Bribery Cartels: Collusive Corruption in Bolivian Street Markets

Many Bolivians engage in corruption through intermediaries, like civil society representatives and lawyers, instead of paying officials directly. People vocally resent that intermediaries add an extra layer of costs and opaqueness to corruption but still choose to pay bribes through intermediaries that knowingly take advantage of them. Why do intermediaries facilitate corruption? While other studies on corrupt intermediaries find that they reduce uncertainty and transaction costs, this study contributes to corruption research by finding that intermediaries engage in cartel-like behavior by disproportionally helping officials and intentionally increasing uncertainty and costs for the average citizen. Ethnographic evidence from street markets in La Paz demonstrates that civil society actors like street vendors’ union representatives advance their careers by collecting and delivering bribes to specific bureaucrats. Collusive relationships between bureaucrats and intermediaries hide and perpetuate corruption, while giving the appearance of a transparent government that is responsive to civil society. officials and civil society leaders have powerful incentives to enter into mutually beneficial corruption agreements. This study worked with street vendor representatives as cases of civil society actors who become intermediaries in corruption; future research could investigate if similar incentive structures exist for other types of civil society leaders, like NGO directors, members of the media, or prominent lawyers. Future research could also examine if intermediary roles change in different economic contexts.


Introduction
conceptualizes corruption as an illicit exchange between a citizen and a public official (e.g., Rose-Ackerman 1999;Treisman 2007;Olken and Pande 2012). However, a small number of studies posit that Eliana's strategy is common: researchers have documented intermediaries handling bribes across Latin America and Europe (Della Porta and Vannucci 1999;Lambsdorff 2002Lambsdorff , 2013, and in settings ranging from the Nigerian business sector (Ufere et al. 2012) to Indian licensing offices (Bertrand et al. 2007). Why do intermediaries facilitate corruption?
Research on intermediaries in corruption argues that they reduce transaction costs and uncertainty (Della Porta and Vannucci 1999;Drugov, Hamman, and Serra 2014;Lambsdorff 2002). Researchers find that intermediaries reduce uncertainty by clarifying prices and expectations (Bertrand et al. 2007), and remove the citizen from the transaction with the official, reducing the citizen's legal (Lambsdorff 2013) and moral corruption costs (Drugov, Hamman, and Serra 2014). In contrast to past work, this project finds that intermediaries in La Paz's street markets use their position to enrich and protect officials-at the expense of the citizens that they claim to serve. As one of the first studies from the point of view of intermediaries, this study contributes to corruption research by finding that intermediaries in La Paz's street markets produce uncertainty in order to preserve their role as middlemen, block competitors, and keep bribes high.
I argue that civil society representatives, like street vendor leaders, become corrupt intermediaries because it advances their careers and increases their incomes. Representatives like union leaders have plenty of incentives to become intermediaries in corruption: intermediaries make money from corruption and become gatekeepers to specific officials, which increases their standing in their organizations. Officials support intermediaries because they prefer working with a small number of known representatives instead of hundreds or thousands of individuals. Intermediaries bolster officials' careers and income by reducing risks, bringing officials work, and keeping bribe rates high. Intermediaries and officials reach informal, cartel-like agreements to channel corruption through the intermediaries; I term these agreements "bribery cartels." Bribery cartels exist beyond Bolivia: the 2015 FIFA scandal uncovered a bribery cartel between officials and contractors, and in the 1990s prosecutors uncovered a bribery cartel in São Paulo that channeled millions from street vendors to mayor Celso Pitta's reelection fund (Itikawa 2006). Bribery cartels advance officials' and intermediaries' careers at the expense of the citizens they serve.
This study contributes to research on corruption by demonstrating how intermediaries create uncertainty about corruption and then profit from hierarchies between citizens and officials. I support the argument with ethnographic evidence from nine months of working with street vendors and their intermediaries in La Paz, Bolivia. I analyze corruption in La Paz's street markets as a case of intermediaries facilitating corruption. The following section discusses the causes of corruption and the role of intermediaries, particularly in informal markets. The theoretical section develops why and how representatives collude with officials to extract bribes from ordinary citizens like street vendors. The theory explains why intermediaries want to keep costs and uncertainty high, contrary to the expectations of other studies, and why citizens knowingly work with intermediaries who take advantage of them. The paper concludes with implications for reducing corruption and evaluating reforms in places where intermediaries facilitate corruption.

Corruption, Intermediaries, and Markets
Corruption is the misuse of public office for private gain (Rose-Ackerman 1999, 91;Treisman 2000, 400). People engage in corruption where they have the opportunity and where they face little risk of punishment; when conditions change, they adjust (Balán 2011;Della Porta and Vannucci 1999;Rose-Ackerman 1999). Some economic and political conditions-like new regimes, state-run monopolies, and natural resource revenues-create more opportunities for corruption than do others (Mauro 1995;Treisman 2007;Ufere et al. 2012;Weyland 1998); weak or unstable institutions encourage corruption by making punishment less likely (Bardhan 2006;Rose-Ackerman 1999).
With the notable exceptions of Lambsdorff (2002) and Drugov et al. (2014), few studies or policies address the role of intermediaries in corruption, despite evidence that they play an important part (Bertrand et al. 2007;Della Porta and Vannucci 1999;Ufere et al. 2012). More general theories establish that intermediary agents facilitate exchange in contexts with high transaction costs and specialized information; Della Porta and Vannucci (1999) and Lambsdorff (2002) posit that illegal transactions like bribery have these characteristics, creating a role for intermediaries who can then reduce costs and enforce informal contracts. In La Paz's street markets, intermediaries facilitate corrupt transactions in part because they reduce transaction costs and mobilize specialized information, as Lambsdorff (2002) expects. However, these theories largely overlook how intermediaries bring unequal benefits to the citizen and official in the transaction, largely benefitting the politically powerful participant.
Corruption can refer to practices that straddle the line between abuse of power, professional discretion, and good etiquette (Olivier de Sardan 1999;Granovetter 2007;Taylor 2004). Officials, citizens, and researchers may disagree about what constitutes corruption, even with an established definition (Auyero 1999;Olivier de Sardan 1999;Lazar 2007). For example, one intermediary described bribing an official for an upgrade for her stall. Upgrades are legal and there is a clear application process for requesting one, but bureaucrats have discretion over granting them. The woman explained that bureaucrats expect bribes for approving upgrades. She then recounted an expensive lunch that she made for a bureaucrat after an approval, which she called a coima or bribe, and called the upgrade process corrupt. I was surprised to hear that she considered the lunch a bribe, and not an innocuous favor. 2 In this project, I consider an exchange corrupt if the people discussing it signaled that they considered it corrupt by using words such as coima, corrupción, or extorsionar to describe the exchange. In some conversations, people used broader verbs or code words, such as tomar, "to take," or reconocer, "to recognize," and in those cases I rely on the context of the conversation. For example, telling a bureaucrat "Voy a reconocer tu trabajo," which means "I will recognize your work," is widely understood to mean "I will give you a bribe or gift later." 3

Corruption and informal markets in Bolivia
Eighty percent of Bolivian workers labor informally, and an estimated 68 percent of the country's GDP comes from informal work (Schneider, Buehn, and Montenegro 2010). Street markets serve populations around the country and street vending is a common occupation (Goldstein 2016;Kohl and Farthing 2006;Tassi et al. 2014). Informal workers' unions structure and regulate most street markets and sort vendors into a hierarchy based on their legal status and relationship to the state (Goldstein 2016;Lazar 2007). Street markets are almost always semi-legal and on public land, which means that street vendors' relationships with local officials shape the profitability of their businesses (Goldstein 2016), and street vendors select intermediaries like union leaders who have strong relationships with local officials (Hummel 2016). Though profits in many sectors depend on business people's relationship with officials (Boone 1993;Schneider 2004), informal workers' semi-legal work status make them even more vulnerable to extortion by public and private actors (Bhowmik 2010;Fernández-Kelly and Shefner 2006;Goldstein 2016;Itikawa 2006).
In La Paz in the 1980s and 1990s, under the MacLean Abaroa administration, police arbitrarily collected sentaje payments-a daily tax for using public space-from street vendors. Street vendors report that while they did receive sentaje receipts, they paid sentaje and a bribe to the police officer, who would in essence overcharge. Anjaria (2011) documented similar tax and bribery schemes in Mumbai's street markets. After controversial decentralization reforms in 1994 (Kohl and Farthing 2006), the La Paz city government negotiated a new licensing policy with street vendor union leaders. The pago único municipal centralized licensing through the creation of the Markets Office and delegated licensing decisions to a handful of bureaucrats within the office. Street vendors now pay larger bribes in lump sums to a small cadre of bureaucrats instead of small, periodic bribes to a large number of police officers. 4 In the case of street vending in La Paz, street vendor representatives capitalized on the reorganization of licenses to collude with bureaucrats and form bribery cartels.

Bribery Cartels
When actors form cartels, they collude with a small number of other producers to restrict production in order to keep prices artificially high (Della Porta and Vannucci 1999). Cartels benefit members by keeping prices high and competition low, and form in markets or sectors with a small number of players. Many agreements and relationships take on cartel-like features. Government services can take on the characteristics of oligarchic markets, which officials can exploit for private gain (Bardhan 2006;Della Porta and Vannucci 1999): where people must go to a small number of places to obtain a given service-like vending licenses-the officials running those services can extract bribes by threatening to approve or deny requests (Rose-Ackerman 1999).
In the case of street markets in La Paz, representatives manipulate their knowledge of the Markets Office's informal and formal rules to become intermediaries. Informal rules and agreements can usurp formal rules and agreements (Helmke and Levitsky 2006), and informal agreements can take on cartel-like characteristics (Della Porta and Vannucci 1999). Erdmann and Engel (2007, 105) argue that when informal and formal logics exist in bureaucracies, the mix creates "insecurity about the behavior and role of state institutions and agents." I suggest that informal agreements between officials and representatives take on cartel-like features where they restrict access to officials and the services that they are supposed to provide to the public. In the case of street vendors in La Paz, the municipal Markets Office grants licensing requests and a few bureaucrats within the office approve or deny applications. These requests are brought to them by a small number of street vendor union leaders; together, several of these actors form what I term a "bribery cartel." Bribery cartels are informal and illicit agreements within a small network of bureaucrats and representatives. The defining features of a bribery cartel, for the purposes of this project, are that participants (1) collude to extract bribes, and (2) restrict who collects and receives bribes to a small group of known actors. The group must be small enough that it is possible for all members to recognize each other, and certainly smaller than the pool of potential bribers. Cartels are not necessarily formal organizations, though they can be. For example, bureaucrats and businesspeople formed a national bribery cartel in Brazil that siphoned billions by setting bids for public projects; at least one national corporation had a parallel accounting department that functioned as a bribery department (Beauchamp 2016). Itikawa (2006Itikawa ( , 2010 documents a separate and simpler set of agreements between officials and street vendor leaders in downtown São Paulo that siphoned millions to local officials and union leaders. In São Paulo, rank-and-file vendors recognize this behavior as cartel-like and call these agreements máfias, especially máfia da propina, or bribe mafia (Itikawa 2010, 188). Della Porta and Vannucci (1999, 21-22, 183-186) also find that informal agreements in Italy between businesspeople and politicians constitute cartels.
In La Paz's street markets, tens of thousands of street vendors want licenses and could potentially bribe bureaucrats (Hummel 2017). A handful of bureaucrats at the Markets Office control licensing decisions and can use their position to demand rents from these tens of thousands of potential bribers. I estimate that of the roughly five hundred street vendor representatives in the city's vendor unions, perhaps one hundred of them work as bribery intermediaries, with several dozen conducting most of the activity. 5 These relationships constitute cartels because intermediaries and officials restrict access to government goods and services, like street vending licenses, which keeps bribes high, as the empirical section demonstrates.

Cartel winners and losers
Civil society leaders collude with officials to facilitate corruption because they gain political and financial benefits from working as corrupt intermediaries. Civil society leaders who develop working relationships with officials can use those relationships to advance their careers, particularly in sectors like street vending, where profitability depends on relationships with state actors (Fernández-Kelly and Shefner 2006;Goldstein 2016). Leaders become gatekeepers to the officials who they work with, which improves their standing with the people they represent and with other representatives who have a different set of working relationships. Representatives also have a financial incentive to become corrupt intermediaries because they can pocket a portion of the bribes they handle.
Why do officials partner with intermediaries instead of using their position to extort directly? Officials collude with intermediaries because they simplify officials' work and keep bribes high (Della Porta and Vannucci 1999;Auyero 1999). First, intermediaries simplify officials' jobs by bundling bribes. Officials can then limit their illicit dealings to a handful of known associates instead of dealing with hundreds or thousands of individuals that the officials may not know or remember. Second, intermediaries protect officials' careers by limiting who witnesses corrupt transactions. Anti-corruption reforms potentially increase monitoring by encouraging citizens to report corruption (Bardhan 2006;Gray-Molina, Perez de Rada, and Yáñez 2001). 5 These are rough estimates based on fieldwork. The city has 31,906 licensed vendors (GAMLP 2015), and officials and leaders estimate that another 20,000-30,000 work without licenses (Hummel 2017). Fewer than a dozen bureaucrats control licensing decisions, as stated in the city's internal documentation. The largest vendor federation of the city's four federations has 140 unions, but only about 80 regularly attend meetings and other activities. Each union has 1-20 elected representatives, and some representatives hold multiple positions. Thus, I estimate that there are roughly 500 street vendor representatives in the city. Some are more active than others, and I tentatively estimate that there are about 100 active representatives at any time. During interviews and observation, vendors and bureaucrats repeated that the same dozen representatives had illicit relationships with specific bureaucrats. Multiplying that over the organizations I did not work with, I estimate that several dozen street vendor representatives do most of the bribing in the Markets Office.
When private individuals hand money to an intermediary, who then passes it to an official, the individual cannot directly monitor the official. They do not know if the intermediary paid the official or how much each participant's cut was. If the promised good or service is not delivered, individuals do not know whether to blame the official, the intermediate, or both, and a corruption denunciation becomes weaker without a willing eyewitness. Finally, intermediaries keep bribe prices high because private individuals do not know how much the intermediary keeps and how much goes to the official. Why do people work with intermediaries who charge extra fees instead of bribing officials directly? First, where officials limit corrupt transactions to known intermediaries, citizens may not have the option to bribe officials directly. Officials may refuse to take bribes from unknown people because they worry about being denounced, or because intermediaries simplify and streamline their work (Della Porta and Vannucci 1999). Second, officials may benefit from the uncertainty that an intermediary can create about the informal and formal rules of the bureaucracy (Helmke and Levitsky 2006), particularly if uncertainty keeps bribes high. In these ways, informal bribery agreements between officials and representatives constitute bribery cartels because the agreement restricts who can bribe; 6 the average vendor must go through the cartel if they choose to pay a bribe for a service. Additionally, informal bribery agreements between officials and representatives constitute cartel-like behavior because they keep prices high.
In La Paz's street markets, people are often uncertain about how to bribe officials and how much to pay them, but as the empirical section demonstrates, they know that representatives work with officials as middlemen in corruption. Organized street vendors in La Paz also know that their representatives often extort members when they facilitate bureaucratic processes (Goldstein 2016;Lazar 2007). Why do vendors knowingly elect union representatives who extort them? Informal workers' careers depend on their relationship to the state (Fernández-Kelly and Shefner 2006; Goldstein 2016). Therefore, they prefer representatives who have working relationships with key officials over those who do not, even if representatives use their relationships to extort the people they represent. Many vendors prefer service and representation with extortion to no service or representation. Additionally, street vendors acknowledge that representatives' requests for money fall into a morally gray area, since most street vendor representatives provide valuable services but receive no salary and, as a result, expect individuals to compensate them for favors. Finally, many vendors request that officials bend the rules. Rank-and-file vendors can benefit from corrupt representatives in these situations, which gives vendors another reason to reelect corrupt representatives. When vendors pay representatives to bribe officials or reelect corrupt intermediaries, they "reproduce . . . a powerful web of political domination" (Auyero 2001, 84). 7 Representatives like street vendor union leaders become intermediaries in corruption because the role advances their careers and brings in extra income. Intermediaries collude with officials to channel bribes through a small set of intermediaries, forming bribery cartels that enable both representatives and officials to manipulate their positions for private gain. Bribery cartels hide and perpetuate corruption, making it difficult for citizens to hold civil society representatives or public officials accountable.

An Ethnography of Corruption Methodology
Between August 2014 and May 2015, I conducted fieldwork with one of the city's four street vendor federations; I observed meetings, engaged in conversations, interviewed forty-three vendors and bureaucrats, and worked as a street vendor. I worked as a street vendor once a week in a prominent openair market from March to May 2015 and helped out at the stalls of individual street vendors in other markets. 8 All quotes and anecdotes come from the ethnographic data and have been anonymized, due to the illegal nature of the topic. The names I use are pseudonyms, and details like merchandise and location have been changed to ensure confidentiality. 9 Researchers rarely observe corruption because actors have incentives to hide illicit activity (Jensen, Li, and Rahman 2010;Olken and Pande 2012). Interview and survey methods capture perceptions and some self-reporting, but respondents rarely trust enumerators and typically underreport corruption (Olken 2009;Jensen and Rahman 2015). Ethnographic methods rely on routinely triangulating sources' words with their and others' actions, and investigating discrepancies and changes over time (Auyero 2012;Jerolmack and Khan 2014). Many ethnographic projects, like this one, revolve around the significance of documented discrepancies between words and actions (Lazar 2007;López 2007). Additionally, ethnography requires the researcher to build trusting relationships, which are particularly important for research on illicit activities and other topics where sources have an incentive to misrepresent (Auyero 2001;Goffman 2015). In these ways, ethnography overcomes several of the limitations that corruption researchers encounter with survey and interview methods, making the method particularly well suited for studying micro-level illicit activity (Contreras 2012;Nordstrom 2007).
I began fieldwork by approaching the leadership of one federation and asking to observe meetings and work, as well as interview consenting vendors, over nine months. The leadership agreed on the grounds that I would give something back to the federation, and the ampliado-the meeting of the governing council of the federation, consisting of the elected leadership of the 142 affiliated street vendor associations-voted to let me observe. 10 The federation leadership initially suggested that I give back by teaching classes in Marxist philosophy and economics, but the ampliado asked me to teach workshops on street vending policies and regulations in the United States and Brazil instead. I gave two presentations, disseminated a final report on my work, and helped out around the office and in the markets where possible, primarily by running errands, watching stalls, and babysitting, and occasionally by translating or teaching English or computer skills. Most people in La Paz's markets are bilingual or trilingual, and I also learned some Aymara to improve my rapport with older vendors. I made contacts by meeting people at the federation and city offices, and when research participants introduced me to colleagues in the markets.

Context
The main thoroughfare in La Paz winds down a valley in the Andean mountains. Streets creep up all sides of the valley, supplying the city's main markets on its west side, working-class neighborhoods to the north and east, and affluent settlements in the south. La Paz has hundreds of open-air street markets, where fifty to sixty thousand of the city's one million residents work (Hummel 2017). The center of the city hosts the most profitable markets while new vendors regularly create new and less profitable markets in the city's expanding peripheral neighborhoods. The city licenses 31,906 street vendors, and the rest work without documentation, typically as roving ambulante vendors (GAMLP 2015). La Paz's street vendors organize more than most, globally speaking (Bhowmik 2010): 75 percent belong to a street vendor association or union (Hummel 2017). Associations and unions organize geographically, representing several blocks or a market, and affiliate with one of four citywide street vendor federations. Elected association and federation leaders represent their members to other organizations and to the government. Officials rarely take bribes in street markets; instead, vendors give money to representatives in the markets, representatives bundle bribes, and then deliver money and gifts to officials in the Markets Office.

Corrupt incentives for representatives, bureaucrats, and vendors
In La Paz's street markets, the work status and careers of tens of thousands of vendors depend on dozens of minute regulations and decisions made by a handful of bureaucrats in the Markets Office. Bureaucrats interpret local ordinances and decide what and where vendors can sell, how many vendors can work legally, how large their stalls can be down to the tenth of a centimeter, and what they can be made of, to name a few issues. Street vendor organizations and the representatives who work as intermediaries in corruption share similarities with the Peronist problem-solving networks and clientelist brokers that Auyero (2001) documented in Argentina: intermediaries solve problems, for a price, and in doing so perpetuate the asymmetric political power between street vendors and bureaucrats. This section shows how street vendors' dependence on bureaucratic discretion incentivizes street vendor representatives to become intermediaries between bureaucrats and vendors negotiating bribes for favorable, and in some cases illegal, decisions.
Many vendors admit in private to paying bribes. Raquel is a severe, middle-aged union representative who has maintained her position and her union's power through careful political maneuvering. Raquel, like many other union representatives, sees bribery as a standard career and business strategy, and shrugs when she says that her entire association has paid bribes. Silvia is a friendly, approachable woman who, as a former representative, knows how to guide paperwork through the bureaucracy and occasionally helps younger vendors learn the ropes. Silvia believes that actual bribery levels are higher than surveys and other selfreports reveal. Silvia explained, "I have paid, several times. People will deny it when answering survey questions about money. I don't even know how many times I've paid a bribe and I would still deny it." In La Paz, street vendors pay bribes when they interact with the local bureaucracy. When a street vendor has a problem, she goes to a street vendor representative like Raquel and then accompanies the representative to the Markets Office. At the office, the street vendor files a request for a change or an audience with a bureaucrat and enters the official request, called a trámite, into the city's computerized tracking system, along with any required supplemental information. The vendor will follow up with the office until she gives up or receives a meeting with a bureaucrat. Typically, the bureaucrat is a lawyer in charge of licensing decisions, such as Benjamín López or Santiago Pérez. Benjamín López and Santiago Pérez are educated, middle-class men who hold appointed positions in the Markets Office. Street vendors report that bureaucrats like López and Pérez do not ask for bribes directly, even when they expect them, and will not accept bribes from people that they do not have a previous relationship with.
For example, Juanita is an older former vendor representative who prides herself on having turned a profit on everything from unregulated pharmaceuticals to packing tape. As a former representative, she does most of her own paperwork and explains that individual representatives and bureaucrats have mutual bribery agreements. The agreements block rank-and-file vendors from bribing bureaucrats directly, which is why they give money to an established intermediary to resolve their issues with the Markets Office. Juanita explained: "Members give money to their leader to do a specific thing. And this leader will pocket part and give the rest to the official. They have an agreement, like the one that Don Alfonso has with Dr. López. They will not accept just anyone." In private, street vendors confirm that most of them pay bribes to resolve problems, especially legally dubious problems, at the Markets Office. For example, the city issues vendor licenses to individuals, who are legally allowed to transfer the license to another person only if the vendor can no longer work, the family can demonstrate hardship, and the new license holder is a family member. In practice, vendors treat licenses like real estate deeds and resell them to unrelated people for thousands of dollars. A license's value depends on the stall's location and the status of a license: a centrally located stall with a full-day license and no outstanding license issues can sell for $50,000 or more. Open bureaucratic processes can halve the value of a license, while licenses in peripheral markets sell for a few hundred to a few thousand dollars. Bureaucrats accept bribes, typically percentages of the estimated sale price, to rubber-stamp these legally dubious license transfers as officially sanctioned changes to the license. Street vendors resent corruption and extortion in street markets, but most prefer paying a bribe for a favorable and illegal decision, like selling a license, to receiving an unfavorable legal resolution.
Many career street vendor representatives-such as Alfonso García, whom most vendors call Don Alfonsowork for years with the same mid-level bureaucrats. Don Alfonso has a university degree and a union background and started working as a street vendor during the hyperinflation of the 1980s, when vending was one of the only available jobs that enabled people to bring home cash before it lost value. Other vendors quickly elected Don Alfonso, with his union experience and formal education, to a leadership position, where he cultivated relationships with bureaucrats. Vendors know that these specific representatives have muñeca-which literally means "wrist" or "doll" but colloquially means influence-with specific bureaucrats, like Benjamín López and Santiago Pérez. Eliana explains what rank-and-file vendors are expected to do when they ask for Don Alfonso's help with a license: "Here's how it works: keep Don Alfonso company, pay for the taxi there and back, go back with him, pay for food. You pay him at the end but he doesn't ask for it, you give him what you think it's worth. I paid him but it was not that much." Representatives typically charge members for their services. Some expect payment on delivery or that vendors pay for taxis to the Markets Office and lunch, while others charge set amounts for small services like writing official letters or finding supplementary documents. Juanita states that all representatives participate in these exchanges, "All of the leaders take money." I observed representatives accepting taxi rides and lunch from members they were helping and routinely taking small amounts (10-30 bolivianos or US$1.50-$4.50) for typing up supplemental documents and letters on official letterhead. Rank-and-file vendors and representatives agree that this is common and acceptable but that most representatives ask for more from members or take more from association bank accounts. One association member waved away questions about electing clean representatives with the Aymara word for stealing: "They all steal [lunthatapjje]." Juanita followed up with, "They are all extortionists. . . . The problem isn't the extortion, everyone does that, but how it is handled." Juanita went on to explain that when street vendors elect new representatives who are not corrupt, most quickly decide to make extra money from bribes when given the chance. When a representative refuses to participate in corruption, more senior intermediaries and bureaucrats refuse to work with them. The representative then becomes sidelined because they cannot use extralegal channels to help their members resolve problems with the Markets Office. However, if a representative cultivates working relationships with other intermediaries and bureaucrats, they can pull strings for their members and position themselves as indispensable, advancing their careers as civil society representatives.
Once representatives start working as intermediaries in corrupt transactions, they become implicated in and dependent on corruption and are unlikely to denounce the officials they work with. Street vendors are small-business owners who typically work hard for small profits; those who become representatives have little incentive to denounce corruption and lose that additional income. As one vendor put it, "There are people who like being leaders, they get economic benefits from it." Silvia, an occasional intermediary herself, explains that the collusion is too profitable to denounce: "Everyone knows, but they make money off of this." Intermediaries help bureaucrats advance in careers by limiting who witnesses bribery and therefore who can make a compelling denunciation. When representatives take and deliver money, bureaucrats have plausible deniability. In interviews, bureaucrats deny corrupt activities and blame the representatives who take money from their members. For example, Benjamín López places all of the responsibility on representatives: "People think that there are economic agreements, bribery agreements, between me and the leaders and that is a disadvantage. That's another reason why I [was promoted]: transparency. I am more transparent than others and we've made the Markets Office much more transparent. I don't like that [representatives] charge their members so much. We know that they take a lot of money out of them." Privately, Silvia detailed bribing López and many more retold stories of others bribing him. Three vendors independently repeated the same story about one association giving López a car, and he publicly discussed the gifted car in front of me and other vendors. Beyond the accusations, which I could not independently verify, López and others exhibit a level of consumption much higher than their official salaries could allow. For example, López's salary should have been around 9,000 bolivianos or US$1,304 per month, given his position and the city's public salary grades at the time of research. Yet he publicly discussed his international vacations, wore designer shirts that cost hundreds of dollars, wore a designer watch, and owned a large house in a wealthy neighborhood where houses cost hundreds of thousands of dollars-supposedly paid for with a public salary of around US$15,000 a year. Other bureaucrats were less conspicuous but often showed up to work in designer shirts and accessories that would be difficult to afford on a public salary.
Bureaucrats do not ask for money directly, despite prevalent corruption. Uncertain about the informal rules and blocked from bribing bureaucrats directly, street vendors ask their association representatives for help. Representatives then capitalize on their relationships with bureaucrats and their members' uncertainty, and take a cut before passing bribes on to bureaucrats.

Bribery cartels: Uncertainty, prices, and unequal benefits
The city works closely with civil society groups, including organized street vendors. Benjamín López, Santiago Pérez, and other bureaucrats develop working relationships with Don Alfonso, Raquel, and other street vendor representatives, whom they see almost daily. In many cases, working relationships become bribery cartels, in which bureaucrats channel favors and resources to specific representative, who in turn collect money from their members, take a cut, and pass it on to bureaucrats.
Eliana is more outgoing than most vendors and often insists on accompanying representatives to bureaucrats' offices so that she can learn about the process. Despite her attentiveness and frequent visits to bureaucrats, she goes through intermediaries to bribe public employees. She explained, "There are people who work as intermediaries between bureaucrats and police and people that need something resolved. They know how to negotiate and settle on a bribe price and then tell the other person that it'll be more, and then they pocket the difference." The opaque negotiation process between bureaucrats and the known intermediaries that they work with serves to keep the average vendor uncertain about how much the bribe actually is. This uncertainty about the informal rules and the money exchanged keeps bribe prices high, because intermediaries have every incentive to overstate what officials ask for in order to pocket more money, and the average vendor has no reliable way to verify what the official will accept. This dynamic is similar to other types of cartels where insiders intentionally keep consumers uncertain about costs, pricing, and negotiations.
Vendors like Eliana expect that representatives help their members-for a fee-as a matter of course. Sometimes intermediaries charge more for clearly illegal requests or to help more successful vendors or people who are not affiliated with their associations. For example, vendors from the central and most profitable markets tend to pay higher bribes than vendors from peripheral markets, because intermediaries and officials know that people with central stalls make more money. Intermediaries may also charge large amounts if they have monopolized access to a bureaucrat or have a particularly privileged relationship.
For example, Judith is an energetic union representative who leads an umbrella group. As a representative for thousands of members, Judith interacts with the bureaucracy as an intermediary every day and maintains close relationships with several officials. The new leader of Judith's former association alleges that Judith, now in a higher position, has used her privileged relationships with officials to lucrative ends: "Judith was my association leader. And we found out that she was taking a lot of money from people to change documents at the Markets Office. I have receipts with her signature. She was charging 400, 600 [US] dollars for her signature." Vendors occasionally seek out representatives of other associations who have established working relationships with specific bureaucrats and pay more for special favors and connections. For example, when Juanita wanted to upgrade her stall, she first approached her association representative. When her request stalled in the city's paperwork system, she approached Don Alfonso, a representative higher up in the city's street vendor organizations than her association representative, and a better networked intermediary. When the request stalled yet again, she considered going to a different association to find a more effective intermediary. She explained the dynamics of one bribery cartel between a bureaucrat and two established union representatives who work as intermediaries, "Pérez takes bribes and several leaders are in with him: Esther, Maria. They get along well with him and he solves their issues. I may approach Maria or another woman to help me bribe Pérez for the stall, because I am not getting anywhere with Don Alfonso." Juanita usually does her own paperwork or trámites because, as a former representative herself, she knows the formal and informal rules. However, she does not have a preexisting bribery relationship with Santiago Pérez, the bureaucrat assigned to her case, and her representatives have not helped. As the above quote illustrates, at one point in a multiyear process, she considered approaching a different representative with whom she is unaffiliated but who has a known bribery relationship with Santiago Pérez.
Other street vendors report that bureaucrats and representatives occasionally manufacture problems to shake down rank-and-file vendors. For example, bureaucrats and representatives frequently decide which vendors can stay or move during construction projects. These decisions can be worth thousands of dollars to vendors because those who stay face less competition and maintain stalls in high-traffic areas. In contrast, the city sends most of those who move to emptier and less profitable areas. These temporary moves frequently become permanent if the city never finishes the project or refuses to reissue licenses after construction. Judith alleges that decisions are based on which vendors pay the highest bribes, "There are accusations against the leaders [of another organization]. Like on Hidalgo Street, they did a project and moved many stalls, not all but many, and it was the members who paid the most who stayed." This accusation resurfaced in conversations with other street vendors and union representatives.
Vendors report that both street vendor representatives and bureaucrats find out who can pay bribes and allow them to stay. If a vendor cannot afford bribes, the representative will alert the city, which will then revoke the vendor's license. Collusion extends to other projects and informal quid pro quo relationships as well. For example, Don Alfonso discussed how his relationship with Santiago Pérez includes exchanging useful information, which he passed on to Esther, a small grandmother with a reputation for forging stall licenses. In a group conversation at union headquarters, Don Alfonso turned to Esther with, "Don't say you haven't benefited from Dr. Pérez. When they were going to check your stall measurements, he called me and then I called you and said fix your stall! And they went to inspect you many times." The city government campaigned on an anticorruption platform, runs anticorruption publicity campaigns, and established transparency offices inside every city department. Scholars and journalists hail the municipal government's efforts as a success (Zúñiga and Heywood 2015). Nevertheless, collusion between bureaucrats and civil society leaders keep vendors from denouncing corruption. Silvia recounts a conversation with a street vendor who a bureaucrat had extorted: "There is another vendor who asked me for help. I told her that she should denounce this official, but she said no. And it's true: the whole city hall could grab her and she could lose everything. . . . They want us to think that there is a Transparency Unit, but if we make anonymous reports they don't do anything and if we denounce in person, what will happen? Who are we going to open our mouths to?" Instead of publicly denouncing corruption to the Transparency Units, street vendors complain to their representatives, who are themselves implicated in corruption. Sometimes, street vendors directly confront representatives whom they have paid but who have not delivered on their promises, as I witnessed several times at the federation headquarters. Mirroring Benjamín López, intermediaries place all blame on the bureaucrats they work with when members confront them. More frequently, rank-and-file vendors denounce intermediaries to other intermediaries such as Judith, who says, "Many come here, they don't want to denounce publicly, they are afraid of denouncing publicly or to the city, so they come and denounce to me." Corruption enters into street vendors' interactions with officials in La Paz. However, vendors rarely bribe officials directly because the officials will not accept bribes from just anyone. Instead, vendors bribe officials through their civil society representatives. Officials develop illicit relationships with specific representatives, who collect, bundle, and deliver bribes. These relationships enrich and advance the careers of officials and representatives at the expense of rank-and-file street vendors.

Conclusion
Around the world, intermediaries construct their careers by facilitating corrupt transactions (Della Porta and Vannucci 1999;Lambsdorff 2013;Ufere et al. 2012). Existing studies argue that intermediaries specialize in navigating the informal rules of corruption, which helps the citizen who works with the intermediary as well as the official they pay (Bertrand et al. 2007;Drugov, Hamman, and Serra 2014;Lambsdorff 2002). However, I have argued that civil society representatives become intermediaries in corruption to advance their careers as leaders, as well as to make extra money. In La Paz, Bolivia, representatives then collude with officials in cartel-like agreements that take advantage of the citizens that both claim to serve. Citizens reinforce bureaucrats' power over them when they bribe through intermediaries, but they participate because they do not have a better alternative. This study contributes to corruption research by demonstrating why and how intermediaries produce uncertainty, and why ordinary citizens continue to pay bribes through intermediaries who disadvantage them.
Bribery cartels operate around the world: in 2015, investigators uncovered bribery arrangements between FIFA representatives and officials from dozens of countries that featured cartel-like behavior. In the FIFA scandal, officials colluded to keep contracts and kickbacks within a circle of corrupt associates, and taxpayers footed artificially high bills. Della Porta and Vannucci (1999) document cartels of businesspeople and politicians in Italy who collude to keep influence and money circulating between cartel members. Ufere et al. (2012) and Fisman and Gatti (2002) found similar collusion in Nigeria's and Indonesia's business sectors. Itikawa (2006Itikawa ( , 2010 documented a similar arrangement between street vendor leaders and city bureaucrats in São Paulo, which street vendors called a bribery mafia: the city government extorted street vendors in downtown São Paulo to the tune of 1,000,000 reais a month-roughly 500,000 US dollars at the time-which was channeled through street vendor association leaders, who received large kickbacks, to the mayor's reelection slush fund. Countries peripheral to the international economic system and whose economies depend on natural resource extraction, like Bolivia, tend to have few formal sector jobs (Castells and Portes 1989;Fernández-Kelly and Shefner 2007). Additionally, the formal employment that does exist is mostly in minimum wage service, tourism, and resource extraction jobs (Kohl and Farthing 2006). People take jobs in informal markets because they have more career and income options as self-employed workers in the informal economy (de Soto 1989). In these contexts, becoming a bureaucrat or a civil society leader can be one path into the middle class-especially in countries like Bolivia with a heavy international NGO presence (Kohl and Farthing 2006;Kohl 2003). Still, local officials and civil society leaders are often poorly paid and overworked and are happy to increase their income (Lazar 2007). Facing these options, officials and civil society leaders have powerful incentives to enter into mutually beneficial corruption agreements. This study worked with street vendor representatives as cases of civil society actors who become intermediaries in corruption; future research could investigate if similar incentive structures exist for other types of civil society leaders, like NGO directors, members of the media, or prominent lawyers. Future research could also examine if intermediary roles change in different economic contexts.
Where representatives work as corrupt intermediaries, they are unlikely to be allies in corruption reforms that rely on civil society to monitor officials and report corruption, as many reforms do (Berliner and Erlich 2015;Kosack and Fung 2014;Kohl 2003). One option is to pay independent monitors. For example, transparency units inside the city government could be run by civil society organizations that receive grants to do so or include paid civil society representatives. Another option already in practice (Kenny and Savedoff 2013) is compensating officials for clean, high-quality performance; in essence, replacing corruption income with performance pay. Finally, governments need regular, external, and independent audit systems to follow up on reforms and head off any organic attempts to co-opt or subvert them. Independent audits are a low priority with extra costs and little political payoff for most governments (Kohl 2003); international organizations could consider paying for audits and expanding existing auditing programs.
Auditors and corruption scholars should be aware that collusion can look like successful reform. The La Paz Markets Office looks like a success story: the Transparency Unit inside the office is fully staffed, prominent posters state which services are free and what the official fees are and list numbers for anonymously denouncing corruption, and the well-educated bureaucratic staff is fluent in the international discourses around reform, transparency, and accountability. Yet Markets Office employees enjoy a standard of living well above their official salaries and the public they serve complains of rampant corruption. When assessing reforms, auditors and corruption scholars should adopt practices from innovative studies like Sequeira and Djankov (2014) and Olken and Barron (2009) and set up systems to directly observe potential corruption, and talk with users of public systems instead of only reform architects, managers, and employees.
Corruption research has largely overlooked how intermediaries perpetuate corruption and why representatives face strong incentives to facilitate corruption. This project contributes to work on intermediaries in corruption by confirming the role that intermediaries play in facilitating corruption but, contrary to existing theory, finding that intermediaries primarily benefit the politically powerful. Additionally, this study cautions reformers that intermediaries can frustrate accountability reforms by restricting the number of people who witness corruption and making witnesses dependent on corruption. Intermediaries' strategies work to insulate officials while extracting money and power from ordinary citizens.