelism. This is necessary since there is some debate in the literature on the causal direction of the relationship. Some authors argue that, based on small case studies, clientelism can foster social capital through its possibility to create networks. We will show that the networks that clientelism produces are often of a hierarchical nature and therefore do not foster social capital.

After that, a methodological section will be provided to illustrate how this research is carried out. The research is carried out by analyzing time-series cross sectional data in a linear mixed model in SPSS. In this section we will operationalize the variables and talk about specific methodological choices, advantages and shortcomings. The variables that we will discuss logically follow from the theoretical framework in a way that they represent multiple aspects of the concept of clientelism. The three dependent variables in this research are Election vote buying, particularistic vs public goods and party linkages. The control variables for this research are GDP per capita and the clean elections index. These variables will be measured over time and therefore I also introduce a time component, the variable year. I will elaborate on the specific ideas behind these choices in this chapter.

In the fourth section, I will present an initial model of the dependent variables. Then, I will draw on the first model that is presented in this chapter by adding the control variables and the time variable to that model. After investigating the model fit, I will present the empirical results which logically follow from data analysis. The data analysis will be done along the lines of the three hypotheses that will be given at the end of the following section.

Finally, a conclusion on the relation between social capital and clientelism will be provided. In this conclusion I will ask myself what this research contributes and what it tells us exactly. Also, what can this section not tell us perhaps with regard to our modelling choices or case selection.

2.Theoretical Framework

2.1 Clientelism

Clientelism actually comes from the Latin word ‘cluere’, which means ‘to listen and obey’. Also, patrons in ancient Rome were often people that spoke for a group of clients who were connected to this patron through certain codes of conduct and mutual benefit (Muno, 2010). The concept of clientelism thus dates from ancient times and is often confused along other hierarchical concepts that describe relations such as feudalism. It is no surprise that there is sometimes confusion over the meaning of the concept, since there seems to be no general accepted definition of clientelism as an indisputable concept.

Clientelism can simply be defined ‘as the proffering of material goods in return for electoral support, where the criterion of distribution that the patron uses is simply: did you (will you) support me? (Stokes, 2011, p. 2). Scott (1972) thought of clientelism as ‘an instrumental friendship in which an individual of higher socioeconomic status (patron) uses his own influence and resources to provide protection or benefits, or both, for a person of lower status (client) who, for his part, reciprocates by offering general support and assistance, including personal services, to the patron“ (Scott, 1972, p. 92).

Hopkin (2006) defines political clientelism as ‘the distribution of selective benefits to individuals or clearly defined groups in exchange for political support’. Allen Hicken (2011) tries to sketch the debate on what is and what is not clientelism by distinguishing the concept from other concepts such as programmatic policies and noncontingent vote buying. Hicken makes the distinction on the basis of the question whether or not the benefits of an act by the patron are targeted and if they are contingent or not. He ultimately finds that clientelism, under which he includes patronage, clientelist vote buying and clientelist club goods, is distinguishable from those other benefits on the two questions. As the argument by Hicken already suggests there are many ways to define clientelism. Ordinary vote-buying or patronage are the two concepts mostly linked to clientelism as they also are seen as illegal ways to manipulate elections (Schaffer, 2007, p. 6)

2.2 Five criteria of clientelism

Wolfgang Muno (2010) argues that even though there is debate about the concept of clientelism there are five key characteristics to the concept: 1) The relationship is dyadic; 2) The relationship is asymmetrical; 3) The relationship is personal and enduring; 4) The relationship is reciprocal; 5) The relationship is voluntary.

2.2.1 The dyadic relationship

The dyadic relationship that clientelism holds can be exemplified by comparing it to a pyramid. At the top of the pyramid sits the patron. At the bottom of the pyramid are the clients, who have no mutual connection whatsoever. The patron has a dyadic relationship to his clients. The relation is dyadic since it is a relation that based on the individual ties to a leader (Abers, 1998). A dyadic relation between a patron and a client is the most basic form of the dyadic relation in clientelism. There may however be more levels in the pyramid. The pyramid then still exists of a patron at the top but is followed by brokers in the middle of the pyramid. These brokers have both a relation to the patron as well as to the client. The patron instructs the brokers who then interact with the clients. There may be even more levels of brokers in between the client and the patron but what is most important in the clientelist relationship is that all relations between clients, brokers and the patron are dyadic (Muno, 2010). As is illustrated below a patron can have more brokers and a broker can have multiple clients but all the relationships between patron, brokers and clients remain dyadic. The pyramid below is only a simplified illustration of how complex a clientelist relation might get.

The pyramid is in line with the definition of Eisenstadt and Roniger (1984) that also spoke about an unequal balance between the patron and its client in their definition of clientelism, stating it as ‘a form of personal, dyadic exchange usually characterized by a sense of obligation’ (p.48). The pyramid can definitely become a complex organization but what is important is the hierarchical nature of the

pyramid which will also be described in the next paragraph. Patron

Broker Broker

Client Client

2.2.2 The asymmetry of the relationship

The second criterion described by Muno (2010) is the asymmetric relationship between patron and client. This is often also described as a vertical or hierarchical relationship. The asymmetry is ‘not only marked by an asymmetrical distribution of information, it is marked by an asymmetrical distribution of resources, of power, money, goods, prestige, etc. The patron controls material or immaterial resources, the client does not’ (Muno, 2010). Hopkin (2006) adds that clientelism reflects ‘the unequal, hierarchical exchange characteristic of feudal society, in which patrons and clients were tied to durable relationships by a powerful sense of obligation and duty’ (p.2). The unequal nature of the relationship between patron and client cannot be described as a principal-agent relationship since ‘a patron is not an agent of a client, and the client is not the principal of a patron’ (Muno, 2010). Muno counters the argument that a patron is just an agent with better access to information by elaborating on the asymmetric and hierarchical nature of the relation. Hilgers (2011) adds to Muno that clientelism is ‘an exchange in which individuals maximize their interests’ which involves ‘longevity, diffuseness, face-to-face contact, and inequality’ (Hilgers, 2011, p. 568). By inequality Hilgers (2011) underlines once again that clientelism is a relationship between two people of unequal status.

2.2.3 Personal and enduring relationships

According to Muno (2010) ‘patron and client know each other and trust each other’. Kitschelt and Wilkinson (2007) add that predictability and monitoring are two important components of the enduring relationship of clientelism. By predictability it is meant that both actors have confidence in the reciprocity of the exchange. Monitoring is difficult but necessary in a way that it can boost the confidence of both parties in the relationship. But what is most important to the clientelist relationship is that it requires time to establish a personal and trusted relationship between the patron and the client (Muno, 2010). In line with Muno’s argument Hilgers (2011) states that clientelist relations are often a long-term commitment with regular personal interactions in which both the client and the patron understand their obligations.

When looking back at Nichter’s (2018) differentiation between electoral clientelism and relational clientelism the personal part of Muno’s (2010) third criterion seems to hold. However, the enduring part of a clientelist relation might differ when it comes to electoral clientelism versus relational clientelism. Electoral clientelism is often something that only occurs months, weeks or even days before the election which makes the definition of clientelism as an enduring relationship somewhat problematic. According to Nichter (2018), relational clientelism is an ongoing exchange relationship which is more in line with an enduring relationship.

2.2.4 Reciprocity

‘The reciprocal exchange of material or immaterial goods is an essential feature of clientelism’ (Muno, 2010). The exchange is reciprocal since it involves both patron and client to accept the resources of the other and offer their own resources. The type of resources by no doubt differ, since the patron is often able to exchange material (food, shelter, machinery etc.) or immaterial (access to health care, security etc.) support in return for immaterial support by the client in the form of e.g. a vote. It is often however not the case that the exchange of goods happens on the same day, there may be a lag (Lauth, 2004). A reciprocal relationship holds that promises are contingent. This differs clientelism from ordinary political promises which do not tie anyone to a certain action. What is important for the reciprocity of the exchange however is trust in the relationship which can be built by the personal element of the relationship as mentioned in the paragraph above.

2.2.5 Voluntary engagement

This final characteristic of clientelism as a voluntary relationship between patron and client is heavily debated by scholars such as Lauth (2004), who stated that exiting clientelist relationships is often not cost-free and therefore the relationship can be restrictive instead of voluntary. Muno (2010) mentions that in the basis the clientelist relation is a voluntary relation since the choice to engage in such a relation often comes from a rational choice perspective. This rational choice perspective nicely lines up with Hilgers (2009) view that some people simply view clientelist relations in a utilitarian way. According to Muno (2010), force can also bind clients to their patrons. However, force is a one-way possibility in this situation which underlines the asymmetry of clientelist relationships once again. A sole client could never force a patron, but a patron could force a sole client. Another perspective ‘emphasizes the cognitive dimension and identifies the feelings of loyalty, demerit and obligation as crucial factors for keeping patron-client-relations together’ (Muno, 2010).

The main problem that Lauth (2004) has with the voluntary aspect of the relation is however not the entrance in the relation but the exit. Hirschman (1970) finds that especially material necessities and inherited obligations are problematic for the exit opportunities out of a clientelist relationship. Sometimes material necessities and inherited obligations give the client no other option than to stay loyal to the patron even though they would leave the relationship otherwise. Also, the patron cannot simply promise something and later leave that promises unfulfilled. It is not possible to treat clientelist promises as political promises because of the contingency of the relationship.

According to Hilgers, clients are theoretically free to choose their patron and to exit the relationship as well. She notes however that the idea of free entrance and exit is countered by the fact that clientelist relations are often of exploitative nature. Also, it is problematic that a client’s resource

base or his access to alternative patrons is inherent to his possibilities to choose another patron. According to Hilgers (2009), some people simply “must choose between going without much-needed resources and using personal relationships with local politicians, operators and social leaders to obtain housing, services and work. Under these circumstances, the poor tend to consider patron-client exchanges a strategic mechanism for achieving physical and economic security” (Hilgers, 2009, p. 51). Also, the context of clientelism is extremely important since the degree of competition between patrons for clients is often determining whether the relation is either repressive or participatory (Hilgers, 2011).

What is clear is that the exit out of clientelist relations involves high costs both for patrons as for clients. Patrons may ‘lose credibility and legitimacy in case of breaking promises’ (Muno, 2010), whereas clients risk to lose material necessities. Following the fact that clientelism is often seen as a way to manipulate elections it is clear that the concept finds itself in a negative atmosphere. The exploitative nature that Hilgers (2011) describes doesn’t make the concept more likeable since it suggests that there is often no choice for the client to not engage in a clientelist relation.

2.3 Opportunities and challenges for clientelism

The concept of clientelism will likely remain in its negative sphere. However, part of the question of why people enter into clientelist relationships remains unanswered and asks for more research. Is it just the economic needs of people, the fact that they don’t care about politics or the exploitative nature of the clientelist relation that makes people still engage in these practices?

The empiric evidence of Hilgers’ (2009) research in Mexico helps us to think about this matter. Hilgers (2009) distinguishes three categories of citizen’s views on clientelism by using interview data from Mexican citizens. The first category of people embraces the utilitarian perspective and defines clientelism as a necessary way of establishing conditions under which they can survive and take care of themselves (Hilgers, 2009, p.56). The second category is made up of individuals that have a choice in moving beyond utilitarian relations since they have ‘the economic means, education and/or organizational and political experience’ and therefore see clientelism as a negative concept that clashes with their worldview (ibid, p. 52). The third category of people embrace the idea of clientelism as a concept that can eventually help to build communities and teach democratic values such as mobilization. As a result of clientelism, people find that actively participating democratically is something ‘personally fulfilling and important for society at large’ (ibid., p.66). These people started to think about democratic values after being exposed to clientelism.

The utilitarian perspective on clientelism which aligns with Hilgers’ (2009) first category of people is also described by Mariela Szwarcberg (2011) who argues that clientelism is ‘a problem-solving

network. In this network, the flow of goods is conditional upon its political payment in electoral support’ (Szwarcberg, 2011, p. 1). The findings in this article are based on an Argentinian case that shows how the political effectiveness of brokers in problem-solving networks is increased due to their position in non-political networks in society. Whenever a broker is already involved in a problem-solving network in a community and provides clients with certain resources, it is more likely that these clients will also engage in political networks with that broker. We can conclude from this that the establishment of clientelist relations might be more effective whenever a broker or patron already is nestled in a problem-solving network in a society and therefore leaves many clients no choice but to engage also in his political network.

Hilgers’ (2009) second category of views on clientelism consists of the idea that people start to evaluate their own position in society as soon as they can satisfy their basic needs is followed up by Carlin & Moseley (2015). Whereas Hilgers (2009) finds that people that start to think of their own position in society develop certain ideas that make them think that clientelism is wrong, Carlin & Moseley (2015) found that these people indeed participate less in clientelist relations. The reason for their nonparticipation in clientelist relations is their personal involvement in society and especially democracy. Individuals that are already democratically involved tend have more social capital and therefore typically won’t engage in clientelist relations (Carlin & Moseley, 2015).

Another factor that might explain why clientelism takes place and why not is the idea that voter demographics indicate whether or not a society is likely to be targeted by patrons to engage in clientelist relations. The reasoning behind voter demographics is the idea that clientelism has its costs and benefits which are dependent on the voters in a certain society. Clientelism mostly affects the poor people but the extent to which patrons start clientelist relations often also depends on whether or not the non-poor change their vote behavior away from that patron. This is called the electoral costs of clientelism (Weitz-Shapiro, 2012). This idea can be aligned along the second category of individuals as described by Hilgers (2009) that is able to move beyond utilitarian relations with patrons because of their non-poor status and their political experience which the poor unfortunately often cannot.

Weitz-Shapiro (2014) elaborates her argument by reasoning that non-poor voters might reject clientelism because of firstly their view of ‘clientelism as a negative signal of the quality of government performance’ and secondly their rejection of clientelism ‘on moral or normative grounds’ (Weitz-Shapiro, 2014, p. 54). But, how do non-voters in reality find out about clientelism if they are not targeted themselves? As a matter of fact, the costs for clientelism are not activated if the non-poor never find about it. Weitz-Shapiro (2014) argues that there are three ways in which the

non-poor may observe clientelism: ‘direct observation, opposition effort, and the reports of other actors in society’ (ibid., p. 56-57). Society is thus mentioned as a possible predictor for whether or not the audience costs for clientelism are too high and therefore is a possible indirect predictor of clientelism taking place.

As mentioned earlier on, Nichter (2018) found that four specific events caused challenges for electoral clientelism. Shortly, these challenges were ballot secrecy, compulsory voting, anti-clientelism legislation as well as the shift in focus of partisan strategies to a more program-oriented approach instead of clientelist activities. Those events obviously give, and are meant to give, more political autonomy to voters. Despite these challenges to clientelism, citizens remain vulnerable to clientelism. The argument holds the idea that electoral clientelism might slowly decline but relational clientelism is able to deal with the four challenges. Nevertheless, patrons have to deal with the autonomy of their clients more than ever and therefore they start to lean more towards the agency of the client instead of to their own agency. Nichter (2018) therefore makes a strong argument that the demand side of clientelism (the agency of the client) has become more important to patrons and thereby it relates to the second category of people described by Hilgers (2009) that bases its opinion on clientelism and politics on more than just material rewards. When comparing Nichter (2018) to Hilgers (2009) we should note however that Hilgers’ (2009) mainly talks about people that have moved beyond the material needs, whereas Nichter finds that institutional changes and party strategy changes are the most influential threats to clientelism.

As we have seen, the changing context in which clientelism takes place becomes increasingly important. Kitschelt & Wilkinson (2007) argue that clientelism can only take place if the sociopolitical, economic, and cultural context is conducive to it. As Putnam (1993) and Carlin & Moseley (2015) already argued in different ways, social capital as a predictor for the establishment of a clientelist relationship is definitely interesting. When you flip the concepts around, social capital may be a predictor for the extent to which people engage in clientelist relationships.

2.4 Social capital

The concept of social capital is heavily debated since it is a cross-disciplinary concept. Since multiple disciplines have multiple frameworks there is thus disagreement and contradiction in the definitions of social capital (Adler & Kwon, 2002). What is widely accepted as ‘social capital’ is that it is about values of social networks, interactions between diverse people in regard of norms of reciprocity (Dekker & Uslaner, 2001). This definition is underlined by the fact that people have the idea that the goodwill of others is valuable to other individuals or groups (Adler & Kwon, 2002).

As Inglehart (1997) states, social capital is ‘a culture of trust and tolerance, in which extensive networks of voluntary associations emerge’ (1997, p. 188). In relation to that Thomas argues that it is ‘those voluntary means and processes developed within civil society which promote development for the collective whole’ (1996, p. 11). Since there are many perspectives on the concept of social capital it is hard to define it as a single concept. The core intuition of social capital according to Adler & Kwon (2002) is that goodwill that others have towards us is a valuable resource: ‘If goodwill is the substance of social capital, its effects flow from the information, influence, and solidarity such goodwill makes available’ (Adler & Kwon, 2002, p. 18).

Adler & Kwon have created a conceptual model of social capital which can be found in the figure below. The figure discusses the nature, the sources, the benefits and risks and the contingencies that influence the value of social capital. The model will be discussed along the different aspects of it.

(Adler & Kwon, 2002, p. 23)

2.4.1 The social structure

As one immediately notices when looking at the social structure in Adler & Kwon’s (2002) model, we can distinguish three different aspects of the social structure; market relations, social relations and hierarchical relations. Market relations are specific, explicit and symmetrical relations in which goods and services are exchanged for money. Hierarchical relations involve relations that are typically asymmetrical by nature; One obeys authority for material and spiritual security. These hierarchical

relations are marked by their diffuse nature; the person that is being dominated often only knows that they have to obey but not what to obey (Adler & Kwon, 2002, p.19). What is important to note is that these hierarchical relations which speak about dominance are however slightly different from clientelist relations: in clientelist relations it is often specific and explicit what the terms of exchange are between client and patron. Adler & Kwon (2002) distinguish social relations as the third type of relations in a social structure. Social relations are relations in which favours or gifts are exchanged on a symmetrical basis. The terms of exchange are very diffuse since we can do someone a favour without immediately getting a favour back. Nevertheless, due to our symmetrical social relation we do know that the favour will eventually be returned (Adler & Kwon, 2002, p. 19). Out of the social structure as a whole, both task and symbolic contingencies and complementary abilities arise (Adler & Kwon, 2002, p. 23). These two aspects of the model will be discussed in a later section. The next section will focus on the three concepts that arise from social relations in particular; opportunity, motivation and ability (Adler & Kwon, 2002, p. 23).

2.4.2 Opportunity, motivation and ability

‘An actor's network of social ties creates opportunities for social capital transactions’ (Adler & Kwon, 2002, p. 24). Individually there are opportunities to interact with other people and their resources. Collectively there may be opportunities to act together as a whole. To what extent these opportunities arise is heavily debated since many scholars make a difference between small and large social networks. What is commonly accepted however is that the amount to which relations are contingent is often decisive for the amount of opportunities.

Motivations in social relations differ from motivations in for instance market relations since the objective of the relation is often not to get one simple outcome in an exchange. Leana & Van Buren (1999) argue that ‘the sources of social capital lie in trust and associability – the willingness and ability of individuals to define collective goals that are then enacted collectively (p. 542). Motivations seem to be less dependent on contingency but more dependent on norms and trust (Adler & Kwon, 2002, p. 25).

Abilities are the ‘competencies and resources at the nodes of the network’ (Adler & Kwon, 2002, p. 26). Ability is constitutive to social capital in a way that it is impossible to use social ties when people are not on the same level of resources. To illustrate, parents may have a good connection to their child and may be very motivated to help with their child’s mathematics homework. Nevertheless, if they do not have the resources (mathematic skills) the relation will not be considered helpful. According to Adler & Kwon (2002) all three sources of social capital – opportunity, motivation, ability - must be present to activate social capital (p. 27).

2.4.3 Benefits of social capital

There are three main benefits to social capital: 1. Information; 2. Influence, control & power; 3. Solidarity. First, social capital networks broaden the access to information. Improving the access, quality, relevance and timeliness of information can lead to ‘positive externalities for the broader aggregate’ (Adler & Kwon, 2002, p. 29). This can be illustrated by comparing society to a firm. Increasing information about production processes of food will benefit other firms and society as a whole in providing and having access to food.

Influence, control & power simply stands with the idea that power can get things done. Some people in society may have greater power or influence to get things done. These people are however only this influential since they have obligations to other people in society. Power is thus distributed by and spread across society which enables people to get things done (Adler & Kwon, 2002, p. 29).

The third benefit of social capital is solidarity. According to Adler & Kwon (2002) ‘Strong social norms and beliefs, associated with a high degree of closure of the social network, encourage compliance with local rules and customs and reduce the need for formal controls’ (p. 29). Social capital thus boosts solidarity which in turn takes care of strengthening society as a whole and bringing people closer together. Related to the idea of solidarity is the idea that ‘social capital can be defined simply as the existence of a certain set of informal values or norms shared among members of a group that permit cooperation among them" (Fukuyama, 1997). The informal values or norms that Adler & Kwon define as solidarity is also referred to by Robert Putnam (1995). He states that social capital ‘refers to connections among individuals – social networks and the norms of reciprocity and trustworthiness that arise from them’ (Putnam, 1995, p. 19). He claims that civic virtue is closely related to social capital but opts that social capital shows us that civic virtue can be used optimally when ‘embedded in a dense network of reciprocal social relations’ (p. 19).

2.4.4 Risks of social capital

The risks of social capital are based on the benefits that social capital is argued to have. Hansen (1998) argued that information benefits do not always result in the ability to maintain a social relationship, the argument being that the strong relations with information benefits were to costly to maintain. Secondly, Ahuja (1998) argued that the information benefits may also disturb the power benefits of social capital since the more information distribution and ties there are within society the less dependent society will be on one powerful actor making decisions. Thirdly, too much solidarity may cause a loss of rationality and general overview since people have tunnel visions on who and what are good and bad.

Apart from the direct risks regarding the benefits of social capital, prior unequal distributions of assets might cause trouble too. The abuse of social capital by a dominant part of society that has more information and influence at forehand can never be ruled out as a risk. As Adler & Kwon state, ‘there is no invisible hand that assures that the use of social capital resources in competition among actors will generate an optimal outcome for the broader aggregate’ (p. 31).

Both the benefits and risks of social capital know strong arguments. But, generally speaking, the risks will be levelled out by hierarchy and the market. Hierarchy may level out the risks in the form of general laws that are enforced in society. The market may level out the risks as we might establish either stronger or weaker networks and potentially strengthen our own motivation to prevent the risks from actually happening (Adler & Kwon, 2002, p. 32).

2.4.5 The contingencies and value of social capital

Adler & Kwon (2002) argue that the ‘ultimate value of a given form of social capital also depends on more contextual factors’ (p. 32). These factors are the task and symbolic demands and the availability of complementary resources. First, we will discuss task contingencies. Task contingencies at first help us to determine whether we should have strong or weak ties in our social capital networks. This because task contingencies tell us how much our network contributes to the wider objectives of social capital. It will thus help us to value the social capital.

Secondly, task contingencies will help us evaluate to what extent a network should be a closed network or a sparse network that has many structural holes. There seems to be a trade off between the closed and sparse network. At first, a network may be purely informational and therefore needs input from these structural holes whereas a more stabilized and established network gets more value out of dense, cooperative relations (Adler & Kwon, p. 33). The task contingency can thus help us determine what the social network should look like.

Thirdly, the task may determine to what extent the network requires cross-network interactions. When the interactions that are necessary for the task are mainly within the network itself, the value and bonding of the network of social capital increase. When there are many cross-network interactions needed, the bonding of social capital may reduce (Adler & Kwon, 2002, p. 33).

Next up are symbolic contingencies. Symbolic contingencies are for instance norms and beliefs that shape whether or not social capital accepts certain concepts and ideas in a network (Adler & Kwon, 2002, p. 33). For instance, one society might highly value individuals that start a new business out of self-motivation whereas other societies may feel that this provides too much opportunities for one person compared to the task that social capital as a whole was. Symbolic contingencies may decide whether or not a network values something as ‘legitimate’ (Adler & Kwon, 2002, p. 33).

Complementary abilities are abilities and resources that actors gain from other sources than their social capital. It is the ability to generate new ideas from combinations of other ideas not constitutive of social capital (Adler & Kwon, 2002, p. 34).

2.4.6 Conclusion

Recurring in the definitions are notions of trust, informal values, solidarity, contingency, reciprocity, participation, a collective and voluntarism. Almost all these values counter the values that clientelism is often related with. The following section will elaborate on the relationship between the concepts of social capital and clientelism and will shed light on the causal direction of the relationship between the concepts.

2.5 Clientelism and social capital

Clientelism and social capital seem to be concepts that, dependent on the context (Kitschelt & Wilkinson, 2007), interact and possibly shape each other’s challenges and possibilities (Putnam, 1993; Hilgers, 2009; Carlin & Moseley, 2015). This section will provide an overview of different points of view among different authors on the relation between the concepts.

The link between Putnam’s definition of social capital and its relation to clientelism lies in the fact that Putnam sees clientelism as something that ‘restrains civic involvement and inhibited voluntary, horizontally organized manifestations of social solidarity’ (Putnam, 1993, p. 149). In his book ‘Making Democracy Work’, Putnam (1993) argued that a history of civic organization in the North of Italy provided the basis for strong democratic institutions and a lack of clientelism. In the South of Italy however, clientelist networks have made the emergence of a strong civil society almost impossible. This thought is in line with Putnam’s argument that ‘features of social organization, such as trust, norms, and networks, can improve the efficiency of society by facilitating coordinated actions’ (Putnam, 1993, p. 167). At the aggregated country level, we can define social capital as the reflection of the degree of trust and density of associational activity.

According to Putnam (1993), these conclusions evolve into the idea that the presence of a strong civil society entails a lot of social capital which ultimately predicts whether or not an electoral democracy and clientelist relations within a region succeed or fail. However, through links that are tied from the citizens to local leaders, to then federal leaders and then state leaders, Brazil managed to create a civic network based on clientelist practices (Gay, 1990). Thus, there might be some trade-off between the presence of social capital and the presence of clientelism in a region. Putnam’s approach is a structuralist approach that argues that a certain level of civic organization and social capital enables to establish an electoral democracy in which voters would choose in the public interest (Hopkin, 2006). Social capital thus seems to create a mindset among people that is public-interested instead of self-interested. Kitschelt & Wilkinson (2007) also describe the degree to which clientelism can take place in a country from a structural approach. This approach consists of the idea that ‘politicians are only able to monitor clients if the sociopolitical, economic, and cultural context is conducive to it’ (Warner, 2008).

John Durston (1999) linked clientelism to social capital on the basis of clientelisms’ possibility to create social capital. He writes that the introduction of semi-clientelism, a form of clientelism in which patrons ‘further their interests by providing needed services to their clients or gaining their allegiance in a negotiated, shared political cause’ (Durston, 1999, p. 111), weakened authoritarian clientelism in Guatemala. Durston (1999) found that this shift towards semi-clientelism alongside the

fact that local groups started to establish reciprocating relations created social capital. The shift towards semi-clientelism could also be made because there was room for people to start thinking about norms and politics rather than food and shelter. Ultimately this results in a situation in which patrons constantly have to adapt to the emerging social relationships and institutions. This once again indicates the trade-off between social capital and clientelism.

Patron-client relations are outdated according to Chaves and Scholler (2002) since they affect the development and formation of social capital. These scholars compare Northern Italy, on which Putnam (1993) also touched, with an area in Honduras that had similar characteristics leading to social capital because of mass organizational training in the area. They argue that mass organizational training in order to form and develop social capital in an area will affect the way in which patron-client relations develop and might thus eventually deal with clientelism. Similar research results were found by Emma Grant (2001) who on the basis of data from Guatemala suggested that clientelism had a negative effect on social capital.

Auyero (2000) stresses that our negative look on clientelism as a form of exploitation and domination must be countered by the fact that clientelism also constitutes ‘problem-solving networks’ which link clients to patrons and can therefore contribute to the formation of social capital within poor communities. Social capital can link both members within networks or societies (horizontal) as well as members and political parties or patrons (vertical). Therefore, clientelist networks also seem to some extent help in establishing social capital as we have also seen in the argument of Durston (1999).

There is clearly a lot to say about how the concepts of social capital and clientelism are structured, differ and interact. The table below provides an overview of the similarities and differences:

Clientelism Social capital

Similarities  Problem-solving network

 Feature of social organization  Can be both national and local

Differences Relation between patron 

broker  client

Relations between multiple

persons that are

interconnected Asymmetric relation between

unequal people

Symmetric relation between equal people

Problem-solving network based on domination

Problem-solving network based on empowerment No connection between

people on the same or different levels

Centers on the local and national level, but all are connected

Reciprocal relationship based on contingency

Reciprocal relationship based on trust & values

Relation based on conditional support

Relation based on solidarity

The main difference is the fact that social capital is a way more powerful network than clientelism. Both networks are problem-solving networks. Nevertheless, clientelism is mainly about domination and can therefore not serve as a predictor for a network that is about empowerment and interconnectedness (social capital). Even though both Auyero and Durston have argued that clientelism can help to build social capital within communities, most scholars still find that social capital makes clientelism decline and that the presence of clientelism limits the opportunities for creating social capital. The results that Ayuero and Durston also seem never to have been found on the national level which makes the fact that clientelism helps to build social capital on a national level highly unlikely. However, the fact that many scholars based their research on results from societies within countries instead of countries as a whole leaves us with thoughts and opportunities on researching the role of social capital on the country level as a predictor of clientelism. Therefore, this research will focus on the presence of social capital as a predictor for clientelism within countries.

2.6 What fosters what?

Part of this research consists of making a clear distinction between theories that substantiate the idea that social capital fosters clientelism and theories that claim the opposite. Theories that claim that social capital can be established through clientelist relations or through the effect that clientelism has on people’s mindset in social activities are outnumbered against the theories that claim that social capital has a negative effect on clientelism’s existence and effectiveness. However, this is in no way a convincing argument to believe in advance that social capital always has a negative effect on social capital. Also, the fact that both types of theories are solely based on results on the local level leaves us with many open ends concerning this relationship on the national level. Therefore, we must carve out the differences between clientelism and social capital to see at which levels they operate. This will effectively help in strengthening our research question and the endogeneity that the question has within it.

Clientelism and social capital are both structures. However, both differ in their inferences. Clientelism in all its forms seems to mostly take place at the local level. Therefore, clientelism is a macro-level relationship between a patron and a client. This is substantiated by the fact that researchers found and measured clientelism mostly at this local level (Hopkin, 2006; Hilgers, 2009, 2011; Szwarcberg, 2011). One important factor in the essence of clientelism which relates it mostly to the local level is the individual aspect of it. Clientelist relations are often vertical relations between a patron and a client. This client, that can either be one individual or a group of individuals, gains individual benefits from its relation to the client. These benefits will never benefit society as a whole. This is a result of the vertical structure that clientelism is.

Social capital is a concept that is mostly measured at the national level. It is a macro-level variable. Its impact on the presence of clientelism is therefore clear. On the national level clientelism seems to fail to build social capital. This makes it plausible that the results found by Ayuero (2000) and Durston (1999) were outliers that only proved this theory on the local level. The absence of impact of clientelism on social capital on the national level is likely to be caused by the absence of vast national clientelist networks. Social capital is often measured at the national level and is something that ‘refers to connections among individuals – social networks and the norms of reciprocity and trustworthiness that arise from them’ (Putnam, 1995, p. 19). Therefore, social capital is an example of a horizontal structure which can be measured at the national level. What is also very striking is the difference in impact of social capital and clientelism on society. Clientelism tends to only benefit individuals or groups whereas social capital is something of which an entire society or country can benefit. Social capital according to Putnam’s theory is determinative for the presence of clientelism since clientelism can restrain people from engaging in reciprocal relations. There thus is a direct link between social capital on the national level and the absence of clientelism.

A factor that also suggests that social capital rather predicts clientelism than the other way around is the mechanism of opportunity costs. Opportunity costs are costs that the patron has to make to ensure that people engage in a clientelist relation. Social capital, and social capital organizations raise awareness of the presence of clientelism and the seemingly negative effects clientelism can have on elections and therefore on society. Thus, the more social capital, the more awareness of clientelism. This awareness can raise the opportunity costs which can have a negative effect on clientelism (Weitz-Shapiro, 2012). This is another indication of the negative impact of social capital on clientelism.

Both the fact that clientelism and social capital differ in their structure (vertical/horizontal) and the fact that clientelism only has individual benefits whereas social capital has societal benefits indicate

that it is more likely to find an impact of the national level variable (social capital) on the local level variable (clientelism). The macro-level variable (social capital) is more relevant for society as a whole since the macro-level variable probably determines the micro-level variable instead of the other way around. The argument about opportunity costs also follows this reasoning. At last, it is highly unlikely that a network that is characterized by domination (clientelism) fosters a network that is characterized by its empowering nature (social capital). All added together, this reasoning substantiates the direction in which the research question is phrased and thus follows the theories that find that social capital has a negative effect on the presence of clientelism.

To sum up and conclude our mechanism we must distinguish the different levels that these concepts appear on. In social sciences we can analyze both macro-level and micro-level relations. Macro-level relations are the aggregated whole of the micro-level interactions. In the specific case of clientelism and social capital, the macro-level and the micro-level differ from and interact with each other. We assume that the relationship between social capital and clientelism is a relationship on the macro-level. Nevertheless, we assume that the presence of social capital on the macro-level influences the behavior of individuals on the micro-level. The model below represents the different levels, interactions and assumptions.

Our macro-level association is thus that an increase in social capital will eventually lead to a decrease in clientelism. The situational mechanism that affects the micro-level actors is the fact that there is more social capital. The actors on the micro-level, namely individuals and patrons, will change their course of action which is represented in the action-formation mechanism. This will presumably lead to less clientelist relations. The transformational mechanism represents the aggregation of choices of both individuals and patrons to engage less in clientelist relations. This will lead to less clientelism on the macro-level.

- Clientelism

Individuals _{Less clientelist}

The situation being that there is more social capital at one point is presumably followed by two actions. Those actions are the hypotheses of this research which are based on the dependent variables that will be further discussed in chapter 3:

1. If there is more social capital, there will be less vote buying. 2. If there is more social capital, there will be more public goods.

3. If there is more social capital, party linkages will be more focused on programmatic policies.

The first hypothesis is backed by the idea that social capital may increase the chances of an opt-out for clients due to a raise in awareness of the negative aspects of clientelism. We assume this because social capital creates a new problem-solving network for the poor. The poor that were once vulnerable to clientelism due to their inability to move beyond clientelist relations as a result of their economic situation are now connected in a social capital network. The social capital network, as we have seen, is a network of empowerment which is assumed to strengthen the poor and society as a whole and therefore counters clientelism. Also, clientelism is likely to be countered by the increase of visibility. Therefore, vote buying may become less of an issue due to the presence of social capital.

This brings us to the second hypothesis. The increase of visibility of clientelism due to e.g. reports by social organizations increase the possibility that a patron will opt-out of clientelist relations. Opting-out then means that his focus will shift more towards providing public goods instead of particularistic goods. The clientelist relations may hurt his position because of the visibility and the aversion to the patron this visibility evokes. The idea behind the second hypothesis links with the third hypothesis. The third hypothesis is based on the idea that patrons will opt-out of clientelism due to visibility and fear for loss of position. Therefore, the patron will focus more on programmatic policies instead of direct linkages to people.

Both the actions that individuals as well as patrons take because of the increase of social capital will eventually result in less clientelist relations. The idea is that the situation has changed in the fact that there is more social capital which eventually influences the actions that both patrons and individuals make. All these actions combined that both individuals and patrons make on the micro-level will eventually, when aggregated to the macro-level, lead to a decrease in clientelism.

2.7 Conclusion

The arguments that find clientelism to be a predictor for the presence of social capital such as Gay’s (1990) findings are definitely interesting in investigating the relationship between social capital and clientelism. The main reasoning against these arguments however is the fact that these findings never seemed to be substantiated by large studies on the national level. This makes the assumption that clientelism fosters social capital problematic. Also, the benefits that people or groups gain from

the vertical relation to their patron are likely not benefits for society as a whole. Social capital is something that an entire society can benefit from. This makes it particularly interesting to investigate the impact of social capital on clientelism on the national level since we have determined that the main differences between the two structures are the level on which they operate (macro/micro-level) and the direction in which relations are shaped within the structure (horizontal/vertical). Therefore, this research will focus on the impact of the macro-level and horizontal nature of social capital networks on the micro-level and vertical network of clientelism. The reasoning that the relation between the two networks is shaped in this way is substantiated by the argument about rising opportunity costs for clientelist networks in a society with much social capital.

3.

Research Method

In this thesis I use a quantitative approach to investigate the research question. The choice for a quantitative approach comes from the nature of the research question that is about social capital in different countries within Latin-America as a possible predictor for the extent to which clientelism functions over time. The quantitative design comes also out of own interest, since many studies before have focused on small areas, municipalities or cities within countries in a short period of time (Durston, 1999; Chaves & Scholler, 2002; Grant, 2001). This also indicates that there is a lot of space for possible further research on this subject. This section is divided into multiple parts. The first part is a general overview of the motivation for the case selection. The second part contains the operationalization of the dependent variable. After that, the independent variable will be discussed. Finally, I will discuss the control variables. The section also provides motivations for the choices in data sources and the choice of the research method. Furthermore, this section will discuss the descriptive statistics of the variables and also elaborates on possible shortcomings and problems with the data. Before discussing the three sections I first briefly discuss expert surveys, their advantages and their shortcomings.

3.1 Expert surveys

Since this research uses data from the Varieties of Democracy (V-Dem) dataset it should also discuss the data gathering methods that V-DEM uses. V-DEM uses at least 5 country experts per survey cluster. They use a total of 13 general surveys divided in five survey clusters in which the country indicators are subdivided. V-Dem suggest to the Country Experts that they code at least one cluster. Most of the time the amount of Country Experts is higher than five since over 10 people provide ratings for each country (V-Dem, 2020, p. 16). When Country Experts code they use a specially designed online survey. In order to ensure that every country can be coded adequately, V-Dem provides the questions in six different languages. This enables non-English speakers and writers to code in their own language and thus makes the availability of valid data more likely. This solution to increase validity of the judgments by the experts is also proposed by Benoit & Wiesehomeier (2009, p. 504).

A problem with expert surveys is their validity. Ultimately the validity of the benchmark derives from judgments by political specialists. However, ‘because the political quantities or qualities we are attempting to characterize can never be observed directly, we will never have access to any confirmed, “true” external measures against which to benchmark our findings from expert judgments’ (Benoit & Wiesehomeier, 2009, p. 503). Identifying the right experts for the right country and field of expertise can be difficult however. Benoit & Wiesehomeier (2009) suggest that the typical expert for a country is someone who specializes in a certain topic in her/his country (p. 506). V-Dem seems to do this by inviting experts from countries and by dividing the surveys into clusters so that each expert gives expertise about his field of profession. In fact, V-Dem closely approaches the exact same method that Benoit & Wiesehomeier (2009) suggest by recruiting their experts on the basis of five recruitment criteria. First, they identify a list of potential Country Experts per country (100-200 typically per country) after which regional managers of V-Dem find out whether someone is competent to code or not. Secondly, V-Dem wants that at least three out of five (or 60%) of the country experts are nationals or residents of the country they want to be an expert for. This ensures that there is a connection to the country that the experts code. Thirdly, the regional managers determine whether or not an expert has the time and willingness to devote to the coding. Fourth, V-Dem ensures that all coders are impartial and will therefore not recruit any coders with strong political backgrounds or personal motives. Finally, V-Dem aims to have a diverse pool of professional backgrounds for each country. After applying these five criteria to the potential Country Experts V-Dem chooses which experts can participate (V-V-Dem, 2020, p.13-14).

V-Dem addresses certain concerns that Benoit & Wiesehomeier (2009) express. One of these concerns is the expert bias, which V-Dem counters by their fourth criterion on impartiality. Another

issue with expert surveys is to what extent expert surveys can be relied upon since they might not always reflect the latest changes. V-Dem states nothing about this but brings out a new dataset every year and also reflects on changes through providing time-series ratings (V-Dem, 2020, p. 11). A third concern is the possibility that experts respond to the same questions differently; ‘different experts may use different conceptions’ (Benoit & Wiesehomeier, 2009, p. 510). This problem however seems almost irresolvable since ‘experts may also differ in their interpretation of the same scale, no matter how well defined and precisely worded the end points of the policy dimensions are’ (p. 510). It is irresolvable in a way that the judgment itself might not be wrong but the conception of an observation might differ (p. 511). At last, there is the possibility of expert fatigue. This is the possibility that experts become tired of being asked to fill in the same questionnaires over and over again and eventually will quit participating (p. 511). This apparently is only problematic when there is a limited pool of coders per country. V-Dem has a vast pool of coders from which they can choose but also admit that for some smaller countries the number of available coders is rather limited. Nevertheless, the 2020 dataset by V-Dem has no remarks on expert fatigue or a lack of experts on certain countries (V-Dem, 2020, p. 10).

3.2 Case selection

The sample that is used in this research design consists of 18 countries that are considered to be Latin-American. These countries are: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, the Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay & Venezuela. There are arguably more countries that could have been added for geographical reasons but due to lack of data no more than the mentioned twenty countries are included in this research. I will use the past forty years of available data on these countries. The motivation for this is that the third wave of democratization has started from around 1980 and on in Latin America. This third wave of democratization is important since we are investigating clientelism, a concept that is only worth observing in democratic systems. Since clientelism is a concept that is only useful to research in times of democracy, I have filtered out the years for each country that were considered not to be democratic. To determine whether or not a country was democratic I have used the Polity2 variable from the Polity IV dataset (Polity, 2020). This variable is a revised Combined Polity score which is basically a modified version of Polity IV’s Polity variable that facilitates the use of the Polity variable in time-series analysis (Polity, 2020). The scores on this Polity2 variable range from -10 to +10. For this research I will only use cases that have a score of +5 or higher on the Polity2 variable. This results in the following years for each country that are included in the research design:

Argentina 1983-2019 (37) Guatemala 1996-2019 (24)

Bolivia 1982-2019 (38) Honduras 1982-2019 (38)

Brazil 1985-2019 (35) Mexico 1997-2019 (23)

Chile 1989-2019 (31) Nicaragua 1990-2019 (30)

Colombia 1980-2019 (40) Panama 1989-2019 (31)

Costa Rica 1980-2019 (40) Paraguay 1992-2019 (28)

Dominican Republic 1980-2019 (40) Peru 1980-1991 (12) 2001-2019 (19) Ecuador 1980-2019 (40) Uruguay 1985-2019 (35) El Salvador 1984-2020 (36) Venezuela 1980-2008 (29) Totals

Countries 18 Number of cases 606

As the table above tells us, we are left with 606 cases that will be used in trying to answer our research question. The next section will discuss the dependent variable clientelism.

3.2.1 Dependent variables

The dependent variables in this thesis are based on the concept of clientelism. Clientelism can be measured in multiple ways. We could for instance say that there either is or either is no clientelism. However, in this thesis that tries to explain if social capital influences the presence of clientelism it is way more interesting to find out how strong the effect on clientelism is than to determine whether or not there is clientelism. Also, determining when there is or when there is no clientelism is very hard since the statistics show us that all countries have or have had to deal with clientelism in some way, be it a lot of clientelism or almost no clientelism.

I use the disaggregated Clientelism Index variable to measure the amount of clientelism (V-Dem, 2020, 273-274). The Clientelism Index is a numeric variable which allows us to find out how social capital influences clientelism and is therefore not just a variable that tells us if there is any clientelism or not. In this research there are three dependent variables which link up with the hypotheses that were provided in the previous chapter.

The first dependent variable is Vote buying (v2elvotbuy): In this dataset, ‘vote and turnout buying refers to the distribution of money or gifts to individuals, families, or small groups in order to influence their decision to vote/not vote or whom to vote for. It does not include legislation targeted at specific constituencies, i.e., "porkbarrel" legislation’ (V-Dem, 2020, p. 64). Vote buying or turnout