Elite Capture and Co-optation in Participatory Budgeting in Mexico City

. Participatory Budgeting opens up the allocation of public funds to the public with the intention of developing civic engagement and finding efficient uses for the budget. This openness means participatory budgeting processes are vulnerable to capture, where through subtle or unsubtle means authorities reassert control over the PB budget. With a focus on PB processes in Mexico City, this paper explores areas of vulnerability and approaches used to combat them. Digital approaches can be used to address issues during voting or in the monitoring phase, but these can raise the cost both of administering and of participating in the process, or may skew participation towards certain demographics.


Introduction
Development advocates and institutions have, over the last 20-30 years, invested heavily in creating processes and structures to bring governance materially closer to citizens, and to increase participation as a method of reducing corruption and the monopoly over resources that distinct local elites hold.Participatory Budgeting (PB) has been considered one way of increasing trust in government and reducing institutionalised corruption through creating transparent, open processes that deliver citizen needs.It is perceived (and even described) by some as a "new golden bullet" [1, p 198] in the fight against corruption.However, as a mechanism of public participation it by necessity includes openings that might allow pre-existing institutionalised norms and values to be reasserted, and that may result in the recapture of resources by governments or third parties.
This paper focuses on this concept of elite capture through institutional responses to disruption using empirical evidence from the participatory budgeting program in Mexico City, which exhibits multiple different undermining behaviours relating the the process s both in its digital and physical manifestations.It also discusses the broader implications of conducting PB digitally, examining the various digital democracy e-platforms used in instances of PB around the world, and the trade offs between accessibility and security of the ballot these platforms must make.This paper argues that PB is not in fact a golden bullet that miraculously rids communities of corruption, but a process vulnerable to re-institutionalisation and elite capture in both its digital and physical form.Background to Participatory Budgeting Participatory Budgeting (PB) is an umbrella concept that covers a wide range of potential schemes.
Generally, it refers to the involvement of the public (through deliberative mechanisms or votes) in the setting of spending priorities or selection of individual public projects to implement.While its start in Brazil was tied to a specific moment and political project, Peck and Theodore [1] describe it as an example of "fast policy" -where it has rapidly spread around the world, adapting a broad concept to the service of a diverse range of actors.Notable is its adoption by the World Bank as a method of promoting good governance, which has been a key force in shaping how the intentions of PB are currently understood.Cabannes and Lipietz [2] separate current PB projects into "political" (an increase in deliberative democracy), "good governance" (better connection of civic sphere to government), and "technocratic" (optimised use of resources).Specific instances may move from one strand to another over time, with the example of the original instance in Porto Alegre shifting to more of a good governance model since 2005.While these categories have been discussed within the literature, PB implementing practitioners are less likely to conceptualise their programmes within this framework.
That participatory budgeting should function as a tool to disrupt institutionalised corruption was part of the initial conception of participatory budgeting in Brazil.As Gilman describes: "In its original campaign for PB, the PT [Worker's Party] outlined four basic guiding principles: (1) direct citizen participation in government decision-making processes and oversight; (2) administrative and fiscal transparency as a deterrent for corruption; (3) improvements in urban infrastructure and services, especially in aiding the indigent; and (4) altering political culture so that citizens can serve as democratic agents."[3, p.6]While the evidence collected for this study suggests that the redistributive aspects of PB have been reduced in its international replication, the focus on corruption has sharpened.World Bank publications describe the problems of "the traditional budgeting process [that] can often contribute to social exclusion and poverty due to elite capture, lobbies, and powerful interests" and in another publication that "[t]he enhanced transparency and accountability that participatory budgeting creates can help reduce government inefficiency and curb clientelism, patronage, and corruption" [4, p.5].
This belief is not without evidence, Zamboni [5] uses government audit reports of violations of public management regulations as a corruption measure in Brazilian counties and found better governance indicators in 7 out of 10 PB counties when they were paired with non-PB counties -however when the number of irregularities was scaled by the amount of money being audited, the picture was reversed, reflecting that the relationship is complicated.In addition Goldfrank [6] suggests that Brazil possesses a number of factors that make PB a particularly successful programme, that are not necessarily present in PB instances elsewhere Goldfrank [6,p.66].Sintomer, Herzberg and Allegretti [7, p.30] argue that "[w]hen it is performed seriously, PB increases the transparency of the use of public money as well as popular control, and therefore reduces corruption."However, the key part of that quote is "[w]hen it is performed seriously"and there is an active literature [8][9][10][11] discussing the potential for corruption and cooption of the PB process.

Subverting PB -Institutions and elite capture
In understanding the wider picture of subversion of PB processes, there is an issue in that much relevant discussion in the literature uses several different terms and the influence of institutional behaviours and norms is underexamined.Most commonly, attempts to control a PB process will be described as "elite capture" -but the exact difference between this and "corruption" in often unclear.Hellman, Jones and Kaufmann [12] argue there is a split in the literature between discussion of corruption and capture, with two discussions occurring in parallel without identifying what makes these terms discrete.There have been attempts to identify areas where these tend to be used, and Dutta [13, p. 4] argues that problems of "corruption" tend to apply to central governments while "capture" by elites is described more as an issue of regional or local government, however this is by no means a definitive distinction, and reflects rather the correlation of the implementation of PB processes with devolution of power to local levels [14].Even within the "capture" discussion, Lund and Saito-Jenson [10] argues elite capture is often "ambiguous" about the exact dynamics of the situation in question, and is "used interchangeably to describe elites' control of decision-making processes, their monopolisation and misappropriate of public benefits and malfeasance, and corruption by elites", drawing attention to the call of Beard and Dasgupta [15] and Fritzen [16] to distinguish between "elite control" and "elite capture" -where "elite control" refers to elites controlling decision-making arenas while "elite capture" occurs when elites capture the benefits of the PB process.
Discourse about elite capture is often unclear about the role of institutions.The assumption of elite capture is generally that the "elite" has interests that are served by attempting to capture a process, and the same terminology can apply both to elites who are and aren't officially embedded in existing decision making structures (a wealthy cultural group vs currently elected officials for instance).However this can conceal or minimise the role of embedded institutional interests and indeed the institutionalised societal structure as a whole.When new transparency and accountability initiatives are implemented, pre-existing and institutionalised power structures adapt in response to the disruptive element and move to mitigate potential losses and perpetuate those existing structures [17].
Goldfrank [8] argues that participatory budgeting is a form of competitive institution building, where PB processes are often set up in conscious opposition to existing insti-tutional arrangements.As North [18] put it "the individuals and organisations with bargaining power as a result of the institutional framework have a crucial stake in perpetuating the system" , and so decidedly unelite workers at a state organisation may act in a way to perpetuate the power of that organisation and undermine the PB process.
The agents of this process may in some cases be legitimately be described as elites, but there is a distinct character to the interaction of existing state organisations with PB processes that might operate against the existing interests of those organisations.As such, subversions of PB by state organisations aimed at perpetuating their current power requires a distinct terminology.The term "cooptation" is sometimes used in discussions of subversion of PB in Latin America.Holdo [9] takes the definition of "cooptation" from Selznick [20] -"the process of absorbing new elements into the leadership or policy-determining structure of an organisation as a means of averting threats to its stability or existence" (p.3).
Keeley [11] describes two ways in which elites can undermine PB processes; either by simply ignoring priorities chosen by citizens, or by seeding the room with their supporters or structuring meetings to exclude other factions.These systems may experience change not just through changes of elite, but by escalation of subversion and anti-subversion by competing groups or organisations.Undermining methods of undermining is difficult, as the people and organisations who benefit may simply shift strategy.Sheely [21, p. 252] found that where an experiment to mobilise citizens to attend meetings in Kenya was successful, this led to a change in the meeting format to retain control -"if mobilisation is successful in increasing participation in planning meetings, it may also cause elites to modify the tactics they use to maintain influence over participatory institutions".
Institutional theory, with a focus on elite capture, is therefore a key framework with which to understand the phenomenon through which PB programmes, designed to disrupt the existing hegemony, are subverted.

Research Methods
This paper utilises a portion of a larger pool of evidence collected for a study into the current operation of PB internationally.Specific case studies were carried out in Mexico and Kenya involving a mix of semi-structured interviews, and collection of data and documentary (hard copy and digital) evidence.In addition, a technological review of several digital platforms operating internationally was conducted.Finally, semi-structured interviews were carried out with a wide range of stakeholders in the field of PB, including implementers, practitioners, academics, NGOs, local officials, philanthropic organisations funding PB, and international development professionals situated in key PB promoting development organisations.
To quantitatively explore the Mexico City PB process, results for each of the 1,183 different PB ballots ran in 2017 were scraped from the Electoral Institute of Mexico City (IEDF) website (http://sistemas2.iecm.mx/consulta2017/resultados/index.php).These were analysed to examine variation by delegation in engagement with online voting and the victory margin of the winning project.The current status of all projects selected to be implemented was also scraped from the IEDF monitoring website (https://portal.iedf.org.mx/seguimientoproyectos/presupuesto2017/index.php) to examine variations by delegation in engagement with the platform.

Findings
The discussion of the research findings are broken down into several sections.First, PB in Mexico City is examined, with a focus on the structure of PB programmes, the institutional pressure points exerted upon them, and potential points within the programme that elite capture can be experienced.This discussion also provides voting data taken from the PB programmes which demonstrates the distribution of votes and what they suggest about project competitiveness, including potential issues with the digital format.Second, a discussion of the specific digital aspects of PB programmes in general is provided, encompassing a number of international implementations and outlining the potential points of subversion, but also highlighting the technological aspects of digital PB that may reduce the potential for re-institutionalisation and elite capture of PB programmes.Third, a discussion of the weakness of PB monitoring is provided, citing examples of elite capture and re-institutionalisation of PB distribution activities.

PB in Mexico City
Early PB schemes were run in two parts of Mexico City (Cuahtemoc and Tlaplan) in the early 2000s, but PB in its current form in Mexico starts in 2010 with the passage of the Citizen Participation Law of the Federal District (LPCDF).This mandates a process that each of the 16 local authorities will run for neighbourhoods in their areas, with the voting administered independently by the Electoral Institute for Mexico City (IEDF).This was amended in 2012 to introduce a mandate for 3% of each local authority's budget to be spent on PB [22].The funds are restricted to certain types of project, Wiemann and Fuchs [23] (who has previously written on the PB process in Mexico City) call this "pre-structuring", meaning that substantial decisions on what kind of projects in what area are effectively possible, are decided by the rules of process before any project proposals have even been invited.Pre-structuring in this context provides scope for institutional influence and potential elite capture in the very design of the process itself.
Overall, 3% of city budgets are committed to PB.In 2016 this was $874,920,194 pesos or around US$ 47 million [24].However, because Mexico City is geographically large and contains a very dense population, the budgets represent quite a small amount of money per head.As the majority of the budget of these local authorities is used for salaries, 3% represents a large proportion of the remainder, and is therefore a substantial imposition on the actual services budget.This creates an incentive for local authorities as institutions to attempt to capture the process of project allocation.

Selecting Projects
In Mexico City, proposals are submitted in writing on a standardised form, and expert councils comprised of academics and other technical experts are assembled by each local authority to determine which projects can proceed to a vote.Ideally, this gives citizens a stake in projects proposed and leverages local knowledge to suggest better fitting solutions to local issues.
However, this first step of the process is vulnerable to subversion, as while a project may be proposed by a citizen, this is not necessarily where the idea originated.Rather than being citizen sourced, many projects are "citizen-washed", that is, projects are proposed by the local authority itself, using a citizen as an intermediary.This undercuts any benefit of local knowledge in the citizen sourcing of proposals and provides citizens with a choice of projects that are likely to have been implemented by the authority without the PB process.In cases such as this, the local authority can remain in effective control of its budget if the choice of projects is controlled.
Another concern raised in Mexico City was that the written formalised process might disadvantage or deter project ideas proposed by illiterate or less literate citizens.Similarly, in Solo (Indonesia) Grillos found that the poorest subunits were less likely to submit proposals overall [25].

Voting
For electronic voting in Mexico City, fluctuating numbers of voters over time reflect changes in the security of the ballot.As an EMPATIA Project presentation noted "The chance to directly influence public expenditures can generate deceitful or abusive behaviors in PB.ICT vulnerabilities increase this risk, which is limited in face-to-face interactions."[27].In Mexico City, all that was required to vote online was information on the voting card.Interviewees mentioned incidents where authorities or third parties asked for information from residents to register them, but then discovered when they came to vote that their vote had already been cast.There was a substantial problem with online fraudulent votes in 2015, leading to security additions in future rounds reducing the number of online votes from around 100,000 to 5,000.This mirrors the similar transition in the PB program of Bela Horizonte.In 2008, the online PB included 124,000 participants (three times more than the offline component), however complaints of fraud led to increased security standards and by 2011 only 44,000 people participated online [28].
In the 2017 Participatory Budgeting exercise there were 279,023 valid votes cast offline and 4,554 online over 1,183 neighbourhood votes.Only 15 projects had the winner changed as a result of the online vote.There is substantial variation in the number of votes cast in each delegation (due uneven population size as well as uneven turnout).Online voting was unevenly distributed, Iztapalapa accounted for 18.6% of the total offline vote, but 43.4% of the online vote.
Issues of ballot security were not limited to online voting.Interviewees mentioned problems of vote buying, where hundreds of street vendors might be registered in a single building by organised crime groups for the purposes of generating voting cards in an area.This is interesting, as while this can be considered "elite capture" of PB, and subject to influence by institutionally powerful actors, the organised criminal element here presents an additional layer of opaqueness and corruption previously unconsidered.A potential concern for elections with large numbers of projects on the ballot is that votes may be too widely distributed and result in winners with low popularity.However, this is in most cases not the case in Mexico City.Despite there being an average of 8.8 options on the ballot, the winning option tended to receive majority support and the average percentage of the total vote was 63.2%.To make comparisons between ballots with different numbers of options, the winning percentage can be expressed as a ratio to the percentage if all votes were distributed evenly.For example, in a four option ballot the even distribution would be 25% -so a winner with 75% would have a ratio of 3. Higher numbers reflect a wider discrepancy between reality and the "even" result.Table 2 shows there is a large gap between first and second place, with an average ratio of 4.74 for first place and 1.39 for second place.This varies between delegations, with Iztaclco having a ratio of 8.80 and Benito Juárez a ratio of 2.82.To express this as a percentage of total votes -the average distance between first and second place was 45% of the total vote.This reflects that in most local PB ballots, there was no serious competition, one project accounted for a large percentage of the vote and won by a significant margin.This may reflect genuine local support, but might also be indicative of areas where there is a strong official project and/or vote buying, and where institutional norms facilitate elite capture of programmes.

Implementation and Monitoring
Once a project is selected through a vote, there remains the process of converting it from idea into reality.Without sufficient post-award monitoring, there is scope for funds to go missing or for the project to fail to materialise altogether.Plata argues that one of the deficits of the Mexico City system is "the lack of clarity about the institutional mechanisms for citizens to participate in the implementation and evaluation" of the projects.[24,p. 66, translated] After the vote, there is no formal role for the submitters of the project.It is essentially in the hands of local authorities to arrange contracts and complete.In these conditions, the choice of contractor can serve as a conduit for patronage.A citizen group interviewed complained about an incorrect (and cheaper) construction technique being used in a successful sidewalk repair project a few years earlier.In the context of amorphous projects like road repair, there is significant room for money to go missing through selection of cheaper materials.
In these instances corruption would result in an inferior final project -but it is still completed.In other cases, the local authority has reported that the project is complete when in fact it wasn't even started.A commonly overlooked issue in Participatory Budgeting is a formal way of tracking progress of the projects.In Mexico City the IEDF have built a website that encourages social media reporting of project statuses so that there is some room for popular monitoring of the construction process.While this website is an interesting approach, it suffers from low engagement that is typical of this kind of platform (see Bailard et al. [29]).Table 2 shows that 67% of projects are still marked as unstarted.While 25% are marked as completed, these are mostly concentrated inside a few delegations, 9/16 delegations have no engagement on any project.The utility of this tool in Mexico City is limited by the fact that IEDF has no legal powers of enforcement, but it has partnered with other offices of the government of Mexico City to attempt to hold corrupt officials to account.Where budgets or projects were changed (for instance if a local authority says they have constructed a computer center in a school, but it was never built), files are passed over to other offices and around 40-50 people in the local authorities have been sanctioned or removed from office as a result of this process.While her research pre-dates this approach and so doesn't pass judgement directly on its success, Sánchez recommends the law related to participatory budgeting in Mexico City be amended to add explicit penalties on officials for corruption or non-compliance in the process [22].
Kossow and Dykes [30, p. 28] argue that truly effective anti-corruption ICTs "need a functioning accountability framework that includes an independent judicial system, press freedom and an active civil society".In Mexico City, while there is some framework of accountability, this is ad-hoc and depends on the organisations involved taking a wide view of their remit.While the monitoring platform is generally under-used, it is not clear that a more trafficked platform would lead to greater accountability or delivery.It is the tool that can be implemented by the current organisational actors, but not necessarily sufficient to achieve the goal of ensuring delivery of high quality projects.
Olken [31] argues crowd-sourced approaches to PB monitoring have problems with elite capture (where the monitoring is undermined by the elite benefiting from the process) and free-riders (where people who might hypothetically benefit from monitoring do not participate, hoping to benefit from other people's work, to the detriment of the overall monitoring).In a field experiment, he found that expert audits of road surfaces decreased discrepancies between official and estimated costs of road projects.Grassroots monitoring groups were ineffective at this, but effective at detecting discrepancies in Labour.

Conclusions
This paper has examined the risks of cooptation and elite capture in the PB process in Mexico City.The attitudes and incentives of key organisational players such as the local authorities, the electoral institute and civil society groups illuminate the current conflicts (or lack of conflicts) over the operation of the PB system and demonstrate the institutional influences that facilitate them.At the proposal stage, weak citizen engagement with PB leaves the local authorities as the major actor capable of creating high quality projects.While individual projects from citizen groups can succeed, projects "citizen-washed" by councils are at a distinct advantage over other projects.That there is a clear winner in the majority of project selections despite the range of projects on offer reflects a lack of competition at the local level.For the voting stage, the IEDF as an external organisation exists in conflict with groups attempt to affect the outcome.The IEDF is able to adjust the voting process over time to address risk of subversion, this is however at the expense of a wider base of participation and cannot address the problem if effective choice has already been constrained at the proposal stage.
The IEDF's monitoring website deserves consideration as an example of how "crowd-sourcing" monitoring sites can supplement the current pre-vote focus of electronic PB implementations.However, the wider literature on anti-corruption platforms suggest limits to this approach -while citizens may be in a position to recognise if a project was delivered, they are far less able to detect mis-allocation of resources in construction (via materials or employment).For monitoring websites to be successful they must be embedded in a system that can make use of complaints.In the absence of strong sanctions for non-delivery, even a well-used website may not accomplish its goal of ensuring projects are delivered.
This paper has demonstrated how pre-existing institutional frameworks have enabled the elite capture of PB programmes in Mexico, and should be considered a first step in better critiquing and testing how such programmes may be made "tamper-proof" to institutional pressures.

Table 1 .
Valid Votes by Delegation

Table 2 .
Vote Ratios by Delegation

Table 3 .
Project Progress (as reported) by Delegation