NGO-Government Collaboration

Collaboration between NGOs and local governments is advocated by scholars and policy practitioners as an important step to avoid the problems so common in previous “eras” of service delivery. In the decades of the 1950s, 1960s, and 1970s, service provision was largely assumed to be the purview of the state. However, many governments in developing countries struggled to improve service delivery, facing challenges of corruption and weak capacity. In the 1980s, support for state-led development declined, and a wave of neoliberal policies curtailing the size of government ushered in an era of NGO proliferation (Agg Reference Agg2006; Edwards and Hulme Reference Edwards and David1996). NGOs began to step into service delivery areas where government provision had receded or had failed altogether (Salamon Reference Salamon1994).

According to the Union of International Associations (2014), from 1980 to 2000, the number of NGOs operating globally increased by 245 percent. Their financial support has continued to grow: aid channeled to and through NGOs climbed to 37 percent of all official development assistance by 2012 (OECD 2014). The result of the increase in NGO activity is a patchwork of service provision, with governments providing the bulk of services in some contexts, NGOs in others, and in many cases, both engaging in parallel service delivery.

Parallel service provision raises concerns about the effectiveness of aid and the impact of NGO activity on government engagement in service delivery. Scholars worry that as NGOs garner greater support and step into underserved areas, governments’ inability to provide services can become a “self-perpetuating reality” (Farrington and Lewis Reference Farrington and Anthony1993, 333). When NGOs work independently of government, the lack of cooperation can lead to coordination failures that reduce the efficiency of aid in achieving development outcomes (Bräutigam Reference Bräutigam2000). Absent strong leadership, either from a coordinating body or from capable governments, NGOs risk competing among themselves and with governments and duplicating efforts by carrying out redundant or conflicting projects (Mercer Reference Mercer2003). Furthermore, because many NGOs rely on funding streams that are highly variable, the long-term sustainability of their efforts may be questionable, whereas governments presumably offer a better guarantee of long-term presence.

After the mid-1990s, the discourse on aid and development shifted to recognize the centrality of the state in service provision (Agg Reference Agg2006; Batley and Rose Reference Batley and Pauline2011). Attention to the role of good governance in development outcomes led to policies that prioritized government capacity building (Bräutigam and Knack Reference Bräutigam and Stephen2004). Collaboration between NGOs and governments thus became a critical way to avoid the pitfalls of parallel service delivery and to channel aid into activities that encouraged, rather than deterred, government engagement in service delivery. At successive high-level forums on aid, donors and aid-receiving countries agreed on the importance of aligning donor investments with government priorities. Incrementally, donors have made funding flows contingent on NGOs’ partnering with governments (Batley and Rose Reference Batley and Pauline2010, Reference Batley and Pauline2011). Thus, at least in theory, if not in practice, the pendulum shifted away from the extreme of NGO service provision that arose in the 1980s to a middle ground of “collaboration” between state and nonstate providers.

The effect of NGO-government collaboration on the quality of service delivery is still a matter of debate. The consensus among scholars and practitioners is that collaboration yields benefits for both NGOs and governments beyond addressing the pitfalls of parallel service delivery. When NGOs have connections to international organizations, they can be conduits for the spread of norms and best practices to governments to enable more efficient or informed service delivery (Murdie and Hicks Reference Murdie and Alexander2013; Keck and Sikkink Reference Keck and Kathryn1998). For example, Martínez Nogueira (Reference Martínez Nogueira and Charles1995, 61) details how collaboration between NGOs and municipal governments in Rosario, Argentina spurred governments to replicate an NGO’s “work style” and provided staff with “resources, methodologies, and professionalism.” To the extent that NGOs have meaningful connections to local communities and leaders, they can facilitate community-government interactions (Devine Reference Devine2006; Goldman and Little Reference Goldman and Jani2015). Moreover, collaboration with NGOs can give local governments access to additional funding.

For NGOs, collaboration with governments can increase their legitimacy and potentially increase the sustainability of projects after the contract period ends. For example, Keese and Argudo (Reference Keese and Marco Freire2006) document how collaboration between an international health NGO and a municipal government in Ecuador improved the municipal governments’ provision of health services with technical assistance and financial support while enabling the NGO to extend its programming across the municipality, increase its efficiency, and sustain its activities over time.

In contrast to the optimistic view of collaboration, some scholars warn that collaboration can have consequences for NGOs and for the quality of democracy. For example, Cook et al. (Reference Cook, Glenn and Krister2017) find that NGO influence on governments has unintended consequences for government’s responsiveness to local needs and interests. Others assert that collaboration urges NGOs to shift away from their diverse approaches, which can be more contentious in challenging the state, to a more professionalized, hierarchical organization acting as a contractor for state-supported service delivery (Álvarez Reference Álvarez1999). Moreover, government collaboration with more technical NGOs or those with stronger ties to international donors can overshadow the voice of community groups that may not be well represented by such organizations (Álvarez Reference Álvarez2009; Viterna et al. Reference Viterna, Emily and Killian2015).

Twenty years into the era of prioritizing partnership, clarity regarding what factors encourage or discourage governments and NGOs to collaborate is still lacking, despite the shift in theoretical best practices and the contingencies in funding for such collaborations (Mcloughlin Reference Mcloughlin2011; Najam Reference Najam2000). The bulk of what we do know about NGO-government collaboration stems from research on relationships between central governments and the NGO sector (Bawole and Hossain Reference Bawole and Farhad2014). At the national level, scholars have described the relationship between governments and NGOs as a strategic interplay, determined on one side by the state’s incentives to control NGO operations while benefiting from the resources they contribute, and on the other, NGOs’ incentives to maximize their role in government agenda setting and resource allocation while avoiding government control (Fowler Reference Fowler1991). Bratton (Reference Bratton1989) suggests that characteristics of state institutions are the primary factors in shaping these relationships. He argues that democracies with more established political party systems are more likely to see NGOs as collaborators rather than challengers to state power (see also Robinson and White Reference Robinson and Gordon1997). Additionally, states facing budget constraints due to neoliberal policies or higher levels of poverty are more likely to allow NGOs more autonomy in order to benefit from the resources they can contribute.

This tug of war between states and NGOs becomes more complex when considering the reality that governments are not unitary, but are instead a network of multiple overlapping government bodies (Asad and Kay Reference Asad and Tamara2014). This is particularly true for decentralized settings, yet variation in NGO-government relationships at local levels is an understudied phenomenon (Appe Reference Appe2010; Fisher Reference Fisher1997, 452). Collaboration that central government agencies might expect to occur at local levels does not necessarily happen (Soeters and Griffiths Reference Soeters and Fred2003). While there is general agreement that local-level NGO-government collaboration is more complex (Mercer Reference Mercer1999), there are competing views about the nature of collaboration. Some scholars have posited that relationships at the local level are closer and collaboration occurs more easily (Clark Reference Clark1995), while others argue that these relationships are characterized by high levels of mistrust and unease (Bawole and Hossain Reference Bawole and Farhad2014; Najam Reference Najam2000).

These dual perspectives are probably generated by a large degree of variation in local-level state-NGO relationships. For example, where Brass (Reference Brass2012a, Reference Brassb, Reference Brass2016) finds NGOs and local governments in Kenya achieving new levels of coordination and integration, others find experiences of struggle and even competition between NGOs and local governments attempting to provide services (Bawole and Hossain Reference Bawole and Farhad2014; Najam Reference Najam2000). The political context in which NGOs and governments are interacting is a key factor. The work of Oyugi (Reference Oyugi, Walter2004), which focuses on variation in the Kenyan government’s relationship with different NGOs, suggests that cooperative relationships are less likely in areas of the country where security and political stability are weaker. Asad and Kay (Reference Asad and Tamara2014), who recognize the multitude of government organizations within a state, explore NGO strategies according to given characteristics of the state. They find that the likelihood of local governments’ and NGOs’ partnering on projects depends on how much local governments depend on external resources. These findings provide a starting place for the question of collaboration and where it is likely to emerge in countries.

Political Strategies of Collaboration

This article argues that variation in patterns of collaboration in local government–NGO relations depends on politics, shifting away from arguments focusing on geography, donor needs, or poverty. In particular, it argues that the strategic decisions of mayors and NGOs shape the trajectory of NGO-government relations.

Our model makes the following assumptions. We assume that governments benefit from collaboration because NGOs provide resources, including connections to citizens or community groups, technical capacity, and even financial support. Yet governments face trade-offs in working with NGOs, including losing the ability to claim credit for public investments, being forced to negotiate the distribution of resources, and providing NGOs, which can pose future threats to incumbent governments, with resources that could ultimately be used against them. Like the revolving door between business and government, instances of NGO employees’ becoming political rival candidates are well documented (Gill Reference Gill1997, Reference Gill2000).

Moreover, NGOs can be well positioned to contest government decisions and to amplify the voices of rival parties or community organizations critical of government, as documented in Bolivia by Córdoba and Jansen (Reference Córdoba and Kees2016). For example, in our own fieldwork in the Peruvian Amazon, which helped formulate our ideas, one newly elected mayor dissolved a partnership with an NGO that had been established under the previous administration, citing a lack of trust in having that NGO work with his constituents and wanting to give proper credit to his contributions to the project (Author interview, June 9, 2011). Thus, while NGOs represent lucrative opportunities for mayors, they can also increase vulnerability and introduce limitations to authority.

We further assume that NGOs face similar strategic challenges in choosing whether to collaborate with local governments. Some NGOs face upward pressure from donors or home offices to establish formal collaboration. Others may partner to gain access to valuable resources, including credibility and information. However, collaboration can come at a cost. NGOs face reputational risks of having political affiliations and can become subject to greater bureaucratic entanglements and political influence. Association with one party may threaten NGO projects where local actors are aligned with the opposition or when elections bring new parties to power. Conversely, operating independently enables more control over budgets and timelines and optimizes decisionmaking freedom around processes and recipients.

One instance of an NGO in Peru illustrates these pressures well. The NGO, which provides drinking water systems to rural communities, avoided interacting with the government out of concern that collaboration might slow down the implementation of projects and introduce a political element to its work when the NGO wanted to be viewed as politically neutral. With time, however, the NGO became interested in the possibility of collaboration. The change was driven by two benefits: financial support to maintain the NGO’s ongoing support of existing infrastructure it was struggling to fund, and the ability to reach new communities with failed water systems built by the government or other NGOs. The relationship took a year to nurture, after which the mayor and the NGO signed an agreement. The mayor cited the benefits of the NGO’s technical expertise to fix the broken systems. With the agreement, the mayor was able to demonstrate expanding water access for several communities (Author interview, July 11, 2014).

The Case of Bolivia

In Bolivia, the roles of both NGOs and local governments have changed immensely in the last 30 years. In the 1980s, the country’s fiscal crisis made it a testing ground for neoliberal policies. The retrenchment of the state, along with a new Social Emergency Fund provided by the state and international donors, created a new demand for NGO services, along with the supply of resources to support their expansion (Boulding Reference Boulding2014; Gill Reference Gill2000).

This increase of NGOs was further facilitated by the political opening of democratization. The Bolivian official registry of NGOs charts an exponential increase from fewer than one hundred NGOs in the 1980s to more than six hundred by 2000 (Boulding Reference Boulding2014). The growth can be partially attributed to existing NGOs that chose to register in order to capitalize on the opportunities of increased funding and a politically freer atmosphere, but it was mainly driven by the entrance of a multitude of new development actors (Boulding Reference Boulding2014). In this era, the importance of local government was limited; not only were few local governments receiving central government funding, but much of the country was not yet organized into municipal districts.

Like many countries in the 1990s, Bolivia enacted decentralization reform to enhance local governance. Packaged in neoliberal policy prescriptions, the policy of decentralization aimed to achieve greater democratization but also created structural changes necessary to reduce the size of central government, and to shift large spending programs that created openings for political corruption to local levels, where citizens might have greater oversight and voice (Wunsch Reference Wunsch1991). Bolivian decentralization reforms included a suite of policy changes, devolving decisionmaking power to local levels and deconcentrating administrative duties at the national level, dispersing them to local actors (Faguet Reference Faguet2009). These institutional shifts created opportunities for further “privatization” of state-owned enterprises to private firms and public services to NGOs.

Bolivia’s Law of Popular Participation (LPP), passed in 1994, created more than 250 new municipalities and mandated local elections, a right previously accorded only to the largest cities (Van Cott Reference Van Cott2000). To fund the new local governments, the central government transferred 20 percent of the budget to local authorities (Faguet Reference Faguet2012). The result of the LPP was a shift in responsibility for critical service provision to local levels, which were often unequipped to step into such roles. The LPP not only created openings for NGO-government collaboration at the local level, but in many ways, it relied on this collaboration to implement the structural transition. NGOs were deeply involved in the process of building fledgling local governments. International donors channeled substantial resources through NGOs to help local governments create municipal development plans (Kohl and Farthing Reference Kohl and Linda2006).Footnote ² The result was a mix of successful collaborative efforts in some municipalities and a lack of collaboration in others (Kohl Reference Kohl2003).

The LPP created considerably more space for NGOs and CBOs to engage with government. Community organizations that represented specific neighborhoods or regions became officially recognized as Territorial Based Organizations, which participated in municipal planning and elected the oversight committee that monitored municipal spending. CBOs and NGOs could participate in budgeting meetings, and many NGOs, particularly in the years after decentralization, became key actors as advisers and subcontractors in the creation of municipal development plans (Kohl and Farthing Reference Kohl and Linda2006).

Throughout these changes, NGOs played important roles in service provision as well as political mobilization. They were enmeshed with indigenous movements and other human rights advocacy efforts. After the LPP, NGOs continued to engage in service delivery, and did so with more interaction with state actors. NGOs became subcontractors in carrying out services and forged collaborative campaigns. Reports from donors describe collaboration between NGOs and local governments across many sectors, from irrigation infrastructure to campaigns to fight Chagas disease. NGO-municipal support fluctuated considerably. For example, in its review of just one region of Bolivia, the Dutch government found that NGO budget support varied so dramatically that in some municipalities NGOs contributed 0 percent, while in one, their support constituted a full third of the municipal budget (IOB 2012).

In 2007, when our data were gathered, the newly elected MAS leadership had mixed views of NGOs, and policies toward NGOs began to shift soon thereafter. On the one hand, many NGOs had contributed to the rise of the MAS party and to the election of Evo Morales. Some NGOs engaged in politics directly and were outwardly supportive of the MAS (Shakow Reference Shakow2014; Córdoba and Jansen Reference Brass2016) while others supported the mobilization of historically marginalized groups that became a powerful political base for the first indigenous president (Boulding Reference Boulding2014). On the other hand, the flow of international funding through and to NGOs had created divisions between some NGOs and the community groups they claimed to represent (Álvarez Reference Álvarez1999). In contesting the neoliberal model, the MAS supported a more central role of the state in development and service provision and rejected NGO-driven development, particularly funded by international actors. While state-NGO relations at the federal level remained unchanged in the first years after the election of Morales, starting in 2008, the Morales administration incrementally implemented policies that curtailed NGOs’ ability to operate independently, and its rhetoric toward NGOs, particularly those critical of the administration’s policies, became increasingly hostile.

This raises the question of the generalizability of local NGO-state relationships in Bolivia. Two important dimensions of comparability are the decentralization of service provision to local levels, which increases the responsibility and pressure on local government actors and, in turn, heightens the need to negotiate with nonstate actors; and a relatively open national policy toward NGOs, so that NGOs have latitude to opt out of collaborations. The case of Bolivia in 2007 is likely to be relevant to contexts with relatively substantive external aid flow, in which local officials have been politically empowered enough to make decisions about collaborations, and in which central governments do not employ restrictive practices toward NGOs.

Data and Variables

Data used for this study were compiled from several sources, generously shared by Carew Boulding and Krister Andersson (see table 1). Primary data on the perception of NGO-municipal collaboration come from a 2007 survey conducted by Andersson (see Andersson Reference Andersson2013; Andersson et al. Reference Andersson, Gibson and Fabrice2006; Wright et al. Reference Wright, Krister, Clark and Tom2015). The surveys were conducted with mayors from 100 randomly selected Bolivian municipalities (see map 1) (Andersson Reference Andersson2013).Footnote ³ In addition to demographic information for mayors and municipalities, the survey included 258 questions covering policy priorities, relationships with central government and nongovernmental agencies, and dynamics with citizens. These data were supplemented by government data from multiple projects, including separate projects conducted by Andersson and by Boulding (Andersson Reference Andersson2013; Boulding Reference Boulding2010, Reference Boulding2014; Boulding and Gibson Reference Boulding and Gibson2009).

Table 1 Summary Statistics

Data source: https://archive.org/details/GisDataMunicipalBoundariesOfBolivia

Map 1 2007 Survey of Mayors in Bolivian Municipalities

NGO–municipal government collaboration. Partnerships between local governments and NGOs could include a variety of interactions, from communication to contractually established partnerships. We work from a conceptualization of collaboration as a formal act as opposed to more informal interplay, such as cooperation or complementarity (Coston Reference Coston1998). We are most interested in formal collaboration, specifically the joint decision to undertake collaborative projects (Farrington and Bebbington Reference Farrington and Anthony1993). Our conceptualization of collaboration includes cases of NGOs’ contributing resources to local government projects, joint agreements to deliver services, or projects to which both parties are devoting resources. The nature of the survey limits the analysis to the role of formal collaboration, but it can still inform future research considering a broader scope of forms of collaboration.

The dependent variable, collaboration between NGOs and municipalities, is reported by the local mayor. It is a summative index based on three dichotomous questions on collaboration from the survey: whether NGOs work with the government in exchange for salaries or other forms of payment, whether NGOs offer volunteers to work on priority areas of need for the municipality, or whether NGOs donate money or resources to the municipality for work on priority areas of need. The index captures whether municipalities report no collaboration, collaboration in one of these areas, or collaboration in more than one of these areas.

One concern could be overreporting bias of instances of collaboration, but as shown in figure 1, the data are well distributed, with slightly more than a third of mayors reporting no collaboration. Only one municipality reported all forms of collaboration, which, as a sole outlier, was recoded to 2 instead of 3. Of those reporting some degree of collaboration, about 44 percent of the municipalities reported one form of collaboration and about 20 percent reported a higher degree of collaboration, indicating that at least two forms of NGO collaboration were occurring in those municipalities. Notably, the measure tells us nothing about the types of NGOs with which mayors are collaborating or the number of NGOs with which mayors have partnerships.

Figure 1 Distribution of NGO–Local Government Collaboration

Results and Discussion

In the results shown in table 2, a positive coefficient indicates that an increase in that measure is associated with more NGO-government collaboration. The baseline probability of no collaboration is 26 percent, while collaboration of one form is 58 percent and more than one form is 16 percent.Footnote ⁸

Table 2 Explaining NGO–Local Government Collaboration

^*** p < 0.01

^** p < 0.05

^* p < 0.1

Log-odds coefficients with standard errors in parentheses.

In relation to political volatility, the results fully support our first hypothesis but only partially the second. Mayoral instability is significant and negative in both models. The probability of no collaboration occurring substantially rises with the number of mayors since 2000, from a probability of .28 with only one mayor to a probability of 0.83 with the maximum of eight mayors, as depicted in figure 2. Collaboration depends on the evolution of relationships. Theoretically, mayoral stability gives both NGOs and mayors time to cultivate trust and opportunities necessary for collaboration. Political upheaval deters collaboration. While we do not have evidence of which actor (the NGO or the mayor) is less likely to invest in collaboration, due to the nature of these data, these findings support additional investigations along those lines.

Figure 2 Predictive Margins

The effect of political competition is only weakly significant in the first model and drops to insignificance when controlling for municipal revenue. If a strong opposition party competed in the last election, it might signal a more politically contentious context, but it may or may not lead to turnover of incumbent mayors. We find that turnover is a more important factor than the size of the shadow of the political opposition in the previous election.

These results suggest that, consistent with our third hypothesis, the revenue streams of municipal governments can be important factors in NGO-government collaboration. The probability of collaboration diminishes as municipalities’ per capita income increases. This finding is consistent with the work of Asad and Kay (Reference Asad and Tamara2014). At low levels of revenue, the probability of no collaboration is 0.25, compared to 0.72 for municipalities with the greatest resources. Municipal funding is not correlated with measures of poverty or urbanization, primarily because it includes transfers from central government that compensate for these factors and can affect the size of municipal budgets. We believe that this relationship between municipal budgets and collaboration is primarily driven by access to finances, rather than being an indicator of capacity. Further tests of the portion of the budget that comprises tax revenue were insignificant.

We further find confirmation of hypothesis 4, that CBO engagement significantly and negatively affects NGO-government collaboration. As mayors report greater expression of opinions by community organizations, they also report lower levels of collaboration with NGOs. For mayors reporting the minimum amount of CBO engagement across all issue areas, the predicted probability of one or more forms of collaboration is 0.49 and 0.42, respectively, whereas no collaboration is 0.09. However, at the highest levels of reported CBO engagement, the probability of no collaboration increases to 0.66, and falls, for one form of collaboration, to 0.3 and more than one form of collaboration to 0.04. On the one hand, it may be that engagement indicates more contentious relationships, making partnerships less likely. On the other hand, it may be that NGOs are playing a substitutive role for civil society, which supports the literature that suggests that NGOs and civil society are sometimes at odds (Viterna et al. Reference Viterna, Emily and Killian2015; Álvarez Reference Álvarez1999).

The last central hypothesis, that collaboration is influenced by a mayor’s political party affiliation, is rejected. This is an interesting finding, given that social organizations that were supported by NGOs aided the rise of the MAS, yet we do not have evidence that this translates to greater NGO cooperation down the road. We further tested other political parties and found no effect. Like political affiliation, education levels do not appear to be a driver of NGO collaboration.

We do not find strong evidence that high levels of need based on poverty or rural development are associated with higher levels of collaboration. This result is consistent with the work of Brass in Kenya (Reference Brass2012a, Reference Brassb). Neither HDI nor rurality is statistically significant. Moreover, we find that collaboration is not necessarily affected by the availability of NGOs. In the first model, the number of NGOs registered in a municipality is only weakly significant, and the effect drops out with the inclusion of municipal revenue in the second model. We should caution, however, that because municipalities in the sample have slightly fewer NGOs on average than the full population, the variation could be truncated, leading us to falsely disconfirm a relationship.

To test the robustness of the model and explore alternative explanations, we tested additional controls and explored the possibility of a spatial effect of collaboration. Our tests for the presence of spatial autocorrelation, either a dispersed pattern or a clustered pattern, were insignificant (Moran’s I of 0.036, p = 0.22). Furthermore, in additional modeling, we found that controlling for population, percent indigenous, percent Quechua, and size of municipality yielded insignificant results and did not affect the key findings (see online appendix).

The methodological strengths of this study derive from the large random sample of primary survey data complemented by public data from multiple government sites. With this perspective, we can see and explain patterns of collaboration across the country. While the data permit this new approach, they are also limiting. First, we must rely on mayors as informants of collaboration. Without any means to confirm the accuracy of their reporting, this analysis is vulnerable to error from recall limitation and bias if mayors have personal or political reasons not to portray accurately their relationships with NGOs. Second, the measure of collaboration is limited. It omits more informal collaboration, as well as information about the types of NGOs with which governments collaborate. NGOs are a heterogeneous group, with varying objectives, orientations, and capacities, all of which may influence their own incentives in collaborating. Dyadic analysis could be an area for future research.

Furthermore, because the data come from a specific country and a snapshot in time, they cannot answer questions of how central government policies or other contextual factors alter the nature of NGO-government relationships at the local level. Our data originate from the early years of the Morales administration and more than a decade into the implementation of decentralization reforms. Relationships between government and NGOs change and are influenced by the national political context. In Bolivia, they have become more tenuous as the Morales administration has grown wary of NGOs that have been critical of the government. The government has gone so far as to expel NGOs suspected to be affiliated with USAID or the Danish government and has threatened other NGOs that have openly opposed the state’s position on natural resource extraction (Córdoba and Jansen Reference Brass2016; Wolff Reference Wolff2012). Further analysis of the impact of central government policies on local state-NGO relationships will help confirm the generalizability of our findings.

Conclusions

This project contributes to the scholarship on NGO-state relations by adding evidence on local-level patterns through a large-N quantitative approach. Strikingly, it finds that the lack of collaboration is common in a country where many observers would assume that NGOs and local governments are actively collaborating, given the high profile of NGOs in Bolivia, the active role they played in the establishment of municipal governments after decentralization reforms, and even their role in helping to elect the first indigenous president. Perhaps unsurprisingly, this study finds that collaborative relationships are not occurring where they are needed most in terms of areas with relatively high levels of poverty; instead, it finds that political context influences these relationships.

This research has implications for projects that assess the impact of NGOs on development outcomes. Researchers have found both positive (Murdie and Hicks Reference Murdie and Alexander2013) and negative (Cook et al. Reference Cook, Glenn and Krister2017) impacts of NGO activity. The findings presented in this article speak to this research, suggesting that such outcomes of NGOs’ work may be influenced by the antecedent conditions that shape whether NGOs collaborate or work independently of government.

One key finding—that collaboration is more likely where mayoral turnover is lower—has important policy implications. This finding echoes what some scholars have found in analyses of central governments and the NGOs: where politics are contentious, NGOs and governments are more likely to “go it alone.” This study shows that mayoral volatility in governance disrupts not only municipal administration but also partnerships, limiting the government’s ability to subcontract and mayors’ ability to be more engaged in the work of NGOs. Policies promoted by donors or governments that stipulate that NGOs collaborate with local governments may inadvertently encourage more uneven geographies of NGO activity. If NGOs gravitate to places where collaboration is easier, those places are likely to be ones with greater political stability and more effective governments. If NGOs and governments want to reduce inequities in governance, revising policies to incentivize collaboration in places where it is least likely will be important.

Another notable finding is the negative effect of CBO engagement on collaboration between mayors and NGOs. This has two interpretations. One is that NGOs are opportunistic and even supplant CBOs. This is especially disconcerting if NGOs direct the government’s attention to the interests of donors at the expense of the community’s interests. This concern is salient in Bolivia and has been validated by research (Andersson Reference Andersson2004). The other interpretation is that NGOs are going into places where community organizational capacity is weaker—perhaps where there is greater need. Our cross-sectional data cannot tell us whether NGO activity is in the process of building community efficacy or undermining it (or both simultaneously). One area for future research is to understand the conditions under which NGOs are supplanting or building community-government relationships. Perhaps more important is the need for more cross-national research comparing the subnational realities of how NGOs, CBOs, and mayors are interacting and how central policies affect these triadic relationships.